DEV Community: Michael Inghilterra

The RevOps stat nobody can trace

Michael Inghilterra — Fri, 16 Oct 2026 14:00:00 +0000

"Companies with RevOps grow revenue 19% faster and are 15% more profitable." If you work in revenue operations, you've seen that line on a slide. I traced it. It's a 2015 best case about something else, and the original page now returns a 404.

Where it starts

The number starts in May 2015, at the SiriusDecisions Summit in Nashville. SiriusDecisions was a B2B research firm (Forrester bought it in 2019). Its co-founder gave a talk on "the economics of alignment," and the firm's blog recapped it.

The recap describes benchmark data on 400 B2B organizations, collected from 2006 to 2014, plus a separate survey of 300 sales, marketing and product leaders about why alignment is hard. The claim, in the recap's words: organizations that maintain this focus "achieve up to 19 percent faster revenue growth" and "15 percent higher profitability" than other companies.

That's it. That's the whole foundation.

What it actually measured

Read the original sentence carefully and three things stand out.

"Up to." 19% was a best case, not an average. Nobody quoting it today says "up to."

The subject. The study was about aligning sales, marketing and product management. It never mentions revenue operations. The talk measured a kind of cooperation between three teams, not a function with a charter, a forecast and a tech stack.

The method. The public recap never explains how they scored "alignment," what the comparison group looked like, or how they separated alignment from everything else that makes a company grow. Maybe the full research answered those questions. The public record doesn't.

How it mutated

A statistic doesn't usually get invented. It drifts.

2015: "up to 19 percent faster revenue growth" and "15 percent higher profitability," from aligned sales, marketing and product teams.
2019: a Forrester blog post drops "up to" and uses the number to introduce revenue operations.
Today: vendor blogs say "companies already operating with RevOps report 19% faster revenue growth and 15% higher profitability."
2026: one vendor's RevOps report turns the profitability claim into "15% higher win rates."

Each step is small. Together they turn a 2015 best case about team alignment into a present-tense fact about RevOps win rates. And the one page that could settle it, the original 2015 recap, now returns a 404. The Internet Archive still has it.

How a 2015 best case became a 2026 fact

The other RevOps numbers on the slide

Two more show up in nearly every RevOps pitch deck. Neither is fake. Neither is proof.

"Grew revenue nearly three times faster." From Forrester's 2019 revenue operations research survey, run together with three sales-technology vendors. The published write-up says organizations that "executed it well" and "deployed revenue ops in some form" grew nearly three times faster. It gives no sample size.
"10% to 20% increases in sales productivity." From a 2020 BCG article, describing benefits that top B2B technology companies "are reporting." It gives no sample and no method.

Why this matters more for RevOps than for anyone else

A few weeks ago I wrote that the RevOps seat is shifting from scorekeeper to operator. Whichever seat you're in, you keep one asset through every reorg and every miss: people trust your numbers.

That's why this stat is dangerous for us specifically. Sales can quote an unsourced number and survive it. The function whose whole value is measurement can't. A CFO who checks the 19% will find a 2015 best case about a different question, and every number you show after that gets discounted.

I'm not saying alignment doesn't pay. I've spent my career in revenue operations, sales development and analytics, and I've watched alignment pay off. I'm saying we don't need a borrowed number to prove it.

What to use instead

Make the case for RevOps with your own before-and-after numbers. They're harder to get and much harder to argue with.

Forecast error, by quarter, before and after RevOps owned the process.
Commit-to-close, the share of committed deals that closed in the quarter they were committed.
Time from lead to first touch, because speed on a real signal compounds.
Days lost at handoffs, SDR to AE and AE to customer success. The gap between owners, not the time inside stages.
Report adoption, the share of executive reports that someone opened last month.

Four quarters of those numbers from your own business beat any statistic from someone else's.

What's the RevOps stat you've quoted most, and have you read the study behind it?

Sources

The original: Drew Zalucky, "Summit 2015 Highlights: The Economics of Alignment," SiriusDecisions (now Forrester) blog, May 13, 2015. A recap of co-founder John Neeson's Summit 2015 talk. Wording: organizations that maintain this focus "achieve up to 19 percent faster revenue growth" and "15 percent higher profitability" than other companies. Original URL (returns 404): https://www.forrester.com/blogs/summit-2015-highlights-the-economics-of-alignment. Archived copy: http://web.archive.org/web/20260316032404/https://www.forrester.com/blogs/summit-2015-highlights-the-economics-of-alignment/
2019 restatement without "up to": Forrester blog, June 18, 2019: https://www.forrester.com/blogs/how-to-align-revenue-engine/
"Nearly three times faster": Megan Heuer, Forrester blog, June 13, 2019, from the 2019 Revenue Operations Research Survey run with LeanData, InsightSquared and Outreach. No sample size published. https://www.forrester.com/blogs/revenue-operations-and-cmos/
"10% to 20% increases in sales productivity": BCG, "Revving Up Go-to-Market Operations in B2B," May 13, 2020: https://www.bcg.com/publications/2020/revving-up-go-to-market-operations-b2b
The two present-day vendor versions: checked on the vendors' own pages on 2026-09-27. They're left unnamed on purpose; the point is the drift, not the vendors.

Originally published at michaelinghilterra.com.

The RevOps job description doubled. Nobody wrote the new one down.

Michael Inghilterra — Thu, 15 Oct 2026 14:00:00 +0000

98% of RevOps leaders say their function grew in scope over the past year. 89% say it lacks clearly defined strategic goals.

The state of the function, in two numbers

Those two figures come from the same 2025 survey: Salesloft and Wakefield Research asked 400 US RevOps and executive decision-makers about the function. Salesloft sells to RevOps teams, so weigh it accordingly. But the pairing is hard to ignore.

Read together, they describe RevOps in one line: the job got bigger, and nobody wrote down the new one.

How big the job got

The growth isn't only in scope. It's in seniority.

Head of Revenue Operations was the #1 fastest-growing job in the US in LinkedIn's 2023 Jobs on the Rise list. (LinkedIn grouped Chief Revenue Officer and Head of Revenue Management into that entry, so read it as revenue leadership broadly.)
Director of Revenue Operations ranked #4 on LinkedIn's 2024 list.
73% of companies now have a C-suite role dedicated to RevOps, in the Salesloft survey. RevOps teams there work most closely with COOs (26%), CEOs (25%) and CFOs (24%).
Pipeline generation is now a top-five priority for both sales leaders and RevOps, in Rattle's 2025 State of RevOps survey of 199 US participants.

A function that used to sit under sales operations now works next to the COO, the CEO and the CFO, and it's expected to care about pipeline, not just report it.

The job got bigger. Nobody wrote the new one down.

What didn't grow with it

Three things stayed the same while the scope expanded.

