DEV Community: Rachel_EShell

What a DevRel agency costs in 2026, and what it actually does

Rachel_EShell — Tue, 15 Sep 2026 16:00:08 +0000

Search for a devrel agency and you get a page of firms all claiming to speak developer. The useful question is what a DevRel programme is supposed to produce, and what the published numbers say it costs.

Developer marketing agencies price between $3,000 and $25,000 a month, and funding stage decides the band. Flat-retainer models start near $1,250 a month for a single channel, and the full-time alternative costs $200,000 to $260,000 a year all-in.

What a devrel agency does

Developer relations is the discipline of building relationships with the developers who use a product, through technical content, community, documentation, conferences and product feedback. The line that matters is between that and developer marketing. DevRel makes developers successful and trusts adoption to follow. Developer marketing creates demand and drives conversion (saashero.net, 2026; moesif.com, 2026).

A DevRel agency sells execution inside the first definition, staffed by people who can write code and distributing through GitHub, Stack Overflow, Hacker News and developer newsletters rather than LinkedIn ads. The specialist exists because developers evaluate tools through documentation, peer recommendations and code, and run ad blockers at roughly three times the rate of general business buyers (saashero.net, 2026). 47% of enterprise technology buyers now begin vendor research with an AI assistant, which makes the pages a model can quote part of the shortlist process (business.daily.dev, 2026).

The 2026 price bands

Stage / ARR	Typical DevRel agency range	What it buys
Seed (under $3M ARR)	$3,000 to $10,000 per month	SEO foundations, technical audits, a couple of pillar pieces per quarter
Series A ($3M to $15M ARR)	$10,000 to $25,000 per month	Original research, GEO and AEO work, technical posts, competitor conquesting
Series B+ (over $15M ARR)	$15,000 to $35,000+ per month	Custom research, CRO, dedicated account director, multi-channel scaling
Fractional consultant	A few thousand per month	Senior judgement on one build, no ongoing team
Full-time senior DevRel hire	$200,000 to $260,000 all-in per year	Day-to-day ownership once the work justifies it

The retainer bands sit against a hiring benchmark that makes the agency route rational for most funded companies. A full-time senior developer advocate costs $200,000 to $260,000 all-in annually, and the 10th annual State of Developer Relations report puts median base salary at $175,000 (orbytjobs.ai, 2026; develocity.io, 2026). Strategic Nerds puts a full DevRel function at $250,000 to $500,000 a year, which is the number to hold in mind when an $8,000 monthly quote starts to feel large (strategicnerds.com, 2026).

The pricing model matters as much as the price. Percentage-of-spend billing gives an agency a reason to grow your budget regardless of efficiency. A flat monthly retainer decouples the fee from ad volume, so every recommendation rests on performance data (saashero.net, 2026).

The other structural choice is generalist against vertical specialist. Generalist B2B agencies underperform for devtools because their defaults do not match how developers evaluate software. Vertical specialists hold distribution in developer-native channels and know the difference between a demo request and a free-trial activation (marketerhire.com, 2026; saashero.net, 2026).

The pitfalls that eat the budget

Vanity metric reporting. Blog views, Discord member counts and conference attendance are the easiest things to report and the hardest to tie to revenue. DevRel teams were cut disproportionately in the 2023 and 2024 layoff waves because leadership could not connect community engagement to pipeline (techmeetups.io, 2026). Long lock-in contracts and junior execution after a senior pitch are the other two patterns. Ask for the named senior strategist, their weekly hours, and confirmation they handle no more than eight clients at once (saashero.net, 2026).

How to measure it

Track Net New ARR from developer-sourced or developer-influenced opportunities in CRM, with payback period as the discipline behind it: reducing time-to-value by 30% can lift free-to-paid conversion from 5% to 7%, a 40% revenue increase without extra acquisition spend (business.daily.dev, 2026). Activation metrics such as time-to-first-successful-API-call and SDK adoption by language are the leading indicators (moesif.com, 2026).

