- +44(0)28 9099 8390
Most AEO programs start with prompt lists and visibility dashboards. That’s the obvious place to start; however, before jumping into prompt tracking or other AEO tactics, you need to answer three strategic questions. Each one helps inform your AEO program to help drive business impact vs a vanity visibility percentage.
Published: 27 Aug 2026
Author: Adam Grant
Map your candidate territories against two axes: commercial value and viability.
Decide the descriptors and positioning you want to be known for. Being visible in an AI answer is half the battle; having it represent your brand the way you want it to be is the other. Decide the descriptors and positioning you want to be known for.
Every business has unique advantages that are hard to replicate: proprietary data, customer proof, analyst relationships, community, personal brand. Pick the ones that you can leverage to drive authority.
If you haven’t yet made the internal case for why AEO deserves investment, start with [Understanding the Business Impact of AI Search].
SEO trained a generation of marketers to believe that if you could find the keywords and build the pages, traffic would follow. A single well-optimized article could rank on the strength of the page itself, regardless of whether it aligned with your brand’s positioning or commercial priorities. HubSpot built an empire on this: thousands of articles answering questions that had nothing to do with their core product, all funnelling awareness and email captures back to the ICP.
AEO doesn’t work that way. When a buyer asks an LLM for a solution, they don’t get ten blue links. They get a small set of recommendations, described in the model’s own words, drawn from a category the model has already decided you belong to (or don’t). Showing up in that answer requires the entire commercial apparatus behind your brand, including product depth, customer proof, analyst coverage, press, and third-party validation, to have already established you as a credible answer. AEO harvests that alignment. It doesn’t create it.
That’s why the onboarding work matters. If you can’t answer the three questions clearly, you will spend resources trying to be shortlisted in a category/job to be done that isn’t associated with your business (And therefore extremely costly to break into) or worse, you build AI visibility that doesn’t drive business outcomes.
The prompt universe is effectively infinite. Every product, ICP, use case, integration, geography, and buyer role generates its own set of relevant queries. For any business bigger than a single-product startup, you cannot meaningfully compete across all of them, and trying to do so dilutes your entity signal, the pattern of associations the LLM uses to place you in a category. Optimising for everything is the same as optimising for nothing.
So, the first question is really about prioritisation, and the best way to do that is to inherit the overall product and GTM strategy.
If you decide you want to show up for “AI contract review” but product isn’t investing there, and marketing isn’t running campaigns around it, your AEO program will be paddling in the wrong direction. Best case, you get visibility that doesn’t convert. Worst case, you generate demand the rest of the org can’t service, and you erode trust with buyers and internally.
Your first job in onboarding is to understand direction and strategy that already exists: the product roadmap, the ICP definitions sales operate against, the segments marketing is funding, the geographies with headcount. If that is clear, your priorities fall out of them. If they’re not clear or they contradict each other, that’s the finding, and it needs to go back to leadership before you touch a prompt list.
For each candidate territory (a product line, an ICP, a use case, a geography, an integration) ask two things:
Not just “is it a big market,” but: does it capture buyers who fit your ICP, who have real purchase intent, whose deals you can close and retain? Value is the intersection of commercial priority, buyer quality, meaningful ACV, and fit with what product and GTM are already investing in. A big market where you don’t close buyers is worth less than a small one where you do.
Does the LLM already see you as a plausible player in this territory today? You can get a rough answer in fifteen minutes by running a few category-level prompts across ChatGPT, Perplexity, and Gemini and watching whether your brand appears unprompted. If it does, you have a foundation to build on. If it doesn’t, and no third-party sources meaningfully associate you with this category, winnability is low, and no prompt-level tactic will change that in a quarter.
Winnability is largely determined by the LLM’s existing entity associations, which are shaped by years of how the web has talked about you. LLMs recommend brands based on learned co-occurrence patterns between entities and categories.
Plotting your territories against these two dimensions:
Low winnability | High winnability | |
High value | Reposition, reframe, or wait. Multi-year investment if you commit. Don’t pretend it’s a quarterly play. | Invest heavily. This is your Tier 1. Commercial priorities and the LLM’s view of you already align. |
Low value | Ignore. | Often your highest ROI. Small universe, but you can own it, and the buyers who arrive convert. |
The counterintuitive quadrant is the bottom-right, and it’s where a lot of pipeline comes from. Small, specific territories where you’re already the obvious answer often convert better than glamorous ones where you’re fighting to be considered at all. (Even if your boss or board wants to show up for the generic Best X in category)
For the top-left, high-value, low-winnability territories, you have three honest options:
Product marketing, PR, analyst relations, customer advocacy, and content all pulling in the same direction for 6 to 18 months. Expensive but sometimes the right call.
This is especially common in tech with integrations and marketplaces (e.g. policy management as a standalone vs. policy management on SharePoint). Use your existing category strength rather than fighting it.
If the LLM’s view of you can’t be shifted in a timeframe that matches the commercial opportunity, it doesn’t belong on the priority list, regardless of how attractive the market looks.
The output of Question 1 isn’t prompts. It’s a ranked, bucketed list of territories that the rest of the org has signed off on. Something like:
Flagship products, core ICP, highest-ACV use cases. Missing here is a revenue problem.
Adjacent use cases, secondary ICPs, comparison and alternatives queries. Missing here is a pipeline leak.
Long tail, exploratory buyers, tangential integrations. Track but don’t invest.
For a single-product company solving one job for one ICP, this exercise is trivial. The whole business already agrees on the territory, and you can move to prompts quickly. For a multi-product enterprise, it’s the opposite: every BU thinks their line deserves priority, and without arbitration you end up with a prompt list that reflects internal politics rather than commercial reality.