The written mandate. 89% say RevOps lacks clearly defined strategic goals. The scope grew by accretion: one more report, one more system, one more problem nobody else owned. Nobody sat down and rewrote the charter.

The time. A separate 2024 survey of 1,200+ revenue professionals, from BoostUp.ai and RevOps Co-op, found only 23% could prioritize GTM strategy, with many still spending substantial time on systems administration. A bigger mandate on top of the same admin load doesn't produce strategy. It produces a longer week.

The authority. Being asked to care about pipeline is not the same as owning it. In most companies I've seen, nobody has said where RevOps' job ends and sales leadership's begins.

The trap

Put those together and you get the most common way good RevOps leaders get set up to fail: measured on outcomes they were never given the time, the authority or the written goals to move.

I wrote about this in September as the scorekeeper-or-operator question. My own seat was a scorekeeper seat that kept getting handed operator problems. That's the pattern the 98% and the 89% describe at scale.

Write the job description yourself

If your scope grew this year, don't wait for someone else to rewrite the charter. Draft it yourself, one page, and walk it through with your boss before your next review.

1. What you own. The outcomes you're accountable for, where a miss is yours to explain.

2. What you influence. The outcomes you shape but don't control, like pipeline generation when sales leadership owns the reps.

3. What you only measure. The numbers you report and keep honest but don't answer for.

4. The three numbers you'll be judged on. Written down, with how each one is calculated.

5. The admin you'll stop doing, or the headcount or tooling that takes it off your plate. A bigger mandate needs time from somewhere.

The conversation is uncomfortable once. Not having it is uncomfortable every quarter.

RevOps leaders: did your scope grow this year, and did anyone write the new one down?

A correction to something I posted in September

In September I opened a post with "'VP of Revenue Operations' as a title is up roughly 300 percent in 18 months." I traced it for this piece. It comes from a 2021 vendor infographic that counted LinkedIn job titles over roughly late 2019 to early 2021, with no published method. It isn't a study, and it isn't current. LinkedIn's own rankings, cited above, are the better evidence for the same point.

Sources

98% say scope grew; 89% say RevOps lacks clearly defined strategic goals; 73% have a C-suite RevOps role; closest to COOs (26%), CEOs (25%), CFOs (24%): Salesloft and Wakefield Research, press release July 22, 2025. 400 US RevOps and executive decision-makers in four industries; vendor-funded; no fieldwork dates published. https://www.salesloft.com/company/newsroom/revenue-operations-makes-leap-to-csuite-study-says
Head of Revenue Operations #1 (2023): LinkedIn News, "LinkedIn Jobs on the Rise 2023," January 18, 2023. Method: jobs started by LinkedIn members, January 1, 2018 to July 31, 2022. The entry notes the title "can also be known as Chief Revenue Officer or Head of Revenue Management." https://www.linkedin.com/pulse/linkedin-jobs-rise-2023-25-us-roles-growing-demand-linkedin-news
Director of Revenue Operations #4 (2024): LinkedIn News, "LinkedIn Jobs on the Rise 2024," January 17, 2024 (January 2019 to July 2023). https://www.linkedin.com/pulse/linkedin-jobs-rise-2024-25-fastest-growing-roles-us-linkedin-news-dxmie
Pipeline generation a top-five priority for sales leaders and RevOps: Rattle, State of RevOps 2025, 199 US participants (publication date not shown). https://21423273.fs1.hubspotusercontent-na1.net/hubfs/21423273/Guides%20and%20eBook%20PDFs/State%20of%20RevOps%202025.pdf
Only 23% can prioritize GTM strategy: BoostUp.ai and RevOps Co-op, RevOps compensation and impact survey, press release November 14, 2024. "More than 1,200 revenue professionals." Wording: "only 23% of respondents can prioritize GTM strategy, with many still spending substantial time on tasks like systems administration." prnewswire.com press release
The September correction: Clari blog, "The Rise of Revenue Operations: A RevOps Infographic (2021)," February 24, 2021, archived: https://web.archive.org/web/20210524160247/https://www.clari.com/blog/the-rise-of-revenue-operations-infographic

Originally published at michaelinghilterra.com.

Buyers want to skip the rep. Until they need to trust something.

Michael Inghilterra — Wed, 14 Oct 2026 14:00:00 +0000

67% of B2B buyers say they'd prefer a buying experience without a sales rep. 69% say they want a sales rep to validate what AI tells them. Same Gartner survey, 645 buyers.

Not a contradiction

Read those two numbers together and they look like buyers can't make up their minds. I think they describe exactly what buyers want.

Buyers don't want a person for information anymore. AI gives them information faster, at any hour, without a sales pitch attached. What they still want a person for is trust: someone who can tell them whether the AI-generated answer is actually right for their situation, their budget and their risk.

That's the new job description for sales development. Less "deliver the message," more "confirm the truth."

Why the AI SDR story keeps disappointing

The volume of AI in sales is about to explode. Gartner predicts that by 2028, AI agents will outnumber human sellers 10 to 1, and that fewer than 40% of sellers will say those agents improved their productivity.

Both halves are predictions, not measurements. But the gap between them is the whole story. More agents, not much more productivity.

The wider evidence on AI projects points the same way:

Gartner predicts more than 40% of agentic AI projects will be canceled by the end of 2027, due to escalating costs, unclear business value or inadequate risk controls.
MIT's NANDA initiative reported in 2025 that of the organizations that evaluated custom or embedded AI tools, 60% evaluated, 20% reached a pilot, and 5% reached production. The report calls its numbers preliminary and "directionally accurate," so treat it as a direction, not a measurement.

None of this says AI can't do sales development work. It says most deployments point it at the wrong part of the job.

The part to automate, and the part not to

Most AI SDR pilots automate the part buyers were happy to skip: account research, first drafts, sequences and follow-up. That part should be automated. It's where the machine is fast, tireless and good enough.

The failure is expecting the same machine to do the part buyers still want a human for: reading whether an account is actually in pain, noticing hesitation, and telling a buyer the truth about fit.

There's a hint in Salesforce's latest State of Sales survey (weigh it knowing Salesforce sells AI agents). High performers were 1.7 times more likely than underperformers to use prospecting agents. The best teams use AI. They just don't use it to replace judgment.

I led sales development for about seven years, including a 47-person North America org. The best reps I managed were never the fastest dialers. They were the ones a buyer trusted to tell them the truth. That skill used to be one part of the job. Now it's most of it.

Where I'd draw the line

AI owns: account research, signal monitoring, first-draft outreach, sequencing, and follow-up that never gets tired.

A person owns: any reply that doesn't fit the script, any account worth a real theory of why it's in pain, and every moment where a buyer needs to trust something.

The handoff is the design. Route to a person the moment the interaction needs judgment, not after the sequence ends. Then coach that person on exactly that moment, because it's their entire value now.