The takeaway

DevRel agency pricing in 2026 is knowable: seed programmes at $3,000 to $10,000 a month, Series A at $10,000 to $25,000, Series B+ above $15,000, against a full-time hire at $200,000 to $260,000 a year. What separates proposals is the billing model and whether the agency reports pipeline or reach.

The full breakdown is on the EShell blog: What a DevRel agency costs in 2026

SaaS lead generation agency pricing in 2026: the models behind the number

Rachel_EShell — Fri, 11 Sep 2026 03:19:29 +0000

Ask five lead generation agencies what they charge and you will get five answers, and more than one will be "book a call." Pricing is a competitive advantage, and vague pricing pages are a feature. But the public numbers do exist if you assemble them from enough 2026 sources, and the spread is the story: managed B2B lead generation runs from about $2,500 to $15,000+ per month, and a qualified lead can cost anywhere from $84 to $400+ depending on industry, channel, and how strictly you define qualified (saleshive.com, updated July 2026).

The spread is not quality alone. It is three things: the pricing model, what the price includes, and whether the agency measures cost per lead or cost per opportunity.

The published bands

Managed retainer programs run roughly $2,500 to $15,000+ per month, with most monthly retainers between $3,000 and $12,000 (saleshive.com, 2026). Cleverly's 2026 guide splits it further: single-channel cold email at $2,000 to $8,000, LinkedIn-led programs at $3,000 to $10,000, multi-channel outbound at $5,000 to $15,000+, and enterprise at $15,000 to $40,000+ (cleverly.co, March 2026). Cost per lead models run $200 to $500 per qualified lead. Pay per meeting runs $150 to $600 for mainstream targets, and above $900 for enterprise. For comparison, a fully loaded in-house SDR costs roughly $110,000 to $160,000 per year and takes three to six months to become productive (saleshive.com, 2026).

The four models

A monthly retainer is a fixed fee for an agreed scope, the most common structure, and it buys predictability. Its risk is paying for activity instead of pipeline when the success metric stays vague. Cost per lead charges only for leads that meet defined criteria, which shifts risk to the vendor but creates a volume incentive. Pay per appointment charges only when a qualified meeting is booked, which shifts more risk but rewards volume over quality at scale unless qualified is airtight. A hybrid combines a smaller base retainer, often 40% to 60% of a full retainer, with performance bonuses. Skip commission-only: no sustainable agency funds data, infrastructure, and reps on closed-deal commissions alone.

Cost per lead vs cost per opportunity

This is the comparison most proposals avoid. SaaSHero's 2026 benchmarks put healthy blended CPL for B2B SaaS at $137 to $237, but the number that predicts performance is cost per opportunity: about $860 for organic channels up to $2,968 for LinkedIn, with healthy payback periods of one to three months (saashero.net, 2026). A $50 lead that closes one in a hundred costs $5,000 per opportunity. A $300 lead that closes one in ten costs $3,000. The rule of thumb across the guides is that a healthy CPL stays under 10% to 20% of your annual contract value.

What drives the price

Channel is the biggest lever. Multi-channel prospecting averages around $188 per lead, LinkedIn ads average $408, and trade shows top out near $840 (saleshive.com, 2026). Target complexity is next: a tight ICP costs more to reach than broad targeting, and enterprise campaigns often cost two to three times more per meeting than SMB ones (cleverly.co, 2026). Then come the hidden costs that ambush budgets. Setup fees of $1,500 to $5,000 are common, and extra domains, data enrichment, and tool subscriptions can add 30% to 50% on top of a base retainer, turning a $5,000 engagement into $7,500.

What to demand before signing

A written qualification definition in the main contract, with a rejection window and a replacement policy. One all-in monthly number. Reporting anchored to pipeline and closed-won revenue, not impressions. Month-to-month or performance-linked terms over long lock-ins. And a clear answer on whether the agency has ever recommended cutting budget for efficiency, because that answer tells you what the fee is really aligned to.

