Most B2B teams don't have a production problem anymore. Between freelancers, templates and AI drafting, getting words on a page is cheap. The expensive part is the decision that comes before: what should we create next, for whom, and why?
When that decision is made on instinct, you get the familiar symptoms. Posting regularly, but no engagement. A content calendar that exists mainly to be filled. "Top 5 trends" pieces written to hit a quota. AI drafts that sound like everyone else's. And the quiet question every marketer eventually asks: is there any point in this?
I've spent more than 15 years in B2B marketing, and for the last couple of years I've been building a product around this exact problem as a solo, bootstrapped founder in Paris. This post isn't about the product. It's the framework I use to decide what to publish next, and you can run it with a spreadsheet.
The core idea: decide before you produce
The framework has six steps:
- Write down your marketing context once.
- Audit what you already have.
- Read your own performance signals.
- Listen to what's happening outside.
- Score, decide, and reject topics on purpose.
- Plan to capacity, and approve a brief before anyone (or anything) drafts.
None of it is revolutionary. The point is doing it in that order, every planning cycle.
Step 1: Write down your marketing context once
The most underrated input in content marketing is the stuff everyone "already knows" but nobody has written down:
- Who you sell to. Your ICP (the type of company) and the personas inside it (the people who read, share and sign off).
- What you sell. Products, the problems they solve, and what they deliberately don't do.
- How you sound. Positioning, tone, words you use and words you avoid.
- Who you're compared to. Competitors, and where you genuinely differ.
- Source material. Sales decks, customer interviews, support FAQs and product docs.
This matters twice. First, it filters every topic idea: if you can't say which persona a piece is for, it probably shouldn't exist. Second, it fixes most of the "AI content sounds generic" problem. The issue usually isn't the model, it's the inputs. A prompt with no context gets you the average of the internet; a prompt carrying your ICP, objections, tone and sources gets you closer to your company.
The key word is persistent. Context that lives in someone's head, or in a doc nobody opens, gets re-explained (badly) in every brief. Write it once and reuse it everywhere.
Step 2: Audit what you already have
Before adding anything, look at what's already published. In an audit, I look for three things:
- Freshness: which pieces are outdated (old screenshots, old pricing, old claims)?
- Depth: which topics do you only cover superficially, with one thin post where a buyer would need three?
- Distinctiveness: which pieces say nothing a competitor's page doesn't already say?
Then look for overlap: two articles targeting the same question compete with each other. Many "new" ideas turn out to be an existing article that needs updating, merging, or a better angle.
The output of this step is a short list: update, merge, retire, and gaps.
Step 3: Read your own signals (Search Console and GA4)
Your analytics already show what your audience is trying to find.
In Google Search Console:
- Queries where you get impressions but few clicks. People are searching, you're showing up, but your page isn't the answer they wanted. Often that means a rewrite or a dedicated piece.
- Queries that bring you traffic for topics you never intended to cover. That's demand you didn't plan for.
- Pages slowly losing clicks over time. They're candidates for the "update" list from step 2.
In GA4:
- Which pages people actually read, not just land on.
- Which pages sit on the path to a signup, a demo request or a contact form.
- Which topics bring visitors who come back.
A caveat, because this is where content teams get burned: these signals are useful, but they are not attribution. With long B2B sales cycles, nobody can honestly say which blog post "closed" a deal. Use analytics to decide what to create and improve, not to promise revenue numbers you can't defend. For the revenue conversation, ask new customers how they heard about you. It's crude, but it's honest.
Step 4: Listen to what's happening outside
Internal data only shows the questions people already associate with you. To find the rest, listen in three places:
- Communities (social listening). Reddit and LinkedIn are full of your buyers describing their problems in their own words. Don't go there to promote anything. Go to collect vocabulary. The phrases people use ("busy work", "no engagement", "I don't know what to post") are better headlines than anything you'll brainstorm.
- Your own field teams (voice of the field). Sales and customer success hear objections and questions every week. A simple shared channel or form where they drop "a prospect asked me…" is one of the best idea sources you'll ever have.
- Your market (media watch). Industry news, competitor launches and regulatory changes. These create time-sensitive opportunities that your long-term plan can't predict.
This is also where I keep two editorial speeds: a long-term plan for pillars and SEO groundwork, plus a small reserved slot for reacting to signals (a hot thread, a competitor move, a question sales keeps hearing). Pre-fill every slot and you can't react; plan nothing and you're chasing noise.
Step 5: Score, decide, and reject on purpose
Now you have a long list of candidate topics from steps 2 to 4. Here's where most teams skip the hard part: actually choosing.
For each candidate, I answer five questions:
| Question | Why it matters |
|---|---|
| Which persona is it for? | No persona means no reader |
| Which funnel stage does it serve? | Awareness, consideration or decision |
| Which signal supports it? | Search data, a community thread, a sales question, an audit gap |
| What's our angle that a competitor can't copy? | Distinctiveness |
| What does it cost to produce? | A LinkedIn post and a pillar guide are not the same effort |
Then sort the list. High impact and low effort first. High impact and high effort get planned deliberately. Low impact and high effort get dropped, and you write down why. That last part feels bureaucratic, but it's what stops the same rejected idea from coming back in every planning meeting.
One balance worth naming: automated research is great at surfacing patterns, but people rarely feel conviction about topics a tool picked for them. Treat signals as proposals and keep the final call human. The pattern comes from the data; the conviction comes from you.
I wrote a shorter walkthrough of this prioritization step, including the impact/effort grid, in turning scattered insights into a prioritized editorial calendar.
Step 6: Plan to capacity, then approve the brief before drafting
Capacity first, calendar second
The fastest way to kill a content plan is to make it aspirational. Before placing anything on a calendar, estimate:
- Effort per channel. A newsletter isn't a LinkedIn post, and neither is a long-form guide.
- A weekly maximum your team can actually ship and review, holidays and launches included.
Fill the calendar up to that limit, not beyond. Two strong pieces a month beat an ambitious plan that collapses in week three. When a campaign needs extra capacity, take it from the always-on baseline explicitly, so the trade-off is visible.
The brief is the contract
This is the rule I'd keep if I could only keep one: no drafting, human or AI, without an approved brief.
A good brief fits on one screen:
- Audience: persona and funnel stage
- Goal: what the reader should think or do afterwards
- Angle: the specific point of view, and what makes it ours
- Signals: why this topic, why now (link the data or the thread)
- Constraints: format, channel, length, sources to use, claims to avoid
- Context: tone and positioning, pulled from step 1
Reviewing a brief takes a few minutes. Reviewing a full draft that went in the wrong direction takes an hour, plus a rewrite. AI makes this rule more important, not less: drafting is now so fast that the brief is the only checkpoint left where strategy can still change the output.
Close the loop
After publishing, the cycle restarts: new queries, new GA4 behaviour, new threads and new sales questions feed the next planning period. Over time, "what should we write next?" stops being a weekly crisis.
Where the tooling fits
You can run all of this with a spreadsheet and some discipline. The pain shows up at scale: keeping context current, merging signals from five places, checking overlaps, and respecting capacity across channels and campaigns.
That's what I'm building with CntX. It keeps the marketing context persistent, pulls in signals (audits, Search Console, GA4, social listening and field insights), and turns them into a capacity-aware calendar with briefs you approve before anything is generated. If you'd rather not glue it together yourself, it's a content planning tool for B2B teams built around this framework. If you prefer the spreadsheet, the framework works just as well.
Either way: decide first, then produce. Publishing into the void is a choice, not a law of B2B marketing.
I'm curious how other teams handle this. How do you decide what goes on your calendar next? Let me know in the comments.
Top comments (0)