How to measure a hybrid team

Meetings booked is the one number AI is best at inflating, so it can't be the headline. I'd use three:

Qualified pipeline accepted by sales. A meeting counts once an AE agrees it's real.
Meeting-to-opportunity rate. If AI doubles meetings and halves this rate, nothing improved.
Time from signal to first touch. This is where AI should win clearly. If it doesn't, the tool isn't doing its part of the job.

Measure all three before the pilot starts. Otherwise you'll grade the pilot on whatever number looks best afterward.

If you run sales development: where does the machine stop and a person take over on your team?

A correction to something I posted in August

In August I cited two AI SDR figures: that "only about 2 percent of companies make autonomous AI SDRs stick," and a controlled test where an AI SDR booked 847 meetings against a human-plus-AI team's 312. I traced both for this piece. I couldn't find any source for the 2%: the page it's usually credited to doesn't contain it. The 847 vs 312 comes from one advisor's newsletter about an unnamed company, with no published method. I shouldn't have presented either as data, and I've left both out here.

Sources

67% prefer a rep-free experience; 69% prefer to validate AI-generated insights with sales reps: Gartner press release, May 20, 2026. Survey of 645 B2B buyers, August to September 2025. https://www.gartner.com/en/newsroom/press-releases/2026-05-20-gartner-survey-finds-sixty-nine-percent-of-b-two-b-buyers-turn-to-sales-reps-to-validate-ai-generated-insights
AI agents will outnumber sellers 10 to 1 by 2028; fewer than 40% of sellers will say agents improved productivity (prediction): Gartner press release, November 18, 2025, restated July 28, 2026. https://www.gartner.com/en/newsroom/press-releases/2025-11-18-gartner-predicts-by-2028-ai-agents-will-outnumber-sellers-by-10x-yet-fewer-than-40-percent-of-sellers-will-report-ai-agents-improved-productivity
Over 40% of agentic AI projects canceled by end of 2027 (prediction): Gartner press release, June 25, 2025. https://www.gartner.com/en/newsroom/press-releases/2025-06-25-gartner-predicts-over-40-percent-of-agentic-ai-projects-will-be-canceled-by-end-of-2027
60% evaluated, 20% reached pilot, 5% reached production: MIT NANDA, The GenAI Divide: State of AI in Business 2025, July 2025. 300+ public AI initiatives, interviews at 52 organizations, a survey of 153 leaders, January to June 2025. The report describes itself as preliminary findings, "directionally accurate based on individual interviews."
High performers 1.7x more likely to use prospecting agents: Salesforce, State of Sales, 7th edition, February 3, 2026. 4,050 sales professionals in 22 countries, August to September 2025. https://www.salesforce.com/en-us/wp-content/uploads/sites/4/documents/reports/sales/salesforce-state-of-sales-report-2026.pdf
The August correction: the "2%" appears in a February 2026 blog post that links to a May 2025 page that doesn't contain the figure. The 847 vs 312 appears in the GTM AI Podcast & Newsletter, March 14, 2026, describing "one company I'm advising."

Originally published at michaelinghilterra.com.

The fine-tune that looked fine

Michael Inghilterra — Tue, 13 Oct 2026 14:00:00 +0000

My fine-tuned AI model matched the original on the headline metric and cut false alarms by more than half. It also found 6 of the 14 roles that mattered. The original found 10.

0.707 vs 0.705 Rank correlation with Claude. A tie.
10 vs 22 False alarms. Better.
6 vs 10 of 14 Strong roles found. Worse.

The short version

I fine-tuned a local AI model to judge job descriptions the way Claude does, then tested it on 116 roles it had never seen.
On the averages, it looked like an upgrade. On the roles that mattered, it found 6 of 14. The original found 10.
It had learned to play it safe. It never scored any role above 4.3 out of 5, so the best roles sat just under the bar.
I wrote the pass/fail rule before I trained anything. Three versions later, none passed. The original model stays in production.

The question

I run a small AI lab at home on an NVIDIA DGX Spark. One of its jobs is triage for my job search: a local model gives incoming job descriptions a first read, and only the promising ones go on to Claude for a full evaluation.

That local model works, but it carries a hidden cost. To score like Claude, it needs my full scoring rubric in every prompt: about 22,000 tokens of instructions before it reads a single word of the job description.

Fine-tuning promised a shortcut. Train the model on Claude's past judgments, and the judgment moves into the model itself. The prompt drops from about 22,000 tokens to about 550.

So the question was simple, and it's the same one every revenue team asks about a cheaper tool: can it make the same calls as the expensive one?

The rule I wrote first

Before I trained anything, I wrote down what "better" had to mean. It's the same rule I'd applied to every model that tried to replace the one in production:

It has to rank job descriptions at least as well as the current model (a higher rank correlation with Claude's scores).
It has to find at least as many strong roles. No fewer.

The second condition matters most, because it matches the decision the filter makes. If the filter misses a strong role, I never see a job I'd want. If it raises a false alarm, the cost is one extra evaluation by Claude. Those two mistakes don't cost the same, so I don't count them the same.

What I built

I had 1,160 of Claude's past evaluations to learn from. Each one pairs a job description with Claude's scores and its written reasoning. A strong role is anything Claude scored 4.0 or higher out of 5.

I held back a random 116 of them, 14 of them strong, and never used them for training or tuning. Every result below comes from those 116. The model trained on the other 1,044.

I trained three versions:

Version 1 repeated the strong roles three times in training, so the model couldn't learn to score everything low.
Version 2 pushed harder: strong roles six times, near-strong roles twice.
Version 3 kept version 2's recipe but trained on the scores alone. I left out Claude's written reasoning.

The first run took about six hours on one desktop box. I ran every comparison on the same serving setup, so the only thing that changed between runs was the model.

What happened

	Rank correlation	Strong roles found	False alarms	Big misses
Original	0.705	10 of 14	22 of 101	4
Version 1	0.707	6 of 14	10 of 102	1
Version 2	0.634	9 of 14	16 of 102	0
Version 3	0.472	5 of 14	13 of 102	0

Big miss: a strong role scored 3.0 or lower. The original shows 101 non-strong roles instead of 102 because one of its answers came back malformed.

Version 1 is the one that fooled me, briefly. Its rank correlation tied the original. On a separate test set, the original's own correlation moved by 0.04 between two identical runs, so a gap of 0.002 is a tie.

It also cut false alarms by more than half. On a slide, it was an upgrade.

Then the column that decides anything: it found 6 strong roles out of 14. The original found 10.

Version 2 traded back the other way. It found 9 of 14, with more false alarms than version 1 and a weaker ranking. Both versions gave up strong roles to cut false alarms, just in different amounts. Neither found as many strong roles as the original.

No version found as many strong roles as the original

Scatter plot of four model versions. The original found 10 of 14 strong roles with 22 false alarms. Version 1 found 6 with 10 false alarms, version 2 found 9 with 16, and version 3 found 5 with 13. No version found as many strong roles as the original.

