Decision guide · 10 min read

Your brand is absent. What should you fix first?

“AI doesn’t mention us” is five different problems wearing the same coat. This is how to tell which one you have before you spend a quarter fixing the wrong one.

· By the Askwords team

The most expensive response to being absent from AI answers is to start writing. Content is the fix for exactly one of the five common causes, and if yours is one of the other four, you will publish for three months and move nothing.

Absence is a symptom. Below is how to read it, starting with the check nobody wants to do first, because it is dull and because it is occasionally the whole answer.

01 · Diagnose before you build

Rule out access, then read the sources

Before any strategy, confirm the answer engines can actually read your site. A well-meant robots.txt edit, a blanket bot block at the CDN, or a framework that renders everything client-side can each make an otherwise excellent site invisible to the crawlers behind these answers. It happens more than you would expect, and it costs ten minutes to rule out.

Once access is clear, open the sources the AI answer cited and read them as if you were the buyer. Read them properly, do not skim. The answer to “what did they have that we did not?” is almost always sitting in plain text on those pages.

Five minutes reading the pages that beat you is worth more than five hours of keyword research. The evidence that won is right there, already public.
02 · The decision tree

Four questions, in this order

Work down the list and stop at the first “yes”. That is your gap. The order is deliberate: earlier items are cheaper to fix and block the later ones from working.

GapHow it shows upEffortCharacter
Content gapThe cited pages answer a question you have simply never answered in public.1 to 2 weeks per questionFastest to move
Entity gapThe model describes you vaguely, wrongly, or as the wrong kind of company.Days of editing, weeks to take effectCheap, often ignored
Third-party gapEvery cited source is a roundup, directory or review site you are not on.1 to 3 monthsSlowest, highest ceiling
Proof gapYou are named, but hedged: “smaller”, “newer”, “less established”.OngoingCompounds quietly
03 · Content gap

The question exists. Your answer does not.

This is the most common and the most fixable. Your site sells your product; the cited pages answer the buyer’s question. Those are different documents, and marketing sites are full of the first and short of the second.

The fix is a page that answers one specific question completely, in the first two hundred words, without a preamble about your journey. Then the detail, the caveats, the numbers, the cases where your answer does not apply. That last part matters more than it looks: hedged, honest writing is disproportionately quotable, because it is what an answer engine needs to distinguish situations.

Write this

“For teams under ten people, X costs about £Y a month and takes an afternoon to set up. Above thirty people, the approval workflow starts to matter and Z is usually the better answer.”

Not this

“In today’s fast-moving landscape, choosing the right solution is more important than ever. Our platform empowers teams to unlock their full potential.”

One good page per buying question beats twelve thin ones. If you can only write four things this quarter, write them for the four questions where you are absent and the buyer intent is highest.

04 · Entity gap

The model is unsure who you are

Ask a model directly what your company does. If the answer is vague, dated, geographically wrong, or quietly about a similarly-named company, you have an entity problem, and no amount of new content fixes it, because the model will not confidently recommend something it cannot confidently describe.

Entity work is unfashionable and cheap. It is mostly making the same facts appear the same way everywhere: one company name spelled one way, one canonical description, consistent category language, consistent address and founding details, structured data on the about page, and the same story on the third-party profiles you already have.

The name test

Search your brand name with your category word. If page one is dominated by something else, a band, a town, a bigger company with a similar name, the model is fighting the same ambiguity. That usually means leaning harder on your category words in your own copy, so that “Acme” and “invoicing software for tradespeople” are welded together everywhere you appear.
05 · Third-party gap

The pages that recommend you are not yours

Look at the citations again. If they are overwhelmingly roundups, directories, marketplaces and review sites, then the recommendation is being made somewhere you do not control, and your website is not in that conversation at all.

This is the slowest gap to close and the one with the highest ceiling, because it is the one competitors cannot copy from you overnight. Build a short target list from the domains that keep appearing in your answers, then grind through it: get listed, get reviewed, pitch the roundup author with something genuinely new, publish the data nobody else has.

Category directories and marketplaces

Often the single most-cited source type in software and services answers. A complete, current profile is table stakes.

Review platforms

Volume and recency both matter. Twelve reviews from last year reads as quieter than forty from this one.

Independent roundups

The “best X for Y” articles. Reach out with a specific, checkable claim rather than a request to be added.

Trade and local press

Undervalued. A single credible outlet in your niche can outweigh a dozen generic mentions.

06 · Proof gap

Being named cautiously is its own problem

Sometimes you are in the answer and still losing. The tell is the adjectives: competitors get “established”, “widely used”, “trusted by”; you get “newer”, “smaller”, “also worth considering”. Those adjectives track your evidence base, not your product.

Proof is specific and checkable: customer numbers, named case studies with real outcomes, review counts, certifications, years in business, named clients who agreed to be named. Vague confidence, “industry-leading”, does nothing, because it is unverifiable and every competitor claims it too.

Replace one adjective with one number. “Trusted by hundreds of teams” becomes “used by 1,400 UK plumbing firms”. The second is quotable; the first is noise.
07 · Doing them in order

A realistic first quarter

You will likely have more than one gap. Do them in this order anyway: each step makes the next one work better, and doing the slow one first wastes the fast one.

  1. 1

    Week 1: Access and entity

    Confirm answer bots can crawl you. Fix the name, category and description everywhere they appear. Cheap, quick, and it unblocks everything else.

  2. 2

    Weeks 2 to 6: Content

    Write the two or three pages that answer the buying questions where you are absent. Answer-first, specific, honest about limits.

  3. 3

    Weeks 3 to 12: Third-party

    Start in parallel, because it is slow. Directories and reviews first, roundups and press after. They take longest and need something to point at.

  4. 4

    Week 12: Re-measure

    Same questions, same answer engines, same wording. Compare to the baseline you took at the start. If you did not take one, take it today and start the clock.

The takeaway

Absence is a diagnosis, not a sentence.

Read the sources that beat you, name the gap, and do the cheapest one that blocks the others. The businesses that get recommended in a year are the ones that spent this quarter fixing one specific thing instead of publishing in every direction at once.
Check three buyer questions free

Sources and further reading

Askwords measures a selected set of buyer questions against fresh answers from each engine. AI answers vary by model, time, context and personalization, so a check is evidence of the answers observed, not a universal judgement about a brand.