The full breakdown with the pricing table, FAQ, and all sources is on the EShell blog: SaaS Lead Generation Agency Pricing in 2026

Agency prices for cold email finally make sense in 2026

Rachel_EShell — Wed, 09 Sep 2026 14:56:57 +0000

I have watched the same founder conversation for years: 400 emails sent by hand, four replies, then agency websites showing prices from $397 to nearly $10,000 a month for what looks like the same service. It is not price gouging. The published 2026 pricing finally makes the difference readable.

The market runs from about $1,500 a month for entry-level done-for-you sending to $10,000+ for enterprise programs, with most serious engagements between $2,500 and $6,000 a month (outboundsystem.com, July 2026; leadium.com, 2026). Setup fees of $1,000 to $3,000 are typical, and infrastructure that proposals quietly exclude adds $500 to $2,000 a month on top (leadium.com, 2026).

The three bands are fairly consistent across sources:

Under $2,500 a month: sending capacity. Template copy, basic data sourcing, thin infrastructure. Fine as a cheap test, not the same product as the tier above (outboundsystem.com, July 2026).
$2,500 to $6,000 a month: the real work. Custom copy, verified data, deliverability infrastructure, a dedicated strategist, honest reporting. Multiple guides call this the sweet spot for most B2B companies (outboundsystem.com; leadium.com; cleverly.co).
$5,000 and up: multi-channel. Email plus LinkedIn and sometimes calling on one accountable team, with vendors like Belkins and Callbox publishing $5,000 to $15,000 ranges (sequenzy.com, 2026).

Published vendor prices, read off their own sites in August 2026, include Belkins from $5,000 a month, CIENCE at $7,499 for the first month ($5,000 setup, $2,000 a month team, $499 a month platform), SalesRoads at $9,950 per four-week engagement, and Cleverly from $397 a month for LinkedIn outreach with $1,995 for cold email (outreachbloom.com, retrieved September 9, 2026).

The number that matters more than price is the denominator. Cold email reply rates have no single value because the denominator decides the result. Belkins' study of 7.5 million emails reports a 0.45% average against total sends and notes that 5% against openers can describe the same campaign (belkins.io, June 2026). Instantly's platform average is 3.43% (instantly.ai, 2026). Woodpecker's 20-million-email analysis splits personalization: 18% for personalized campaigns versus 9% for templates (woodpecker.co, 2026).

For agency evaluation this is decisive. If the pitch leads with volume, the data argues against you: campaigns of 50 or fewer recipients average 5.8% reply rates versus 2.1% for campaigns of 500+ (Belkins data summarized at autobound.ai, 2026). The agencies winning in 2026 sell smaller, better-targeted campaigns with signal-based personalization, and they publish their numbers against total sends.

Red flags worth keeping: infrastructure excluded from the retainer, commission tiers without qualification definitions, no written deliverability standard, and guarantees on the wrong metric. A serious agency will put a spam-rate standard under 0.10% with a kill-switch in writing (leadium.com, 2026).

Ask five questions before signing: what denominator do you report replies against; is infrastructure in the price and do we own the domains; what is your written spam-rate standard; show two current same-ICP clients at this price tier with honest numbers; and what happens at month three if the trend is flat.

The DIY line is more generous than most people expect. Under about $2,000 a month, tools plus a small targeted list is usually the better test. In the $2,500 to $6,000 band, an agency earns its fee through verified list strategy, deliverability operations, and reporting you can audit. Whatever you choose, insist on the total-sends denominator, because it is the only one you can verify independently.

The full version with the pricing table, FAQ, and all sources is on the EShell blog: B2B Email Outreach Agencies in 2026: Real Costs and Red Flags

SEO for SaaS in 2026: You Are Playing Three Engines, Not One

Rachel_EShell — Tue, 08 Sep 2026 15:35:13 +0000

I write marketing content for software companies, and the hardest conversation I have with founders right now starts with a dashboard. Their SEO report shows rankings climbing, and they cannot understand why AI answers still never mention them. The data says those two things stopped being the same game.