116 held-out job descriptions, 14 strong. Michael Inghilterra's lab, September 2026.

Why it missed: the squeeze at the top

The fine-tuned model never gave any of the 116 roles a score above 4.3. Not once.

That's the whole failure. The bar for a strong role is 4.0. When a model's highest score is 4.3, every strong role has to land in a narrow band to clear it. Seven of the eight strong roles version 1 missed scored between 3.4 and 3.9. Close, consistently, and under the line.

The original misses differently. Its four misses were big: it scored each of those strong roles a 2 out of 5. So the honest comparison isn't a good model against a bad one. The original is erratic. The fine-tune is timid. Timid looks better on an average, because it never makes a wild call. It's worse at the one job the filter exists to do.

The fine-tune's scores bunch just under the bar

Dot plot of the scores two models gave the same 14 strong roles. The original scored 10 of them at 4.0 or above, spread up to 5.0, but scored four of them at 2.0. The fine-tuned model never scored above 4.3, and seven of its eight misses landed between 3.4 and 3.9, just under the 4.0 bar.

Scores for the 14 strong roles among 116 held-out job descriptions. Michael Inghilterra's lab, September 2026.

What I tried to rescue it

The obvious fix for a model that scores too low is to adjust its scores afterward. I tried four ways: shift every score up, stretch the scores to match Claude's spread, fit a straight line to Claude's scores, and lower the cutoff.

None of them held up on a second test set. The first three collapsed: the adjusted models found at most 1 of the 12 strong roles on that set. Lowering the cutoff kept the strong roles and roughly tripled the false alarms.

One option looked perfect. At a cutoff of 3.5 instead of 4.0, version 1 kept all 12 strong roles on the second set. But I found that cutoff by studying that same set, and a test you've already studied can't certify anything. I didn't count it.

The surprise: the reasoning was doing the work

Version 3 was supposed to be the efficient one. Every evaluation Claude writes includes a summary and evidence for each score, and I wondered whether all that prose was wasting the model's capacity. So I trained a version on the scores alone. Its answers ran about a quarter of the length.

It was the worst of the four. Rank correlation fell from about 0.7 to 0.472.

The likeliest reading: writing out the evidence is how the model reasons its way to a score. Take away the writing and you take away the thinking. It's the same reason I'd rather read a rep's notes on a deal than see only their close probability.

The verdict

No version beat the original. I closed the experiment on September 26, and the original model stays in production, unchanged.

I'll reopen it once I have a fresh set of about 35 strong roles the model has never seen, and roughly 600 more evaluations to train on. The next attempt keeps the reasoning and tells the model to weight the scores more heavily, instead of deleting everything around them.

What this means if you run a revenue team

I'm not an ML engineer. My career is revenue operations, sales development and analytics, and this failure is the most familiar one I know.

Averages hide the deals that matter. A forecast can land in total and miss every big deal inside it. A win rate can hold steady while you lose every enterprise deal. My model was right on average and wrong on the roles I'd actually take. Always ask what the number looks like at the top, not just across the whole list.

Pick the metric that matches the decision. The filter's job is to never lose a strong role. Correlation doesn't measure that. Strong-role recall does. If I had judged on the headline metric, I would have shipped a worse filter and felt good about it.

Write the rule before the test. Once you've seen the results, every metric starts to look like the right one. Writing the rule first is the only way to stop yourself from picking the number that flatters the answer you wanted.

Three questions to ask any AI vendor before a pilot:

What's the hit rate on the cases that matter, not the average?
Did you measure it on data the model never saw?
Who wrote the pass/fail rule, and was it before or after the test?

Limits

The sample is small. With 14 strong roles, one role moves the hit rate by about 7 points. Treat the direction as solid and the exact numbers as rough.
The benchmark is Claude's judgment, not hiring outcomes. I measured whether the local model agrees with Claude, not whether either one predicts who gets hired.
One task, one box. Training on a single desktop machine capped each training example at about 3,000 tokens. More memory would allow longer examples, and that might change the result.

Glossary

Fine-tuning: training an existing model further on your own examples so it picks up your specific judgment. I used LoRA, a lightweight method that trains a small add-on instead of the whole model.
Rank correlation: how closely two scorers agree on the order of a list, from 0 (no agreement) to 1 (the same order).
Strong-role recall: of the roles Claude rated strong, how many the model also rated 4.0 or higher.
False alarm: a role the model rated strong that Claude didn't.

This experiment runs inside trajecktory, the open-source job-search command center I built. It's on GitHub: trajecktory on GitHub

Originally published at michaelinghilterra.com.

AI SDRs and the human in the loop

Michael Inghilterra — Mon, 12 Oct 2026 14:00:00 +0000

Two things are happening to sales development at the same time, and they are easy to read as one grim story. Human headcount is shrinking, with about a third of B2B companies cutting sales development roles in the last year. And AI is moving into the top of the funnel, doing the research, the first drafts, the sequencing that used to fill an SDR's morning. Put together, the obvious conclusion is that AI is replacing the SDR.

I do not think that is what is happening, and I led sales development for about seven years, so I have watched this role get declared dead before. What AI is actually doing is stripping out the mechanical part of the job, the part that was never where the value lived. That does not eliminate the SDR. It exposes what the SDR was always supposed to be, and it puts a hard premium on the human judgment AI cannot replicate. The teams that win will be the ones that design that human-in-the-loop deliberately instead of pretending the machine can run the whole thing.

AI is good at the mechanical, bad at the judgment

Be honest about what the machine does well. AI can research an account faster than any human, draft a competent first-touch email, run a sequence, and handle the sheer volume of the top of funnel without fatigue. Those are real capabilities and they are not coming back to humans. If your SDR's main value was dialing and sending, that value is genuinely being automated, and no amount of nostalgia changes it.

But the machine is weak exactly where the job gets hard. It does not truly know why this account is in pain right now. It cannot read the hesitation in a prospect's voice and get curious about it. It cannot build a theory of a buying group, sense who the real economic buyer is, or catch the objection hiding under the stated one. It generates plausible outreach; it does not generate insight. The mechanical layer is solved. The investigative layer is not, and the investigative layer is where meetings that actually convert come from.

The role moves up, it does not disappear

So the SDR role does not vanish. It moves up the value chain to the work that was always the point: investigation, judgment, and real human conversation. The rep spends less time producing touches and more time deciding which accounts are worth a real theory, reading the signals a model flattens, and handling the moments where a person can tell they are talking to a person who has actually thought about their world.

This is the same argument I make about coaching reps as detectives rather than dialers, and AI sharpens it rather than softening it. When the machine handles the volume, the human's entire job becomes the detective work: the theory, the curiosity, the qualification. The reps who thrive in this next phase are not the ones who could send the most emails. They are the ones who can do the thinking the model cannot, on the accounts the model flagged.