Once you know where you want to show up, the next question is what the LLM should say about you when you do.
This is the real difference from SEO. SEO was built on rankings. You competed for position on a results page, and the click was the conversion event. AI search doesn’t rank; it describes. When a buyer asks “what’s the best tool for X,” the model doesn’t just list you. It characterises you: what you do, who you’re for, what you’re known for, how you compare to the others in the answer. Those descriptors are what turns visibility into pipeline (or not)
And you don’t dictate them. You influence them. The model’s description of you is an averaged, compressed view of how the web talks about you: your own site, third-party coverage, review platforms, forums, analyst reports, customer stories, community threads. If those sources describe you consistently, the model’s description will be sharp and useful. If they describe you inconsistently, the model will hedge, generalise, or default to the language your competitors have positioned you as.
For each priority territory: the category you belong in, the ICP you serve, the specific job you do best, the differentiators that matter, and the associations you want to avoid. This has to come from product marketing, not AEO. Your messaging house, positioning docs, and category narrative are the inputs. If those don’t exist or don’t align across the org, that’s another finding to send back before proceeding.
In the sources the LLM learns from. Your homepage, your G2 profile, your analyst write-ups, the press coverage you’ve earned, the way customers describe you in case studies and community threads. Fragmentation is a problem, and it’s more common than teams realise, because different functions own different surfaces and rarely coordinate on language.
The tactical work of shifting the language across the web is a long game we’ll cover in a future post. But the strategic work, deciding what you want the language to be, and getting internal alignment on it before you start seeding it externally, belongs in onboarding.
The first two questions are about clarity: where you want to show up, and how you want to be described. The third is about leverage: what you already have that competitors don’t, and how you turn it into AEO signal.
One of the most citable content forms for LLMs, because it’s genuinely novel and other sources reference it. If you have data no one else has, publishing it well is one of the highest-leverage moves you can make.
Named customers, quantified outcomes, and detailed case studies create the third-party validation LLMs weight heavily. Volume matters, but so does specificity.
Coverage in the sources LLMs treat as authoritative (category press and analyst firms) disproportionately shifts entity recognition.
Employees, customers, and users talking about you in forums, on LinkedIn, in podcasts. This is the co-occurrence layer that trains the model over time.
Being mentioned alongside adjacent tools in integration docs, partner marketplaces, and comparison content places you in the competitive set you want to be part of.
Smaller teams can publish, react, and get cited faster than enterprise competitors moving through legal review.
Specialised industry knowledge, in-house experts, and a research team can carry disproportionate weight in narrow verticals.
The strategic question isn’t “how do we do more content.” It’s “which of our existing advantages, if we invested in surfacing them, would move the needle on the specific territories we prioritised in Question 1?” A brand with strong analyst relationships should probably lean into that before starting a podcast. A brand with a passionate user community should activate it before commissioning original research. Play your actual hand.
The three questions work in sequence. Question 1 tells you where to compete. Question 2 tells you how to be understood when you get there. Question 3 tells you what to use to get there faster than your competitors.
It’s worth being clear that this isn’t AEO execution. It’s the strategic planning that enables your future AEO program to be directly connected to the most business impact.
Now that your foundation is set, the next step is turning your prioritised territories into the specific prompts you’ll track. → [How to Build Your AEO Prompt Set]
Be The Brand AI Recommends
See where your brand appears in AI search, where your competitors are winning and what it takes to become the answer AI recommends
AEO is the practice of influencing how AI-powered answer engines (ChatGPT, Perplexity, Gemini, Google AI Overviews, and others) describe and recommend your brand when buyers ask them for solutions. Unlike SEO, which competes for position on a results page, AEO competes for inclusion and framing inside a generated answer.
Keyword tracking measures rank for a specific query. Prompt tracking measures probabilistic visibility across a set of representative questions. LLM answers vary run to run, and real users phrase questions in unlimited ways with unique context. Prompt tracking gives you directional signal on where you appear and who’s winning.
For a single-product company with a clear ICP, you can work through the three questions in a week or two. For a multi-product enterprise with multiple BUs, geographies, and buyer segments, expect two to three weeks, most of which is internal alignment and arbitration rather than analysis.
AEO usually sits within marketing (often SEO, content, or product marketing), but it depends on organisational inputs from product, sales, PR, and analyst relations. The owner’s job is less to invent strategy and more to translate existing product and GTM strategy into the AEO context and coordinate across the functions that shape how the web describes your brand.
Depends on the gap between where the LLM currently places you and where you want to be. If you’re already in the competitive set for a territory, you can shift visibility within months. If you’re starting from zero (no third-party sources associate you with the category), expect 6 to 18 months of consistent work across product marketing, PR, analyst relations, and content before you see meaningful movement.
That’s a finding, not a blocker for AEO alone. It means the business hasn’t decided where it’s competing, and no downstream function (AEO, demand gen, content, sales enablement) can operate effectively without that clarity. Surface it to leadership before building a prompt list. An AEO program built on unclear priorities will reflect that ambiguity in its results.
Run five to ten category-level prompts across ChatGPT, Perplexity, and Gemini (e.g. “best tools for X,” “top vendors for Y,” “alternatives to Z”). If your brand appears unprompted in the answer set, you’re in the category. If it doesn’t, check whether any third-party sources (analyst reports, review sites, press, community threads) associate your brand with the category. If those are also silent, you’re starting from zero.
Growth Manager, Visibility Wins
Adam Grant leads the Answer Engine Optimisation (AEO) team at Visibility Wins. Adam specialises in helping B2B IT companies become discoverable across Google and LLMs. His work focuses on building the content and authority signals that get brands recommended as trusted answers.