How to Turn Buyer Language Into an AEO Content Strategy

In 2011 I co-founded DivvyHQ, one of the first content planning and calendaring platforms built for marketing teams. We launched it at the inaugural Content Marketing World in Cleveland, in front of a few hundred marketers who were still deciding whether “content marketing” was a real discipline.

Then I spent the next decade obsessing over how content actually gets planned and produced, watching hundreds of real teams build their campaigns and content calendars inside the software we built.

I saw some things. 😯

Unlike the newsrooms of old, teams rarely sit down and just start “brainstorming” content ideas on a whiteboard. A big chunk of a content plan is already spoken for based on the things that are happening inside the business. Corporate themes are set early in the year. Campaigns and product launches swallow entire quarters of the content calendar.

Then there’s the SEO inputs. Somebody opens a keyword tool, sorts by search volume, and exports the top thirty rows. With that file and a headline generator prompt, an AI tool quickly generates upcoming articles to fill open spots on the calendar.

Both of these scenarios produce a list of content topics. Neither one starts with the buyer.

Most companies have never had a content strategy built around a short list of core buyer problems. The calendar fills up with what’s happening inside the business and what shows up in a keyword tool. A buyer problem you’ve deliberately chosen to own rarely makes the list.

Here’s the uncomfortable part, and I get to say it because I’m the guy who sold you the calendar… Filling the calendar was never the issue. 

We built good infrastructure for executing a decision most teams never actually made: which buyer problems are we the best answer for. 

Nobody ever asked us to solve that decision, and for about fifteen years nobody needed us to.

Covering a wide breadth of topics was a safe play back then. The contest was a human typing a query into a search box and scanning ten blue links. More topics meant more chances to get found. A calendar full of loosely-justified topics worked well enough.

But there’s a second contest running now, and it changed what broad topical coverage means for your marketing outcomes.

The Change No One Saw Coming

Your buyer used to compress their problem down to three or four words and type it into a search box. “Marketing attribution software.” That’s category language, and for fifteen years it was the only input that mattered, which is why most of us obsess over keyword data (to this day).

Now with AI, buyers describe their entire situation out loud, in a paragraph, to a machine. “We’re a 60-person manufacturer, we got acquired last year, and my marketing team still can’t tell me which channels are actually producing revenue.”

Category-based keywords describe a market. That AI-chat language describes a person in a bind, and it’s what’s getting matched now. 

A search engine could bridge the gap between your phrasing and your buyers’ with a synonym index and fifteen years of click data. 

An AI system assembles its answer out of whatever sources speak to the specific situation somebody described, which means the gap doesn’t get bridged for you anymore.

So your buyers’ own language turned into the raw material your content strategy should now be built from. 

☝️Strange sentence to write after fifteen years of keyword exports, but here we are.

Which brings me to this post: a repeatable way to get from the words your buyers actually use to a list of content topics that have a defensible reason to exist.

Narrow Coverage Beats Broad Now

Quick grounding, in case you haven’t read anything else I’ve written on this.

That second contest I mentioned has a different reader: a machine synthesizing everything it has ever seen about you before it decides whether to recommend you at all. 

Your buyer asks it a question in plain language, and it assembles an answer out of whatever it can find and corroborate.

The machines still reward comprehensiveness. Depth, corroboration, and specificity really do win there. The catch is that a marketing team of three isn’t choosing between comprehensive and not comprehensive. It’s choosing between comprehensive on a few buyer problems or thin on all of them.

And thin is worse than it used to be. 

Amongst the 100+ AI visibility audits we’ve run, we rarely find a company in the middle. They’re either largely absent when buyers describe their problems in their own words, or they show up consistently across them. 

Under the old contest, weak coverage got you page two of Google. Annoying, recoverable, fixable next quarter. 

Under the new one, weak coverage on a problem means there’s no consistent story to retrieve when a buyer describes that problem in their own words. You never enter the synthesis at all, which is a much quieter kind of loss than page two.

That’s the whole case for what we call Problem Ownership: pick a small number of buyer problems you have a realistic shot at being the recommended answer for, build genuine depth on those, and deliberately walk away from the rest. 

Three to five problems in most B2B mid-market cases.

The Handoff Nobody Prepares For

Say you buy that argument and go do it.

Leadership gets in a room, argues about it, and comes out the other side with three to five buyer problems the company is going to try to own. These are real problems, articulated in buyer language, and filtered for whether you can realistically win them. 

Everybody nods. Somebody says “this is the clearest we’ve been in two years,” and they mean it.

Then a week later the content lead opens a blank doc and has to turn “we’re struggling to figure out which marketing channels are driving leads” into a publishing plan.

That’s a completely different skill, and nobody handed them a mechanism for it. 