Design the human in the loop on purpose

The failure mode is treating this as all-or-nothing: either keep a big human team dialing, or hand the whole funnel to an AI and hope. Both are wrong. The win is a deliberate division of labor where the machine does what it is good at and the human does what only a human can, with a clear handoff between them.

That is a design problem, and it is worth doing well. Let AI own the research, the first draft, the volume, and the never-tiring follow-up. Route to a human the moment the interaction needs judgment: a reply that does not fit the script, an account worth a real theory, a conversation where reading the person matters. Then coach your smaller human team relentlessly on exactly that judgment, because it is now the entire value they add. The org chart that results is smaller on humans, yes, but the humans in it are doing higher-value work than a sales development rep has ever done, and they are far harder for a competitor to copy than a bank of dialers ever was.

The shrinking headcount and the rising AI are not two halves of a story about replacement. They are the same shift toward a smaller, sharper, more human team sitting on top of a machine that handles the mechanical. The companies that name that shift and design for it will build something genuinely good. The ones that read it as "fire the SDRs, the robot has it" will automate their outreach and wonder why nothing converts, because they will have kept the part AI does well and thrown away the part that was actually working.

I led sales development for about seven years, and I write about where human judgment still wins as AI reshapes revenue teams. I also run a productized Revenue Engine Audit for leaders designing the human-and-machine engine that comes next.

Originally published at michaelinghilterra.com.

The forecast nobody scores

Michael Inghilterra — Fri, 09 Oct 2026 14:00:00 +0000

Four in five sales and finance leaders missed a quarterly forecast in a single year, and more than half missed two or more. That's from Xactly's 2024 benchmark survey of 405 North American sales and finance leaders.

A separate Gong survey of 2,015 US and UK business leaders landed in the same place. 81% of their companies missed a sales forecast in at least one quarter between early 2021 and late 2023. Only 19% hadn't missed once in seven quarters.

So the miss is normal. Markets move. Deals slip. A buyer goes quiet in week eleven. No forecasting method removes that.

What shouldn't be normal is what happens after.

Used every week, scored almost never

Most leadership teams I've worked with used the forecast every week and almost never scored it once the quarter closed. The call ends, the quarter ends, the number either landed or didn't, and the next quarter's forecast starts from a blank page.

That's strange when you think about it. We score reps on attainment. We score marketing on pipeline sourced. The one number that drives hiring plans, spend and board conversations often has no scorecard at all.

And a missed forecast isn't a spreadsheet problem. In Gong's survey, 42% of US respondents said they'd frozen hiring because of missed forecasts, and 28% of all respondents said they'd had to let people go.

The trap: the total can land while everything inside it misses

I owned BI and analytics for more than $850M in revenue across four regions. The question that changed how I ran it was a boring one: how accurate were we last quarter, and where exactly were we wrong?

The honest answer is usually uncomfortable, because the total hides the errors. Say two regions of the same size land 20% over and 20% under. The roll-up reads "on target." Leadership sees a clean number. Nobody sees that the three biggest deals in commit all moved.

On target in total. Wrong everywhere inside it.

Illustration. Region A lands 20 percent over forecast and Region B lands 20 percent under, so the company total reads on target even though both regions missed.

Hypothetical numbers.

A forecast that hits in total and misses on the deals that mattered isn't accurate. It's lucky. Luck doesn't repeat, and it teaches the organization nothing.

Sales and finance often don't even describe the miss the same way. In Xactly's survey, 52% of sales leaders said their forecasts miss by 10% or more, while 66% of finance leaders said theirs usually land within 9%. Two functions, two forecasts, two accuracy stories, and in my experience nobody reconciles them until the quarter is gone.

Three numbers, every quarter

Scoring a forecast doesn't need a new tool. It needs three numbers that you calculate the same way every quarter and put on the first slide of the next forecast call.

1. Total error. The forecast you called in week one against what actually closed. One number, no adjectives. Pick the snapshot date once and never change it, or the number becomes negotiable.

2. Error by region or segment. This is where a lucky quarter shows itself. If the total landed because two regions missed in opposite directions, you'll see it here, and you'll know which part of the forecast process needs work.

3. Commit-to-close. Of the deals in commit at the start of the quarter, how many closed that quarter? This number tells you whether "commit" means anything. If a third of commit slips every quarter, the category is a feeling, not a forecast.

Track all three for four quarters and you'll learn more about your forecast than a new model could tell you. You'll also have something most teams don't: a baseline. Without one, "we're getting better at forecasting" is a claim nobody can check.

Why this lands on RevOps

A few weeks ago I wrote that the RevOps seat is shifting from scorekeeper to operator: the function that used to measure the number is now asked to own it. The forecast is where that shift shows up first.

The function that used to build the forecast now answers for it. When the forecast misses, leadership doesn't ask for a better dashboard. It asks why the number was wrong and what you're doing about it. The only good answer to that question starts with a scored history.

If you're stepping into a bigger RevOps seat, this is the first thing I'd build. It's cheap, it's fast, and it moves the forecast conversation from opinions to evidence in a single quarter.

If you own the forecast: could you say today how accurate it was last quarter, deal by deal, and not just in total?

Sources

Xactly, 2024 Sales Forecasting Benchmark Report. Online survey of 405 North American sales and finance/RevOps leaders, fielded March 19 to 28, 2024 by Regina Corso Consulting. Press release, July 10, 2024: https://www.xactlycorp.com/company/press-room/new-forecasting-research-uncovers-need-collaboration
Gong, sales forecast accuracy report. Censuswide survey of 2,015 business leaders at privately held US (1,015) and UK (1,000) companies, January 3 to 9, 2024. Press release, January 30, 2024: https://www.gong.io/press/gong-report-finds-more-than-80-percent-of-companies-have-missed-revenue-forecasts-over-the-last-two-years

Originally published at michaelinghilterra.com.

The five silent failures breaking your revenue engine

Michael Inghilterra — Thu, 08 Oct 2026 14:00:00 +0000

Revenue operations is having a moment. VP of RevOps titles have grown roughly 300 percent in eighteen months, and LinkedIn now ranks it among the fastest-growing roles in the United States (per LinkedIn hiring data reported across 2025). The base is still small, so this is momentum rather than mass, but the direction is unmistakable: companies are waking up to the fact that revenue predictability is an engineering problem, not a motivational one.

Here is what most of them find when they finally look under the hood. The forecast misses, the reporting does not drive decisions, and everyone reaches for a new tool. But the real damage is almost never one dramatic failure. It is five quiet ones, compounding, and most teams are living with all five without having named a single one. Here they are, in the order they usually bite.

One: your CRM data is quietly rotting

Every downstream number inherits the quality of the records underneath it. Forecasts, territories, comp, reporting, all of it sits on top of your CRM data, and no dashboard on top can fix a bad foundation.

