Book AI Strategy Call
ITARV Technologies · Rankings vs Recommendations
ITARV · AI Search Resource
Rankings vs recommendations

You can rank first and still not be the one AI recommends

Two kinds of brands show up on page one. Only one gets named when a buyer asks ChatGPT who to hire. Here is what separates them, and the five moves that turn a ranking into a recommendation.

Same rankings, different outcomes

Brands that rankBrands that get recommended
What they measureKeyword positions and organic sessionsShare of buyer prompts that name them, per engine
How they describe themselvesSlightly differently on every profileOne descriptor, word for word, everywhere
How pages openA brand story or a headlineA sentence that answers the question alone
Where their name appearsMostly on their own siteOn the directories, roundups and reviews engines cite
When they know it's workingWhen traffic moves, months laterWhen citation share moves, often quarters earlier

The example above is drawn from a published case on a Northeast US architecture practice (delivered by another agency, analysed by ITARV): the firm held strong rankings and was still largely absent from AI answers until the work below was done. Over two years its citation share on ~180 tracked prompts went from 3% to 37%, as reported by that agency.

The five moves

Run them in order. Each one is useful on its own; together they compound.

  1. Track prompts, not only keywords

    A keyword tells you where you rank. A prompt tells you whether you are named. Ranked brands track 50 keywords; recommended brands track 50 to 200 buyer questions across four engines.

    Do this: Write down the 25 questions your best clients asked before they hired you. Those are your first tracked prompts.
  2. Fix the entity before the content

    Engines hesitate to name a business described three different ways. Different service lists on Google, LinkedIn and directories are the most common, and most fixable, reason a brand is skipped.

    Do this: Write one 30-word descriptor of what you do, for whom, and where. Paste it word for word into every profile you control.
  3. Make your pages extractable

    Ranked pages are written to be read top to bottom. Recommended pages open with a sentence that answers the question alone, and put the numbers in tables.

    Do this: Rewrite the first sentence of your five most important service pages so each one answers "what is it and who is it for" on its own.
  4. Get onto the pages engines already trust

    Answer engines lean heavily on third-party sources: directories, roundups, review platforms, trade press. Brands get named from those pages even when their own site is never cited. This is usually the highest-leverage move.

    Do this: From your prompt audit, list every source the engines cited. Count how many you appear on. That number is your real starting point.
  5. Report citation share from month one

    Rankings and traffic lag. Citation share, the percentage of tracked prompts that name you, moves first. Recommended brands report it monthly so the work stays funded long enough to show up in leads.

    Do this: Re-run the same prompts on the first of every month and log one number: prompts that named you, out of total.

What it looks like when it works

37.9 → 18.3Average Google position for a UK private ultrasound clinic over six months, now cited in AI Overviews for its core scan queries. ITARV engagement.
Key ProvidersWhere an AI answer listed Pratiti Technologies for "digital twin solutions UAE". ITARV engagement.
3% → 37%AI citation share on ~180 prompts for the architecture practice above. Industry reference, not an ITARV result.

Find out which side you are on

We will show you, prompt by prompt, where you rank but are not named, and who is named instead.

Book AI Search Strategy Call
  • 25 of your buyer prompts run across ChatGPT, Perplexity, Gemini and Google AI Overviews
  • Who gets named instead of you, and which sources the engines cite
  • The three fixes most likely to move your citation share first
  • You keep the report whether or not we work together

AI visibility varies by prompt, engine, location, account context and date. The examples and calculations on this page are directional and do not guarantee rankings, recommendations or revenue.