Hi everyone, this is Neo.

Ever had this experience? Your product is solid, your domain authority is fine, you’ve done the link building — and yet, when someone asks an AI (ChatGPT, Gemini, AI Overviews) “which solution is best for small businesses,” the AI recommends your competitor every time. You don’t even appear in the conversation.

Most people’s first instinct: we need more authority. More links. More presence.

But Bill Hunt — a global digital strategist and co-founder of Bisan Digital — published an article on SEJ this week with a counterintuitive conclusion: the problem isn’t your authority. It’s your evidence.

He calls the framework Decision Coverage. Today I’m going to break it down properly — I think it’s one of the most valuable GEO concepts independent site owners will encounter this year.

1. First, understand the shift: “best” went from curated to computed

To get Decision Coverage, you first need the insight from an earlier Bill Hunt piece: the meaning of the word “best” has changed.

Back in the day, “best mortgage rates” or “best laptops” lists were curated — an editor decided who made the list, what got highlighted, what got left out. Get on the list, and you were in the game. Qualification was negotiable.

Now? “Best” is computed. Ask Google “best CD rates” and it calculates the answer live: rates clustered at 4.40%–4.60% APY — visible proof that a machine is executing a mechanical definition of “best” in real time. Ask again the next day and you get a different set. And crucially, no links to any of the individual banks.

What this means: you used to buy your way onto a curated list; now you need to be recognized as qualifying inside an AI’s computation. The game changed completely.

2. The core concept: eligibility gates

The heart of Hunt’s new article is the concept of eligibility gates:

Organizations don’t lose AI recommendations because they lack product information. They lose them because they fail to expose the information AI needs to confidently qualify them for a customer’s decision.

He illustrates it brilliantly:

• Someone shopping for a mattress doesn’t start by asking how many coils it contains. They want to know: does it sleep cool? Does it support side sleepers? Will it relieve shoulder pain? Is it worth the extra cost? Can it be delivered before the weekend?
• Travelers rarely compare hotels by amenity lists. They want to know: is it good for families? Can we walk to the attractions? Is it worth paying more than the place next door?

Those are decision variables. AI has to understand them before it can confidently recommend one option over another. Your product pages can be stuffed with dimensions, specs, materials, warranties, and structured data — if they don’t cover the decision variables, AI can’t qualify you.

3. A brutal real-world case: the “highly configurable” SaaS that AI blacklisted

Hunt shared a B2B SaaS case study that will make your blood run cold.

The company served organizations of every size and had invested in AI visibility monitoring and GEO optimization services. The latest reports showed something strange: when users searched for solutions suited to small and medium-sized businesses, AI almost never recommended them — even though SMBs were a meaningful part of their customer base. And lead volume from that segment had started to decline.

The GEO agency’s first instinct: not enough authority. More community participation, more third-party citations.

Hunt stopped them with a much simpler question: “What have you published that demonstrates your product is well suited for small businesses?”

The answer: almost nothing. The website presented the product as universally applicable but never explained the unique challenges of smaller businesses, the specific benefits for lean teams, implementation considerations, testimonials from that segment, or case studies showing successful outcomes.

Then came the painful part. They asked multiple AI labs to explain their selection reasoning. The pattern was consistent: both the AI models and customer reviews interpreted “highly configurable” as “highly complex.” G2 reviews even said the product was best when you had a dedicated admin to enable its functions. In enterprise buying circles, flexibility is an advantage; in the SMB context, it became the reason to exclude them.

The verdict: this company didn’t have an authority problem. It had an evidence problem. Spending heavily on links and community amplification was treating the wrong disease. The fix was to publish SMB-specific evidence — that the product is easy to configure, light to implement, and works for small teams.

If you sell SaaS, high-ticket products, or B2B overseas, this case study was basically written for you.