The failure is rarely dramatic. It is field completeness that quietly decays, records nobody has touched in ninety days still counted as active, deals owned by people who left, and an economic buyer who is not in the system at all. Clean it once and it re-rots, because the fix is not a cleanup project. It is enforcement at entry: required fields to move a deal forward, and a standing sweep so staleness never accumulates.

Two: your process taxes every deal

Broken process is a tax on every deal, spread thin enough to ignore. The classic case is an approval routed to a senior exec who rubber-stamps 98 percent of quotes, adding days to every deal and changing nothing. The days hide because cycle time is rarely decomposed, so nobody sees where they actually go.

The signal to watch is exception volume. When reps route around the process, the process is wrong, not the reps. Their workaround is a map of where the design failed. And the biggest leaks are not inside stages but in the handoffs between owners, the silence between SDR and AE where a qualified deal sits while each side assumes the other has it.

Three: everyone counts differently

If each region computes open pipeline, forecast, and activity its own way, leadership ends up arguing about whose report is right instead of what to do. Six regions, six definitions of pipeline, and a roll-up total that means nothing.

The tell is a forecast meeting that runs long and generates heat instead of decisions, because people are silently using different definitions and talking past each other. The fix is cheap and boring and enormously effective: a one-page metrics dictionary, one canonical definition per metric, owned and used everywhere. It removes the argument at its source.

Four: your reporting is decoration

A dashboard that does not change a decision is decoration. Most reporting shows the number, colors it, and stops, leaving the viewer to do exactly what they were going to do anyway.

The test is two questions the viewer should answer in one breath about every tile: why does this matter, and what do I do about it. If they cannot, the tile is noise. And the honest measure of a reporting function is not how much it produces but how much gets opened. The most-built, least-opened report in your company is evidence the reporting optimizes for looking comprehensive instead of driving action.

Five: your pipeline is part fiction

Forecast accuracy is downstream of pipeline reality. Fluff, sandbagging, and unqualified commits turn the forecast into a hope with a number attached. Dead deals padding coverage, your best deals deliberately misdated, and commits with no identified economic buyer underneath them.

This is usually the root cause of failures three and four. You cannot report your way out of a pipeline that is not true, and you cannot trust a forecast built on deals that were never real. The fix is evidence gates on stage entry and one firm rule most pipelines break: no economic buyer, no commit.

They compound, which is why they are worth naming

The reason these are silent is that each one is survivable alone. Together they form a chain. Rotting data feeds inconsistent definitions, which feed reporting nobody trusts, which sits on a pipeline that is partly fiction, and the forecast at the end of that chain misses for reasons no single dashboard can explain. Buying another tool does not fix a chain. Naming the weak links does.

Most teams are surprised by how many of the five they have. The good news is that they are all fixable, and the fixes compound in the same direction the failures do. You do not have to solve everything. You have to find the few links quietly costing you the most, and start there.

I built a short self-assessment so you can score your own revenue engine against these five. I also run a productized Revenue Engine Audit that does it rigorously: a scorecard across all five dimensions plus a prioritized fix plan. If your forecast is not trusted, the cause is almost always somewhere in these five.

Originally published at michaelinghilterra.com.

Ramp that actually sticks

Michael Inghilterra — Wed, 07 Oct 2026 14:00:00 +0000

Ask most sales development teams how long a new rep takes to get productive and you will hear a number like six months, delivered with a shrug, as if it were weather. It is not weather. It is the predictable result of treating onboarding as an event instead of a system, and it is one of the most expensive things a team quietly tolerates.

I led sales development for about seven years, and slow ramp was the problem I underestimated the longest. When you can hire steadily, you paper over it: a rep who takes half a year to contribute is just a cost you absorb while the pipeline gets filled by the people already ramped. Take away easy hiring, which most teams have had to, and ramp becomes the highest-leverage thing you can fix, because now the only way to grow output is to make each new rep productive faster and keep them.

Onboarding is an event; ramp is a system

Here is the distinction that changes everything. Onboarding is the week of orientation: the tools, the product overview, the certifications, the swag. Ramp is the months-long journey from that first week to actually carrying weight. Most teams invest heavily in the first and leave the second to chance.

So the new rep finishes a polished onboarding week, gets pointed at the phones, and then disappears into a fog. Nobody quite knows how they are doing until a number is either there or not there a quarter or two later. By then it is late to intervene, and the rep has spent months building habits, some good, some quietly wrong, with no checkpoint to catch the difference.

A ramp that sticks is a system with structure the whole way through, not a strong start followed by a void.

Build ramp as checkpoints, not a calendar

The core of it is simple: define the path from day one to full productivity as a sequence of checkpoints, and make each checkpoint a competency the rep has to demonstrate, not a box to check or a day on a calendar.

Not "week three: cold calling." Instead, "the rep can run a live discovery conversation that surfaces a real pain, and here is what good looks like." Not "completed the objection-handling module," but "the rep can handle the objection under the objection on a recorded call." The checkpoint is a demonstration, observed, against a standard everyone can see.

That one shift does something powerful: it makes ramp legible. When progress is a set of demonstrated competencies, you can look at any new rep and know exactly where they are and exactly what is next. You see the rep who sailed through research and account selection but is stuck on discovery, and you can put your coaching precisely there, in week five, instead of discovering the gap in month six when the pipeline comes up short.

Legible ramp is early-warning for retention, too

The hidden payoff is that a checkpoint system tells you when a rep is struggling long before the results do. A rep stuck at the same competency for weeks is a signal, and it is a signal you can act on while it still matters: more coaching, a different approach, an honest conversation. Without checkpoints, that same rep just looks fine until suddenly they do not, and by then you are often past the point where help lands.

This matters because the other half of the ramp problem is retention. Reps who feel lost and unsupported in their first few months are the ones who leave at month ten, right as they should have been hitting their stride. A ramp that makes progress visible does not just speed people up, it holds onto them, because a rep who can see themselves getting better and can feel the team investing in that progress has a reason to stay. Legible progress is its own kind of retention.

What good ramp is worth

Put it together and the case is straightforward. A ramp built as demonstrated competencies gets reps productive faster, because the path is clear and the gaps get caught early. It makes coaching precise, because you always know the one thing to work on next. And it keeps people longer, because visible progress and real support are what turn a burnout seat into the start of a career.

None of this requires a platform or a big budget. It requires deciding what the checkpoints are, what good looks like at each one, and who observes them. That is a weekend of design work and a standing commitment, and it pays back every single time you hire. The teams still shrugging about six-month ramp are not stuck with it. They just never built the system that would have made it three.

I led sales development for about seven years, and I write about building revenue teams that ramp fast and stay. I also run a productized Revenue Engine Audit for leaders who want the systems under their team to be as deliberate as the hiring.