Only 38% of pages cited in Google AI Overviews also rank in the top 10 organic results, and 36.7% of cited URLs do not rank in the top 100 at all, based on Ahrefs' analysis of 863,000 keyword SERPs and 4 million AI Overview URLs (ahrefs.com/blog/ai-overview-citations-top-10/, March 2026). Ranking is no longer the gate to being cited. That is the difference between engine one and engine two.

Engine One: Google Organic Is Still the Traffic Leader

Traditional SEO still pays the bills. Organic search remains the highest-intent channel for most SaaS products, and the technical floor has not changed: pages must be crawlable, must render without JavaScript-only text, must pass Core Web Vitals (LCP under 2.5s, INP under 200ms, CLS under 0.1 per web.dev/vitals), and must keep canonical URLs clean. The trap is stopping there. Many teams spend a year reaching page one for money keywords and then discover AI answers do not cite page-one results anyway.

Engine Two: AI Answers Run on Freshness and Quotability

AI engines evaluate sources differently. An outdated fact in a generated answer is a direct quality failure for the model, so recency is weighted aggressively. Pages untouched for more than three months are over 3x more likely to lose AI citations than recently refreshed pages, and more than 70% of cited pages were updated within the past 12 months (AirOps and Kevin Indig, "The 2026 State of AI Search," airops.com, December 2025). Content updated within 30 days earns 3.2x more AI citations than older content (ConvertMate, 80 million citations, 2026).

Quotability is the second property. Models cite pages they can lift clean answers from: the answer near the top of each section, sequential headings, tables and lists that survive extraction, and every number traceable to a named source. A pricing page written for humans only is unquotable in AI terms. The operational change is a refresh cadence, commercial pages quarterly, evergreen annually, plus a quotability pass on every page.

Engine Three: Brand Presence Predicts Citations Better Than Links

The uncomfortable 2026 data: across 75,000 brands, Ahrefs measured branded web mentions at 0.664 correlation with AI Overview visibility, branded anchor text at 0.527, and branded search volume at 0.392, versus 0.218 for backlinks (ahrefs.com/blog/ai-overview-brand-correlation/, 2026). The distribution is winner-takes-most: the top quartile of brands by mention volume averages 169 AI Overview mentions versus 14 for the next quartile. AI engines were trained on the whole web, so they can observe which entities people actually search for and mention. Link graphs were Google's proxy for reputation. AI engines do not need the proxy.

For a small team this means earned presence: digital PR that puts the brand name in context, genuine community answers, review platform presence, and professional articles. Community content also remains a direct citation source on Perplexity and Google AI Overviews, and LinkedIn articles climbed to rank 5 on ChatGPT by early 2026 (GEORaiser, Q1 2026 cross-platform analysis).

The Dashboard That Sees All Three

A 2026 report needs three layers: Google rankings and clicks, AI citations tracked per engine over time, and brand signals such as branded search volume and earned mentions. Without the citation layer, none of the other work is observable. Pages hold rankings while losing citations, and the reverse happens too.

Sequencing on a Small Team

Fix the technical floor and instrument citation tracking first (month 1-2). Run the quotability and freshness pass on pricing, comparison, and top-of-funnel pages (month 2-4). Start the earned-presence loop with two to four digital PR or community plays per month (month 4-6). Refresh on cadence forever, and let the citation layer decide what to double down on. The engines compound in that order: technical soundness keeps you eligible, quotable content makes you citable, and brand presence decides whether the model reaches for you at all.

The full canonical version, with the FAQ and sources, is on the EShell blog; this is the condensed engineering read.

Sources: Ahrefs (ahrefs.com/blog/ai-overview-citations-top-10/, March 2026; ahrefs.com/blog/ai-overview-brand-correlation/, 2026); Google Web Vitals (web.dev/vitals); AirOps and Kevin Indig (airops.com, December 2025); ConvertMate (80M citations, 2026); GEORaiser (georaiser.com, Q1 2026 analysis).

