Building Mentions for SaaS Brands

When ChatGPT names a software product, 41.72% of the time none of that product's own pages is among the sources (Google AI Mode: 40.17%), and for products named sixth or later, 73.25%. Who keeps the pages that carry the name, and what the weekly corrections cost.

On this page
  1. Who Keeps the Pages Behind Your Name
  2. What the Weekly List Costs You
  3. What Qvery’s Agents Do Every Week
  4. What Stays With Your Team
  5. What We Take on in the First Two Weeks
  6. Frequently asked questions

For a software company of a few hundred people, the answers buyers read are now part of how they find you: in a Gartner survey of B2B buyers, 45% had used generative AI during a recent purchase, mostly to gather information on vendors and products.

When an answer names your product, the further down the list you sit, the more often the pages behind your name are other people's: a roundup, a rival's comparison post, an app marketplace, a review site. Nobody on your team owns finding them.

41.72%of ChatGPT's namings of a software product had none of that product's own pages among the sources
40.17%of Google's chat-style answers' namings did
73.25%for products ChatGPT named sixth or later
Qvery research: the GEO Agent's daily records of software-buying answers, August 2026

At Qvery, we help software companies free up marketers for the work only people can do, with custom AI agents that sort the pages that miss you and draft corrections.

Who Keeps the Pages Behind Your Name

A mention here is your product's name on a page an answer reads; a citation is the link the answer shows under it. When we opened the cited pages for a sample of the namings with none of the product's own pages, the name was on at least one of them in 75.74% of cases.

So the name is usually on a page someone else keeps, and that keeper, the person or site that owns the page, sets the terms: editors, other software companies, marketplaces, review sites and communities.

Google's guide to its AI features warns that "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem," so the work is getting onto pages other people keep, by means their keepers accept.

Vlad ShvetsPartner @ Qvery

For a product named sixth or later, 73.25% of ChatGPT's namings had none of its own pages among the sources, so lower down the list your name rests on pages other people keep. Getting onto them is mostly hours of corrections and listings, not placements you buy. Treat any paid roundup as an ad with its own budget owner, and decide who owns the weekly list of pages before you decide on any placement.

What the Weekly List Costs You

The agents' drafts go first to the cheapest rows: your marketplace listing (your company's page inside another product's app directory, the one page someone else hosts that you write yourself), then a factual correction to a page that gets your price or plans wrong, which costs hours, not a pitch.

The calculator prices the correction work for a software company of about 250 people: one marketing specialist finding the pages that get you wrong and writing to the sites behind them, about 14% of the week. At about $127,800 a year, the Bureau of Labor Statistics’ mean pay for marketing specialists with benefits, that is about $17,892 a year.

With the agents taking about 60% of it, at about $200 a month to run, it costs about $9,557, saving about $8,335 a year and 175 hours; each $10,000 of the build earns itself back in about 14.4 months. The share and the running cost are our assumptions, to replace with your own.

Fact Correction ROI Calculator

What Tracing Wrong Facts About Your Company And Chasing The Corrections Costs You, And What An Agent Saves

The marketing specialist who keeps your company's facts straight.
Pay plus benefits per person. Market research analysts and marketing specialists earned a mean $89,490 in May 2025, the Bureau of Labor Statistics reports; with benefits at 30% of private employer costs, about $127,800.
How much of their week goes to finding the pages that repeat a wrong fact, fixing your own and writing to the other sites: 14% is about five and a half hours a week.
Tracing every wrong line to the pages that repeat it and drafting each correction. The owner signs each one before anyone edits a page or writes to another site.

Saved Per Year

$8,335

175 hours handed back, after $2,400 a year to run it

Hours Handed Back
175 hours a year
Agent Running Cost (Estimate)
$200 a month
Payback Per $10,000 Of Build
14.4 months

On these numbers, the agent saves $8,335 a year after its own running cost, and every $10,000 of the build is earned back in 14.4 months.

Example numbers. Type over them with yours.

What Qvery’s Agents Do Every Week

The agents read the week's answers from the GEO Agent's records, sort the cited pages that name a rival but not you or get your facts wrong, and draft each correction against your facts sheet, in one weekly list for the person who signs it off.