Originally published at michaelinghilterra.com.

The Quote Approval That Costs You Three Days a Deal

Michael Inghilterra — Tue, 29 Sep 2026 13:00:00 +0000

Somewhere in your quote-to-cash process there is probably an approval step that adds days to every deal and changes almost nothing. I have seen this pattern in company after company: quote approvals routed to a senior executive who approves 98 percent of them without a single edit, waves them through between meetings, and never sees the cost of the wait they are creating.

Nobody measures that cost, which is exactly why it survives. The approval feels like control. What it actually is, most of the time, is a three-day tax on every deal, paid in cycle time, collected by a bottleneck that rubber-stamps its way through the queue.

The tax nobody puts on the books

Broken process is a tax on every deal, and like most taxes it is easy to ignore because it is spread thin. No single deal screams. Each one just sits an extra couple of days waiting on a signature that was never in doubt. Multiply that across every deal in a quarter and you have real revenue moving slower, reps sitting on their hands, and an executive spending calendar time on approvals that a threshold could have handled.

The reason it hides is that cycle time is rarely decomposed. Everyone knows deals take a while. Almost nobody knows where the days actually go. When you finally lay the quote-to-cash flow out and measure each step, the surprise is never the negotiation. It is the queue in front of an approver who approves everything.

The workaround is the diagnosis

Here is the part most process conversations get backwards. When reps route around a process, the instinct is to enforce compliance, to remind everyone to follow the steps. That is treating the symptom.

Exception volume is a diagnostic. Every workaround your reps invented is them telling you, in the only language they have, where the design failed. If a large share of deals are going through as exceptions, or reps are quietly finding the fast path around the official one, the process is wrong, not the reps. They are not being difficult. They are being efficient in spite of you.

So when I look at operational design, I count exceptions on purpose. A high exception rate is not a discipline problem to stamp out. It is a map of where the process fights the work, and it points straight at the steps worth redesigning first.

Deals stall in handoffs, not stages

The other place the tax hides is between owners. Everyone watches for deals stuck in a stage. Almost nobody watches the handoffs: SDR to AE, AE to customer success, sales to deal desk. The real leaks live in those gaps, and no stage report shows them, because a handoff is not a stage. It is the silence between two stages while each side assumes the other has it.

A deal does not usually die inside a well-run stage. It dies in the day it sat after the SDR marked it qualified and before the AE picked it up. Those days are invisible on a funnel chart and expensive in reality.

The fix is thresholds, then redesign

The quick win is almost embarrassingly simple: approval thresholds, so only genuine exceptions escalate. If 98 percent of quotes get approved unchanged, then 98 percent of quotes should not require that approval at all. Set a threshold, auto-approve everything inside it, and route only the real outliers to a human. The executive gets their calendar back and the deals get their days back, in the same move.

The foundational version is to redesign the quote-to-cash flow around cycle time and clear ownership, so the right approver sees the right things at the right threshold and the handoffs have an owner instead of a gap. But you do not need the full redesign to stop the bleeding. You need to find the step that approves everything and ask why it is a step at all.

Start by measuring where the days go. The bottleneck is usually hiding in plain sight, wearing the costume of control.

I write about making revenue operations legible, and I run a productized Revenue Engine Audit that scores operational design alongside four other dimensions. The approval that rubber-stamps everything is one of the first things I go looking for.

Activity Is Not the Outcome

Michael Inghilterra — Fri, 25 Sep 2026 16:21:39 +0000

Most sales development teams manage the wrong number. They manage activity, dials, emails, touches, because activity is easy to count and easy to demand. But activity is an input, not an outcome, and confusing the two is how you end up with a team that is very busy and a pipeline that is very thin.

I led sales development for about seven years, and the trap caught me early. When results are soft, the reflex is to turn up the volume: more calls, more sequences, more activity on the board. Sometimes it works, because you were under the minimum. Usually it does not, because volume was never the constraint. The constraint was that the activity was not the kind that leads anywhere. You cannot fix a quality problem by adding quantity, and a dial count will never tell you which one you have.

Why activity became the default metric

Activity metrics win by default for a simple reason: they are the easiest thing in the whole funnel to measure. A dial is a fact. An email sent is a fact. You can put them on a dashboard by lunchtime and hold a standup around them tomorrow.

Outcomes are harder. A qualified meeting that becomes real pipeline is the thing you actually want, but it lags, it depends on the AE's judgment, and it is messier to attribute. So teams quietly substitute the easy proxy for the hard truth and manage the proxy. The dashboard fills with green, the standups have numbers to discuss, and everyone feels productive, whether or not any of it is producing.

That is the reporting failure underneath the activity trap. A dashboard that does not change a decision is decoration, and a dashboard full of dial counts usually changes nothing except how guilty a rep feels. It measures effort, not effect.

Find the leading indicators that actually predict pipeline

The fix is not to throw activity metrics away. It is to find the specific activities that actually predict qualified pipeline for your motion, and manage those instead of raw volume.

This is an analysis, not a guess. Look back at the meetings that became real, qualified pipeline, and work backwards. What did the reps who sourced them actually do differently? It is almost never "made more dials." It is usually something more specific: multi-threading into more than one contact at an account, personalizing the first touch off a real trigger, reaching a particular seniority of title, following up a precise number of times before giving up. Those are the leading indicators worth their name, because they have a real relationship to the outcome, not just a place on a spreadsheet.

Once you know them, you can coach and measure them. "Multi-thread every target account within the first week" is a behavior that predicts pipeline. "Make 60 dials" is a behavior that predicts fatigue. One of those belongs on the scorecard.

Build the scorecard around effect, not effort

So the SDR scorecard I want has almost no raw activity on it. It has the handful of leading indicators that actually correlate to qualified pipeline, plus the outcome itself. Activity is still visible underneath, as a diagnostic when something looks off, but it is not the thing the team is managed to.

That change does two things at once. It points rep effort at the behaviors that work instead of the behaviors that are easy to count, and it makes the reporting honest, because now the top-line number on the board is one that actually moves the business. When a rep is behind, the conversation is about which predictive behavior is missing, not about why their dial count dipped on Tuesday.

The point is not that effort does not matter. Of course it does. The point is that effort pointed at the wrong activity is just noise you paid for, and the only way to tell the difference is to measure the activities that predict the outcome, not the activities that are easy to see. Manage the outcome, coach the leading indicators, and let the dial count go back to being what it always was: a diagnostic, not a destination.

I led sales development for about seven years, and I write about measuring revenue teams by what actually predicts the number. I also run a productized Revenue Engine Audit for leaders who want their reporting to drive decisions instead of just displaying activity.