So they do what any reasonable person does now with a blank doc and a deadline. They paste the problem into ChatGPT and ask for twenty blog topics. Ten seconds later they’ve got a list of posts based on one buyer intent, dressed up twenty different ways.

Nothing in the prompt told it which distinct questions to cover, because that’s the part nobody had worked out. 

Six weeks later the problem now has a few published blog posts, a webinar, and a LinkedIn carousel, all of which say roughly the same thing in a different costume, but the AI visibility score hasn’t moved.

The Coverage Illusion

Over my years in the content industry, I’ve watched multiple SEO teams take a single piece of content and rewrite it over and over with a different headline. Stop it.

The productive approach is a set of pieces that covers a deep range of situations, angles, and moments in your target buyer’s journey to solve one problem.

The difference between those two scenarios, volume on one intent vs. deep coverage, is the Coverage Illusion. 

This can easily go unnoticed from the inside. A dozen different titles, three formats, a webinar in the mix. Nothing looks duplicated. 

Underneath, every one of those pieces answers some version of “what is this problem and why should I care,” and not one of them answers “why is your solution a better fit for my situation vs. the other two I’m considering.” 

The Coverage Grid

The core idea is simple enough to state in one line, and it’s the thing that makes everything else work:

Every owned problem contains multiple distinct buyer intents. It is not one theme you write about repeatedly with different headlines.

A buyer who has just started sensing a problem and a buyer who is comparing three vendors to solve said problem are asking for completely different things. 

If your coverage answers one of them well and the other two not at all, your evidence layer has holes in exactly the places where deals get decided.

The Coverage Grid is a method for turning one owned buyer problem into a complete topic list. Buyer stages run down one axis, distinct content angles across the other. Every topic the grid produces then gets sharpened by the real circumstances of your ideal customer: industry, company size, team size, geography, and the situation they’re in. 

So instead of asking “what could we write about this problem,” you fill in a grid and look at what’s still blank.

(This is adapted from the same discipline we use internally to test AI visibility for clients via AI360™. I’m describing the logic here, not the machinery.)

The Raw Material: Buyer Language

Your AEO content strategy runs on the words your buyers actually use, so that’s where you start. Write the problem the way a buyer would say it out loud at the water cooler, not the way it appears in your sales deck.

As mentioned previously, “marketing attribution software” is category language. “I can’t tell which of our marketing activities is actually producing revenue. How can we improve our tracking methods?” is buyer language. Only one of those is what somebody types into ChatGPT.

The fastest way to get this right is to stop guessing at it. Sales call recordings, support tickets, win/loss interviews, and the questions that come up on demos are all full of buyers describing this problem in their own words. 

Your team already has the raw material. Most of it has never been mined for this purpose.

Everything downstream inherits this. Get it wrong here and you’ll build a very rigorous grid for the wrong problem.

The Rows: Three Buyer Stages

Every owned problem can be covered across three buyer stages:

Problem – They sense something is wrong and describe it in symptoms, in their own words, before they know what the actual problem is or that a solution category exists. No category label, no vendor name, just the felt problem.

Discovery – They’ve named the pain and now understand that there are solutions available. They’re actively researching what those approaches are and which might fit their situation the best.

Comparison – They’ve narrowed their focus down to a shortlist of solutions and are evaluating between them. They’re looking for fit, risk, cost, and proof.

Most content libraries are lopsided toward one stage, and the tilt usually maps to whoever is producing the content. Brand and thought leadership teams pile up at the Problem stage. Product marketing lives at Discovery, writing to jobs to be done and solution fit. Demand gen concentrates at the Comparison stage, where the pipeline math is easiest to defend.

None of those teams is doing anything wrong. But it means coverage gets shaped by your org chart instead of your ICP and buyer’s journey, and nobody owns the arc end to end. 

And when the team is three people under pressure to show pipeline, the stage that gets covered is the one where attribution is easiest to prove. 

The Columns: Eight Distinct Angles

This is where depth actually gets built, and it’s the part that trips people up. 

Most people hear “go deeper” and write longer pieces. A 4,000-word article answering the same question your last three answered adds words without adding coverage. 

Real depth is the number of distinct angles a buyer can come at this problem from and find you already there. The angle set I’d work from:

• Cost / ROI (“how much”, “how long before I see a return”)
• Capability fit (“can this actually do what I need”)
• Use-case fit (“does this work in my situation”)
• Objection / risk (“what goes wrong, what am I exposed to”)
• Best-for-scenario (“vertical fit, team size fit”)
• Alternatives and comparisons (“us vs. competitor”)
• Third-party validation (“does this actually work for people like me”)
• Outcome framing (“what changes if I fix this”)

Not every angle applies at every stage. Cost questions rarely show up at the Problem stage. Symptom questions rarely show up at Comparison. The point is to work the grid deliberately instead of writing whichever angle came to mind first.