Piotr BabelPartner @ Qvery

The agents check each cited page against your facts sheet, line by line: price, plans, integrations, features. When the sheet is out of date, every correction drafted from it is wrong too, so the person who signs off the weekly list also owns the sheet and dates each line. A correction goes out only on a dated line, because a keeper who gets one wrong fix from you reads the next one with less patience.

Their access starts and ends with the GEO Agent's records and your facts sheet, and every row on the weekly list is logged with the person who signed it off.

What Stays With Your Team

Only a named person contacts a keeper, from their own address. The person responsible for new business signs off the list each week; the rows go to partnerships for marketplaces, product marketing for comparisons and customer success for reviews.

What We Take on in the First Two Weeks

Your buyer questions, your facts sheet and the names of who sends what are where the agents begin. Two weeks after they can read the GEO Agent's records, the first weekly list reaches the person who signs it off, with your marketplace listing and the factual corrections at the top.

Frequently asked questions

Should we stop building backlinks?

We measured names on cited pages, not backlinks (links from other sites to yours): the name appeared on a cited page in 75.74% of the namings we checked. Keep link work that puts you on pages buyers read, and judge the rest by whether the page names you.

Is it worth paying to be included in a best-of list?

Treat paid roundup placement as an ad buy, with a budget owner and a page some buyers will read as an ad. Consider it when the page repeatedly appears under your buyer questions and names several rivals, after handling factual corrections and your own marketplace listing.

Can we ask a competitor to fix what their comparison page says about us?

Send a correction when a price, plan, integration, or feature is wrong, with evidence from your pricing page or release note. The rival may decline; if the page stays wrong, publish your own dated comparison and say where each rival wins.

Can we give customers something for a review?

Only if the reward does not depend on the review being positive and is disclosed under the review site's rules. Ask every customer, happy or not, and never your own staff.

How long until a new mention shows up in AI answers?

We have not measured the delay between a new mention and a changed AI answer. Count page changes and answer changes separately, keeping the same buyer questions throughout the test.

Who should own mention building at a company our size?

The person responsible for new business owns the list and signs it off each week. The rows belong to different people: partnerships for marketplaces, product marketing for comparisons, customer success for reviews, the support lead for communities, and marketing for editors.

Is it the same on ChatGPT and Google's chat-style answers?

No. In our August records the pages behind a name differed by engine: Google's chat-style answers leaned on YouTube videos far more than ChatGPT's. Track both engines separately rather than assuming a page will matter equally to each.

What is the difference between a mention and a citation?

A mention here is your product's name on a page an AI answer reads; a citation is the link the answer displays. An answer can recommend your software while citing only pages written by other companies.

Which page should we work on first?

Start with your integration's marketplace listing if it meets the platform's rules. You write that page yourself, so describe the integration specifically, match the prices on your own website, and add it to the monthly check.

What should we send an editor who compares software?

Provide an account the editor can test without a sales call, current plan prices, a capability the comparison misses, and a product limitation. The purpose is to help the editor evaluate the tool, not to ask for a higher ranking.

Do seeded or fake mentions work?

Google's guide to its AI features warns that seeking inauthentic mentions across the web isn't as helpful as it might seem, and review sites bar rewarded praise. Work only by means each page's keeper accepts: a correction with evidence, a review asked of every customer, a listing you write yourself.

How do we decide whether mention building is worth continuing?

Run a quarter-long test with one accountable owner and a fixed weekly time budget. Compare answers to questions whose cited pages you changed with answers to questions whose pages you left alone; if the first group does not improve, stop the outreach and keep the listings and corrections.

Why does ChatGPT name us but cite someone else's page?

Because the page it read was someone else's: in 41.72% of ChatGPT's namings of a software product, none of that product's own pages was among the sources. The name usually sat on a roundup, a rival's comparison or a marketplace. Find which page it was before you change your own.

Keep Reading

35 Domain Authority and ChatGPT Statistics for Software (2026)

35 statistics on whether domain authority predicts which software brands ChatGPT names: the cited best-of lists tracked it far more closely.

45 Law Firm AI Search Statistics (2026)

45 law firm AI search statistics: ChatGPT opened 83.72% of lawyer answers with a rated card, and 33% of firms answered an emailed inquiry.

50 Personal Injury AI Search Statistics for Law Firms

50 personal injury AI search statistics: accident questions and lawyer questions barely share a source (1.70% and 7.65%).