Six Regions, Six Definitions of Pipeline

Michael Inghilterra — Thu, 17 Sep 2026 13:00:00 +0000

The most expensive meeting in a lot of companies is the one where leadership argues about whose number is right. Not what to do about the number. Whose number is even real. I have sat in that meeting more than once, and it almost always traces back to a single unglamorous cause: the regions are each computing the basics differently, so nobody is comparing the same thing.

Six regions, six definitions of "pipeline." One counts everything with a dollar amount. One counts only deals past a qualification stage. One includes renewals, one does not. One dates by created, one by expected close. Each definition is defensible on its own. Together they guarantee that when the six roll up, the total means nothing, and everyone in the room can feel it even if they cannot name it.

The tell is the argument itself

You know you have this problem when the forecast call runs long and generates heat instead of decisions. People are not really disagreeing about the business. They are silently using different definitions and talking past each other, each certain their own report is the honest one.

That is the quiet cost. Not just wasted meeting time, though there is plenty of that. It is that leadership stops trusting any of the numbers, because they have learned the numbers do not reconcile. And once the forecast is not trusted, every downstream decision gets slower and more political, because it has to survive a debate about the data before it can even be a debate about the choice.

Why this is upstream of everything

This is what I look for when a team tells me their reporting does not drive action or their forecast is not believed. Usually the reporting is fine and the forecast math is fine. The definitions underneath them are not shared. You cannot build trusted reporting on top of metrics that mean six different things, any more than you can average six currencies without agreeing on an exchange rate.

It also compounds. Every new dashboard, every new region, every new leader inherits the ambiguity and often adds a seventh definition trying to fix it. The drift is not malicious. It is what happens when definitions live in people's heads instead of in one governed place.

The one-page fix

The good news is that the highest-leverage move here is also one of the cheapest in all of revenue operations: a metrics dictionary. One page. One canonical definition per metric. Open pipeline, forecast categories, activity, win rate, conversion, stage meanings, each defined once, owned by someone, and used everywhere.

It sounds too simple to matter. It matters more than almost anything else you can do in a quarter, because it removes the argument at its source. When everyone computes open pipeline the same way, the forecast call stops being a definitions fight and becomes an actual conversation about the business. The dictionary does not need a platform or a project. It needs a decision and an owner.

The foundational version, once the dictionary exists, is to push those definitions down into the systems themselves, a governed semantic layer so every report draws from the same source instead of each analyst rebuilding the logic their own way. But that is the upgrade. The dictionary is the fix, and you can start it this week.

What changes when the number means one thing

When there is one definition per metric, the reconciliation cycles disappear. The regions roll up to a total that actually means something. Leadership argues about the business instead of the data. And the forecast becomes something people are willing to stand behind, because for the first time they are all standing on the same ground.

If your forecast meeting is long and hot and never quite lands, listen for the tell. It is probably not a disagreement about strategy. It is six definitions in a room pretending to be one.

I write about making revenue operations legible, and I run a productized Revenue Engine Audit that scores single-source-of-truth alongside four other dimensions. A metrics dictionary is usually the first thing I hand a client.

Coach Your SDRs Like Market Detectives

Michael Inghilterra — Tue, 15 Sep 2026 13:00:00 +0000

The best sales development rep I ever managed was not the smoothest talker on the team. She was the most curious. Before a call she knew why the account might be in pain, what had changed there recently, and who actually felt it. She did not open with a pitch. She opened with a question that told the prospect she had already been paying attention. Her meetings booked at a rate the polished talkers could not touch, and the meetings she booked were real.

That is when it clicked for me: the job of an SDR is not to talk. It is to investigate. And if you coach them as dialers, you get dialing. If you coach them as detectives, you get intelligence, and intelligence is what actually moves a deal. Here is what coaching for that looks like.

Detectives build a theory before they knock

A dialer starts with a list and a script. A detective starts with a theory of the case: given what I can see about this account, here is why they might have this problem right now, and here is who would feel it.

Coaching that theory is a specific habit. Before the reps ever touch the phones, I have them answer three questions about an account out loud. What changed here recently that could create the pain we solve. Who in this org lives with that pain day to day. What would I expect to be true if my theory is right. It takes minutes and it transforms the call, because now the rep is testing a hypothesis instead of reciting a pitch, and prospects can feel the difference immediately. One is a stranger reading a script. The other is someone who did their homework.

The evidence is in the objection under the objection

Detectives listen for what does not fit. When a prospect says "we're all set" or "not right now," a dialer hears a wall and moves on. A detective hears a clue and gets curious about it, because the stated objection is almost never the real one.

So I coach reps to treat objections as evidence, not endings. "We're all set" often means "I don't yet see why this is worth my time," which is a completely different problem to solve. The skill is the gentle follow-up that surfaces the objection under the objection, and it is entirely coachable. On call reviews I am not listening for whether the rep overcame the objection. I am listening for whether they got curious about it. The ones who investigate the pushback instead of steamrolling it are the ones who find the real deal hiding behind the reflex.

Call review is the crime scene

Coaching detectives happens on the tape. A skill ladder tells a rep what to get good at; call review is where they actually get good at it, because it is specific and it repeats. Generic feedback like "sound more confident" changes nothing. "Right here, she gave you a reason and you moved past it, what do you think she actually meant" changes everything, because it is a concrete moment attached to a concrete skill.

The discipline is reviewing the same competency repeatedly until it holds, rather than spraying a dozen tips across one call. Pick the one investigative move the rep is missing this week, find three examples of it in their calls, and work only that. Depth beats breadth. A rep who truly internalizes one detective skill a month is transformed in a quarter. A rep who gets a scattershot of tips every week is just being talked at.

Intelligence is the real output

Here is why this matters beyond the SDR seat. When you coach reps as detectives, their output is not just meetings. It is intelligence: why this account is in pain, who the economic buyer really is, what the buying process looks like, what the competition is doing. That intelligence rides along into the AE's hands and makes the whole deal better qualified from the first real conversation.

A dialer hands off a name and a calendar invite. A detective hands off a case file. One of those makes the AE's job harder and one makes it easier, and over a year the difference in what actually closes is enormous. The meetings look similar on a dashboard. What is underneath them is not.

Curiosity is a hiring and coaching signal

If investigation is the job, then curiosity is the trait to hire for and the skill to coach. In interviews I care less about polish and more about whether a candidate gets genuinely interested in a problem they cannot immediately solve. On the team, I reward the rep who brings back a sharp insight about an account over the rep who simply hit their dial count, because the insight is the leading indicator of the pipeline I actually want.

Sales development got a reputation as a volume game because too many teams coached it as one. The teams doing it well have quietly moved on. They are not building dialers. They are building detectives, and the intelligence those reps generate is worth far more than the noise the old model mistook for productivity.

I led sales development for about seven years, and I write about coaching revenue teams that produce intelligence, not just activity. I also run a productized Revenue Engine Audit for leaders who want the pipeline those teams feed to be as rigorous as the reps filling it.