Full article with FAQ and sources: The 2026 Three-Engine Playbook

Your Six-Month-Old Post Is Quietly Losing AI Citations

Rachel_EShell — Sun, 06 Sep 2026 13:02:17 +0000

If you publish technical content and nobody is updating last quarter's posts, the 2026 citation data says you are losing ground silently.

AirOps, working with Kevin Indig, tracked millions of AI answer data points and found that pages which go more than three months without an update are over 3x more likely to lose their AI citations than recently refreshed pages. More than 70% of all pages cited by AI search engines were updated within the past 12 months, and more than half within six months (source: AirOps, "The 2026 State of AI Search," airops.com/report/the-2026-state-of-ai-search, December 2025).

I write content for software companies, and this changed how we treat every page we ship. Here is what the data says, in engineering terms.

Citations stopped following rankings

The old model assumed AI answers were a new front end on the ranked index. The numbers disagree:

Roughly 60% of Google AI Overview citations come from URLs that do not rank in the top 20 organic results (AirOps, 2025).
Only about 17% of AI Overview citations come from content ranking in the traditional top 10; most come from positions 21-100 or beyond (BrightEdge citation tracking, February 2026, cited in jarredsmith.com).
About 48% of citations come from community platforms such as Reddit and YouTube, and 85% of brand mentions originate on third-party pages (AirOps, 2025).

So ranking is no longer the gate. Being quotable and current is.

Freshness is the gate

For commercial queries the bar is higher: about 83% of commercial citations come from pages updated within the last year, and more than 60% from pages refreshed within six months. In SaaS, finance, and news, pages older than three months see steep citation drops (AirOps, 2025).

That last point is the one that should worry technical teams: a pricing page, a comparison page, or a "best tools" roundup has an expiration date measured in quarters, not years.

What high-citation brands do

HubSpot's AEO research team analyzed citation data across six-plus answer engines and surveyed more than 4,000 marketers for its State of AEO report. Their conclusion: the pattern is quotability. As Botify's AJ Ghergich put it, "You don't rank in AI. It's stochastic." You can only become the kind of source an engine reaches for (source: blog.hubspot.com, updated September 2, 2026).

Quotability decomposes into properties we can build: direct answers near the top of each section, descriptive sequential headings, schema, and every number traceable to a named source. AirOps found sequential headings and schema correlate with 2.8x higher citation rates.

A freshness audit for your content library

The audit we run (and now run on our own blog) is five steps:

List commercial pages: pricing, features, comparisons, buying-intent posts.
Check each page's real update date. Anything older than three months in a fast category is at risk.
Refresh meaningfully: current pricing, current features, new data with sources. A date change alone does nothing.
Make each section liftable: the direct answer in the first sentence, descriptive headings, sourced numbers.
Track citations per engine quarterly. Keyword rankings are the wrong dashboard; citation appearance is the metric.

The bottom line

AI search is a freshness engine. Pages untouched for a quarter lose citations 3x faster, most cited pages were updated within the last year, and citations have stopped following organic rankings anyway. The teams that keep appearing in AI answers run content libraries like software: quarterly releases and a bias toward updating existing pages over shipping new ones.

The full analysis with the complete source list is on our blog (canonical): https://blog.es01.fun/blog/stale-content-loses-ai-citations-2026/

AI Citations Are a Closed Loop: 15 Domains, ~68% of All Answers

Rachel_EShell — Sat, 05 Sep 2026 12:54:52 +0000

If you publish technical content in 2026, one number should reorganize your strategy: fifteen domains capture roughly 68% of every citation that ChatGPT, Claude, Gemini, Perplexity, and Google AI Overviews produce.

The number comes from the AI Citation Source Index 2026, which ranks the fifty most-cited domains in AI answers, synthesized from six independent studies covering 680M+ citations (Everything-PR, updated August 2026).

I write content for software companies for a living, and this index changed where I tell clients to spend effort. Here is the map, the trap, and the play.

