Week 5 · the third dimension

Competitor framing & share of voice

AI rarely describes you alone. It names a set — and where you sit in that set, and how often you're in it at all, is its own dimension.

Makefield · Berlin

The third question

You're rarely described alone

Ask an assistant "what's the best tool for X?" and it doesn't describe one company — it names a set, and frames each against the others: this one's for enterprises, that one's cheaper, this one's the technical pick. So even when a model recognises you and describes you accurately, two more things decide whether that helps: who you're named alongside, and how often you're named at all. That's competitor framing and share of voice — the third dimension.

It's relative, not absolute. Being described well matters little if you're framed as the weak option next to three stronger ones, or if you're simply left out of the set while competitors fill it.

The framing

How you're positioned against the set

Competitor framing is the role the model assigns you within the group it names. The questions to ask of any answer: Are you placed in the right peer set, against genuine competitors — or sorted in with the wrong category? Is the contrast it draws fair and current, or is it repeating an old weakness you've since fixed? And when it implies a "best for…", does the for match who you're actually best for?

These aren't vanity concerns. The framing is the comparison a buyer would otherwise have made themselves — made for them, in a sentence, by the model.

Share of voice — measured per engine

Being in the answer at all, and how often

Share of voice is the blunter measure: across the questions a buyer might ask, how often are you the one named? And here the field's defining fact shows up — there is no single answer to "does AI recommend us."

I tested this directly. Across seven buyer-style questions put to six assistants, no question produced agreement on who to recommend: roughly three-quarters of the companies named appeared in only one engine's answer, and when I asked for the single best specialist, the six engines gave six different names. The sources they draw on barely overlap either. The consequence is concrete: you can hold a strong share of voice in one assistant and be invisible in another, and you'd never know unless you checked each one. Share of voice has to be measured per engine, across a real spread of buyer questions — a single check, on a single assistant, tells you almost nothing.

Lever, not outcome

Why this is something you can work on

Competitor framing and share of voice are levers — things you can move with work, because they come from what the web says about your category and who gets cited in it. The thing they point toward — being recommended — is the outcome, and it's the next lesson. The distinction matters: you can't chase a recommendation directly, but you can raise how often and how well you're framed, and the recommendation follows from that. Work the lever; the outcome is downstream.

Next — Recommendation: the outcome every other dimension feeds, and why you reach it sideways. New here? Start with the overview.

Common questions

Questions this lesson answers

Why does AI recommend my competitor instead of me?

Because an assistant rarely describes you alone — it names a set, and frames each company against the others. Two things then decide where you land: who you're named alongside, and how often you're named at all. A competitor gets recommended over you when the model puts it in the stronger role inside that group, or when it fills the set and you're simply left out of it. Being described well counts for little if you're framed as the weak option next to three stronger ones. It's relative, not absolute — the question is your position in the set the model names, not just the words it uses about you on their own.

What is share of voice in AI answers, and why can't a single check tell me mine?

Share of voice is the blunt measure of how often you're the one named across the range of questions a buyer might ask. It has to be measured per engine, because there is no single answer to "does AI recommend us." When I put seven buyer-style questions to six assistants, no question produced agreement on who to recommend — roughly three-quarters of the companies named appeared in only one engine's answer, and asked for the single best specialist, the six engines gave six different names. So one check, on one assistant, tells you almost nothing about your real share of voice.

Can I be visible in one AI assistant and invisible in another?

Yes, and it's common. Because the engines barely overlap in the sources they draw on, you can hold a strong share of voice in one assistant and be absent from another's answers entirely — and you'd never know unless you checked each one separately. That's why share of voice is measured per engine, across a real spread of buyer questions, rather than from a single query on a single tool.

Is share of voice something I can actually improve, or do I just wait for a recommendation?

It's a lever you can work on — the recommendation itself isn't. Competitor framing and share of voice come from what the web says about your category and who gets cited in it, so you can raise how often and how well you're framed. Being recommended is the outcome that follows, and you reach it sideways: you can't chase a recommendation directly, but working the lever moves the thing it depends on. No tactic guarantees your placement in any given answer.