4. Google is pushing in the same direction: Merchant Center’s new “conversational” attributes

Hunt points out that Google’s own recent moves confirm the trend — the Conversational Attributes enhancements to Google Merchant Center, which added fields like:

• question_and_answer (common questions and answers)
• related_product (related products)
• variant_option (variant options)
• document_link (supporting documentation)
• popularity_rank (popularity)

Individually these look incremental. Collectively, the signal is unmistakable: Google is moving from asking merchants to describe what a product is, toward asking them to expose the decision knowledge around it — when it should be recommended, how it differs from alternatives, which variants fit which needs, what buyers ask before purchasing, and what evidence supports the claims.

That’s essentially the reasoning chain of a good salesperson in a customer consultation. Google turned it into structured data fields.

5. How to audit your own Decision Coverage

Hunt’s definition: Decision Coverage doesn’t measure how much content you’ve published or how many pages carry structured data. It measures whether you’ve exposed enough evidence for AI to evaluate, compare, qualify, and confidently recommend your product or service.

Every unanswered customer question, every unsupported claim, every missing comparison, every undocumented policy, every unexplained trade-off, every absent customer scenario — each one is a gap in your Decision Coverage.

Here’s the practical checklist, translated for independent site owners:

1. Collect the decision questions your customers actually ask — go talk to sales and support

Sales knows the objections. Support knows the recurring questions. Product managers know the compatibility nuances. The knowledge exists inside your company — it’s just scattered across departments and never organized onto the website.

2. Check whether your product pages answer “why,” not just “what”

Are your pages only describing what the product is (specs, materials, pricing, features)? Do they cover: who it’s for, who it’s not for, how it compares to alternatives, which scenarios it fits, what trade-offs to consider?

3. Provide evidence for every specific customer segment

Like the SaaS case: if you serve small businesses, you need small-business-specific content — their challenges, your unique value to them, their case studies, their testimonials. “Universally applicable” translates to “not especially right for anyone” in an AI’s eyes.

4. Test your content the way AI would reason

Ask ChatGPT, Gemini, or AI Overviews the questions your customers ask, and watch how they answer — who gets cited, who gets excluded, and why. You can even directly ask the AI “why don’t you recommend X” — Hunt did exactly that, and the AI’s reasoning exposed the coverage gaps.

5. Get the structured data in place: FAQ, comparisons, attributes

Google’s new Merchant attributes are the direction of travel. E-commerce sites should add question_and_answer, related_product, and variant_option data as soon as possible — it’s like handing AI a checklist of your decision evidence.

Neo’s take

The more I sit with “Decision Coverage,” the more I think it turns GEO from mysticism into engineering.

Most GEO talk so far — citations, brand mentions, entity associations — is basically public-relations thinking dressed up in new clothes. It’s about influence: whether AI is willing to trust you. Hunt’s framework is different: it redefines the AI recommendation problem as an evidence-sufficiency problem. Authority shapes whether AI is willing to believe you; Decision Coverage shapes whether AI is able to recommend you. Those are two different dimensions.

The part that hit me hardest is the SaaS case — “highly configurable” read as an asset to enterprises and a burden to SMBs, and the AI adopted the latter interpretation. It shows AI’s recommendation logic isn’t “who’s best” — it’s “who qualifies for this specific scenario.” Your product can be objectively better, but if the AI can’t find evidence that you fit this customer’s context, you’re invisible.

It also explains a frustration I hear constantly: “I have more links than my competitor, why does AI never mention me?” — because when AI makes a recommendation decision, it reads your evidence library, not your domain authority.

So here’s my advice to site owners: shift a slice of next year’s GEO budget from buying links and citations to closing evidence gaps — build customer-scenario content, build comparison content, get the knowledge out of sales and support and onto your pages, complete the structured data. It’s not expensive, and it’s the moat that actually matters in the AI era.

Read the three posts I’ve published today as one line of argument: Google’s Generative UI is replacing “tool pages” (post one), clicks are structurally declining (post two), and AI recommendations increasingly run on “decision evidence” (post three). A website’s value is shifting from being clicked to being cited and recommended. Once the direction is clear, the moves stop being scary.