The map: AI answers are built from a few super-sources

Strip the index to its load-bearing facts:

Reddit leads the consolidated index at roughly 40% of citations. Across all engines studied, Reddit is the single most-cited domain family.
Wikipedia is #2, with 26-48% of ChatGPT's top-10 citations. In some categories, nearly half of what ChatGPT puts in its top ten comes from one nonprofit encyclopedia.
YouTube is ~19% of Google AI Overviews' top-source share. Google's AI answers lean on video transcripts far more than text-first teams plan for.
LinkedIn and Forbes complete the top five.
Format bias: a separate 2026 analysis found roughly 50% of ChatGPT citations are listicles, and 58% of those are ranked lists (Evertune). AI answers are structurally built from lists.

The trap: aggregate indices hide engine-level cliffs

The consolidated index spans five engines with different retrieval systems. Engine-level behavior diverges hard, and the divergence can be violent.

Cleanest example: ChatGPT versus the rest on Reddit. Reddit's share of ChatGPT Search citations held a steady 3.8% average from July 18 to August 7, 2026. On August 14 it fell below 1%; the August 14-17 average was 0.52%, an 86% relative drop tracked day by day by Promptwatch. Google's AI Overviews declined far more slowly in the same window — which is why the consolidated index still shows Reddit on top.

The mechanism that fits the data: ChatGPT now compiles a shortlist of known brands before it runs its search, reaching for recognized entities instead of scraping forums.

Both readings of Reddit are therefore wrong. It is not dying (it leads the cross-engine index) and it is not king (one engine just abandoned it). Each engine has its own source logic, and those logics change faster than annual indices can track.

The play: what a developer-focused company should do

Break the index down into things you can actually ship this quarter:

1. Fix your entity definition. Wikipedia's citation share is not a content win; it is an entity-definition win. Wikipedia pages are neutral, structured statements of what something is, which is exactly what retrieval models need to answer "what is X." Your replicable version: one sentence describing what your product does, repeated verbatim on your site, GitHub org, docs, LinkedIn, and directories. Machines deduplicate across sources. Give them the same sentence everywhere.

2. Publish the ranked list your category deserves. Half of ChatGPT citations are listicles, and ranked lists dominate those. For developer tools the highest-leverage formats are comparison pages, alternatives pages, and benchmark posts with real, sourced rows. Each entry should stand alone: name, metric, one-line rationale, source. This is not gaming; it is giving a model an answer structure it can quote without inventing.

3. Put FAQ blocks on docs and product pages. Question-answer structure is the most AI-quotable format that exists. If your docs answer "how do I migrate from X" in the first paragraph, both humans and models win.

4. Treat Reddit as a consistency asset, not a traffic hack. Post value, not links. The entity benefit compounds across the engines that still cite Reddit heavily, even when ChatGPT does not.

5. Measure your own citation mix, engine by engine, quarterly. The index is a market snapshot. Your brand's snapshot is different, and it shifts when engines change. Track which sources mention you in ChatGPT versus Perplexity versus Gemini answers for your core queries. A brand that measures its own mix sees cliffs forming; a brand that reads annual indices finds out after the fall.

The short version

AI answers are a closed loop fed by a small set of super-sources, and 68% of citations flow through the top fifteen domains. Stop optimizing for rankings and start optimizing for where answers come from: a consistent, describable entity; ranked and sourced pages in the formats models quote; FAQ-structured answers on every page that sells; and per-engine measurement of your own mix.

The companies that get cited by AI in 2027 will not be the ones that chased the index. They will be the ones that made themselves impossible for an engine to describe wrong.

Sources: Everything-PR, AI Citation Source Index 2026 (680M+ citations, updated Aug 2026); Evertune, AI Search Statistics for Marketers (2026); Promptwatch, Reddit citations are dropping in ChatGPT (Aug 2026).

Originally published on our blog: https://blog.es01.fun/blog/ai-citation-source-index-2026/