Every Cell: Specificity Modifiers From Your ICP

Take each topic and apply the unique attributes that make your ideal customer them. Company stage, industry, ownership structure, team size, the situation they’re in.

Take us, Forge & Fathom, for example. One of our core offerings is our Fathom360 Growth Diagnostic, which analyzes a company’s entire GTM engine to look for issues and optimization opportunities.

“Best GTM diagnostic” is generic, crowded, and hard to win. “I’m a CEO of a PE-backed manufacturing company and we’re struggling with growth. How can I determine if we have a marketing issue or a sales issue?” is winnable, and it matches how a buyer with real context actually asks. 

Buyers describe their problem with their real circumstances attached, which is exactly the specificity an AI system is trying to match against something.

This is where a precise ICP stops being a nice-to-have document and starts being a production input.

If we know that PE-backed manufacturing companies are an ideal fit (our ICP), and our offering best serves the “struggling with growth” situation, we need to specify those distinct attributes in everything we write. 

Last Step: The Duplication Check

This is the step that kills the Coverage Illusion, and it’s the one I’d fight hardest to keep.

Read back through the list and ask one question: are multiple topics answering the same buyer question? The test is the question underneath, not the page. Where two topics share it, keep the sharpest and cut the rest, then spend each freed slot on a cell that’s still blank.

Say your potential content topics list came back with both of these:

1. “The Value of a Growth Diagnostic for B2B Companies”
2. “How to Tell Whether a Growth Diagnostic Will Change Anything, or Just Become Another Deck”

Same buyer question underneath: is this actually worth doing? Keep the second. It’s voiced the way a skeptical buyer thinks, and it earns the click the first one only assumes. Cut the first, reclaim the slot.

Staring at a list of twenty headlines can easily create the illusion of solid coverage. But a duplicate intent check is how a list of twenty quietly becomes a list of twelve, saving you bandwidth for other angles.  

Let’s Actually Build One

Directional advice is easy. Let me run one of our own core problems through the grid so you can see the shape of the output.

The problem, in buyer language: “Our leadership team can’t agree on what’s actually holding growth back.”

Problem stage. The buyer isn’t saying “we’re misaligned.” They’re describing a room that can’t get to yes, and they don’t know a growth diagnostic is a category.

• Why your leadership team keeps having the same growth conversation and never closing it (symptom)
• Everyone’s busy, growth is flat, and no two execs agree on why (symptom / situation)
• When every department has a different theory for why growth stalled (use-case fit)
• What it means when your leadership team can’t agree on the top growth priority (stakes / reframe)

Discovery stage. Now they’ve named it. They’re looking for a way to get the team seeing the same problem.

• How leadership teams actually get aligned on what’s limiting growth (capability)
• Why leadership teams get stuck disagreeing about growth (capability)
• What a growth diagnostic examines that an internal debate never will (capability / use-case)
• Is there a repeatable way to get sales, marketing, and the CEO seeing the same growth picture? (approach)

Comparison stage. They know approaches exist. Now they’re weighing them.

• Growth diagnostic vs. leadership offsite: which produces alignment that outlasts the meeting (alternatives)
• What a diagnostic costs against another year of guessing wrong (cost / ROI)
• Will an outside diagnostic expose our leadership team? What actually happens (objection / risk)
• Hire a CMO now or diagnose the constraint first: which order works (alternatives)

Modifiers applied. The ICP attributes that sharpen these: revenue band ($10M–$100M), ownership structure (PE-backed, where board pressure sharpens the disagreement, vs. founder-owned), the makeup of the leadership team (a founder and a first marketing hire, or a newly assembled exec bench), and the situation (two quarters of missed targets, a recent leadership change, a board asking questions nobody can answer cleanly). 

The modifiers take topics you already have and make them winnable, without lengthening the list by a row.

Duplication check, live. Two entries fail it: 

1. “What it means when your leadership team can’t agree on the top growth priority” (Problem)
2. “Why leadership teams get stuck disagreeing about growth” (Discovery) 

These two look different on the surface. Underneath, a buyer arriving at either one wants the same answer: why are we stuck disagreeing, and what does it mean? 

One good page satisfies both. I cut the second and spend the slot on an angle the grid was missing entirely at Discovery: what a diagnostic can settle, and what leadership still has to decide for themselves. 

Nothing else covers the boundary of the approach, and a skeptical CEO asks exactly that before committing.

That’s twelve topics from one problem, and twelve distinct buyer intents with a real person behind each one.

Run that across three to five owned problems and you have somewhere between forty and sixty topics with a defensible reason to exist. That’s a year of weekly posts, and every piece of it compounds toward the same small set of problems instead of scattering.

Validate the List. Don’t Chase Queries.

One optional gut-check before you publish. The AEO/GEO world has a real conversation going about query fan-out: ask an AI system a question and it quietly fans it into a batch of related sub-queries, then synthesizes across those instead of narrowly answering what you typed. 

Tools now exist to expose those sub-queries, and it’s genuine visibility into behavior that used to be a black box.

But be wary of the trap. Fan-out queries are probabilistic. Run the same prompt twice and you’ll get a meaningfully different set of sub-queries both times. Build a page for every synthetic query a tool surfaces and you’re building pages for noise.

So use it the way the grid already works. Run a few of your real buyer-phrased prompts through a free fan-out tool (DEJAN’s Query Fan-Out tool is solid and FREE), read the output in aggregate, and look for a theme your grid missed or a comparison that keeps surfacing. Fold those into the grid. Then close the tool. 

Do You Fully Cover One Problem Before Starting the Next?

No. Driving one problem to full depth before you touch the next is how you end up with a library that’s deep in one place and absent everywhere else.

I’d resist turning this into a posts-per-month number anyway. The honest answer depends on how contested the problem is and what your evidence layer already looks like.

The principle I hold to is stage coverage before angle saturation. Get real coverage at all three stages on problem one before you go deep on any single stage. 

A library with four excellent Problem-stage pieces and nothing at Comparison leaves the buyer at the exact moment they’re deciding. Breadth across the arc, then depth within it.

So once all three stages have something substantive, move to problem two and come back. Full depth accrues over multiple passes, through corroboration and repetition, which makes the work closer to tending than to shipping.

Knowing When a Problem Has Enough Coverage

To be completely honest here, the industry is still experimenting and building measurement instruments to answer questions like “How much coverage is needed to start showing up?” and “How often should we publish on topic XYZ to maintain our visibility?”

That said, there are signals worth watching, and they’re better than a content count.

Citation patterns. When you describe the problem in buyer language to a few different AI systems, do you surface? More telling: do you surface across multiple phrasings of it, or just the one you optimized for? Consistency across phrasings is the signal that the evidence layer has actual density rather than one lucky page.

Inbound language matching your framing. This is the one I trust most. When prospects start describing their own situation using the words you chose, your framing has propagated. Nobody’s quoting your blog post back at you on a discovery call. They picked up a way of naming the problem somewhere, and that somewhere was you.

Third-party corroboration appearing without you prompting it. Someone else’s post, a podcast mention, a review that uses the problem language near your name. That’s the evidence audit starting to close.

A version of this showed up for us recently. Tracy Scott, a fractional GTM leader in our network, posted her own take, unprompted, and pointed her audience at us by name. Her line: “AI discovery doesn’t create clarity, it exposes whether you ever had it.”

She basically stated our argument, in her words, to her network. These unsolicited mentions are a great indicator that your coverage is gaining traction.

One more consideration, and it’s the one that keeps “enough” from ever being final…

Recency. Practitioners are increasingly seeing what looks like a recency bias in how these engines decide what to surface. A page that corroborated your problem six months ago can quietly stop pulling its weight. The page didn’t get worse. The engine appears to have shifted toward fresher sources. The effect isn’t pinned down yet, and it seems to vary from one engine to the next.

The practical read: coverage isn’t permanent. A problem you owned last year can slide if the evidence layer goes stale and nothing refreshes it. So “enough” carries a maintenance rhythm, not just a one-time build. 

Refresh, update, and add to your strongest problems on a cadence, the way you’d service anything you need to keep running.

Build One Grid This Week

Compressed, for a single owned problem:

1. Write the problem in your buyer’s words, not your category’s.
2. Set your rows: Problem, Discovery, Comparison.
3. Set your columns: cost, capability fit, use-case fit, objection/risk, best-for-scenario, alternatives, third-party validation, outcome framing.
4. Fill the cells, applying your ICP’s real attributes as modifiers. Stage, industry, ownership, team size, situation.
5. Run a handful of buyer-phrased prompts through a free fan-out tool. Read for themes, not for queries.
6. Run the duplication check. Cut every topic a neighbor already satisfies, and spend the slot on a blank cell.

That’s an afternoon, and it’s the difference between a year of evidence and a year of the Coverage Illusion.

Get The Problems Right First

The Coverage Grid amplifies whatever you feed it. Point it at a problem you have no real shot at owning and it will hand you a beautifully structured year of content in defense of the wrong bet.

So the grid is step two. Step one is making sure your problems list clears the bar first: aligned with your strategy, anchored to revenue, and winnable, which means an honest look at how steep the climb is going to be to get consistent mentions and citations in AI.

Remember the room where leadership agreed on the problems short list and called it the clearest they’d been in two years? That clarity was real. It just wasn’t finished. Knowing which problems you own is half the work. Building the coverage that lets you own them is the other half, and now you’ve got the grid for it.  

P.S. If you’d like a clear read on your current AI visibility, or need some strategic guidance on problem identification or AEO execution, give us a shout.