CatalystRead the report
Research report · November 2026

The AI Visibility Gap

Find out your exact problem and how to fix it.

You get the full report on this page and the PDF by email. We’ll also send future Catalyst research. Unsubscribe anytime.

Cover of The AI Visibility Gap
186
Company scans
12
Software categories
4
AI assistants
7
Practitioner interviews

What you’ll learn

Seven senior practitioners from G2, Ramp, Sprout Social, AirOps, Privy, QuotaPath, and Carta contributed to this report. All are well ahead of the field, and still consider the work unfinished.

  1. 1

    How to tell whether your problem is getting named or getting chosen, using only an LLM and an afternoon, and how to fix each.

  2. 2

    Which four signals help only with initial visibility, and which two continue to matter.

  3. 3

    Why the team that usually owns this work is better suited to the smaller part of the problem.

  4. 4

    Why the leading teams still can’t show that this work produced revenue.

  5. 5

    What happened twice in a year to the most cited source in this market, and why it matters.

Seven findings

  1. 01

    AI visibility is two different problems under one label.

    Getting named moves a company from zero visibility to some visibility. Getting chosen moves it from appearing on a handful of questions to appearing on most. Our data associates the two stages with different signals. A G2 profile appears on 16.3% of companies scoring zero and 54.5% of those scoring above zero. That is a 38.2-point difference at that threshold, and it turns slightly negative at higher visibility.

    The two ladders
    NAMED ON MOSTANY VISIBILITY AT ALLNEVER NAMEDG2 profileBlogCrunchbaseReddit presenceOrganization schemaGetting namedMoves a company from zerovisibility to some visibility.Ranks first on own brand nameYouTube channelGetting chosenMoves it from appearing on a handfulof questions to appearing on most.
  2. 02

    Three in ten are invisible, while a larger group has partial visibility.

    Assistants never named 30.1% of our list. A larger group, 59.1%, appears somewhere but misses most of the questions it could win. Most readers of this report are likely in that second group.

    Where companies sit
    30.1%Never named once31.7%1% to 20% of answers27.4%20% to 50%10.8%Above 50%Named somewhere, absent on most · 59.1%VOICE SHARE ACROSS THE LIST0%50%100%Median 12.0%Mean 20.1%

    Shares of the 186 companies in our list, as of 2026-08-26.

  3. 03

    The leading brand takes most of the available answers.

    Across the 12 scanned categories, the top brand holds a median 87.5% of answer slots, compared with a category average of 26.0%. For comparison with SEO, earlier studies put the first organic search result at 27.6% to 39.8% of clicks. AI answers are more concentrated, which makes coverage work more valuable.

    Leader share by category
    39.8%TOP ORGANIC RESULT68.7%TOP THREE RESULTSCRM97.9%27.8%Marketing automation95.8%31.0%Payroll93.8%24.2%Customer data platforms93.8%29.7%AP automation91.7%25.0%Sales engagement87.5%25.9%Expense management87.5%26.1%HRIS81.2%29.4%Employer of record80.0%14.4%Cold email software62.5%28.0%Institutional crypto custody56.7%16.8%Stablecoin payroll40.0%10.7%Leader's shareCategory average0%25%50%75%100%SHARE OF AVAILABLE ANSWER SLOTS

    Twelve fully scanned B2B software categories. The dashed lines mark the top of the range earlier studies give for the click share of the first organic search result and of the top three.

  4. 04

    The default owner is better suited to the smaller part of the problem.

    In one survey of 599 SEO, content, and growth professionals, over half said the team that already owns search runs this work. Barely one in ten had a dedicated person or team for answer engine optimization (AEO). Search teams usually focus on content and page structure, which our data associates mainly with getting named. That makes the default ownership model a better fit for the 30.1% at zero than for the 69.9% already above it.

    Who owns it, and with what
    WHO OWNS AI VISIBILITY52.8%say the team that already ownssearch runs it9.2%have a dedicated specialist or teamTwo in five have no dedicated budget for it.ONE SURVEY OF 599 SEO, CONTENT,AND GROWTH PROFESSIONALSMARKETING TEAMS IN MID-2026Cutting agency or vendor spend47.7%No longer backfilling open roles40.3%Expanded their AI capacity15.5%AIROPS' OWN SURVEY OF 307 MARKETING LEADERS,JULY AND AUGUST 2026, SELF-REPORTED

    The two panels come from different surveys.

  5. 05

    Nobody can connect the work to revenue.

    All seven contributors described a measurement gap. Nearly half of marketers in a second survey said they struggle to measure visibility in AI answers. Fewer than one in ten said they can measure all the metrics that matter. But the demand to prove revenue impact is here.

    Measuring visibility in AI answers
    Struggle to measure it45%Can measure all that matters9%

    Semrush survey of 481 marketers, June 2026.

  6. 06

    The tactics expire faster than teams can copy them.

    In August 2026, the most cited third-party domain in this market lost most of its ChatGPT citations within a week. It was the second collapse in a year. On average, half of a cohort of cited sources drops out of AI answers within 4.5 weeks. A playbook copied from a leader can be out of date before it’s applied.

    Reddit’s August collapse
    0%1%2%3%4%JUL 18AUG 1AUG 7AUG 14AUG 173.83%in the window ending August 70.52%AUG 18 · OUR LAST INTERVIEWAUG 14 · THE BREAKTWO REPORTED WINDOWS. THE DASHED LINE BETWEEN THEM IS DRAWN, WITH NO DATA BEHIND IT.A SIMILAR COLLAPSE IN SEPTEMBER 2025 HAD RECOVERED BY MID-2026.

    Reddit's share of ChatGPT citations. Source: Promptwatch, July 18 to August 17, 2026.

  7. 07

    Markup looks like an early-stage signal.

    Organization schema appears on 55.4% of companies at zero, rises to 74.1% in the 1% to 20% band, then falls to 55.0% among leaders. llms.txt and FAQ schema follow the same pattern. Markup appears to help an assistant identify a company but contributes little beyond that. That result is consistent with a controlled study across 252,000 trials in which formatting alone had almost no effect.

    Organization schema by band
    At zero55.4%In 1% to 20%74.1%Above 50%55.0%

    Share of companies with Organization schema, from the 186 company scans.

Six chapters

Read the report
The signal split
Off the floorThe climb−100+10+20+30+40G2 profile−7.8+38.2Has a blog−3.1+34.1Crunchbase page−8.2+27.0Reddit presence−5.3+22.8Ranks first on own brand name+12.3+16.4YouTube channel+13.9+14.7PERCENTAGE-POINT DIFFERENCE

Off the floor compares companies above zero with companies at zero. The climb compares companies above 50% with companies at 1% to 20%, and rests on 17 to 20 companies, so it is directional.

  1. 1

    Getting named, then getting chosen

    G2 shows a 38.2-point difference at the zero threshold and a negative 7.8-point difference higher up.

    38.2 points
  2. 2

    Never named once

    Across a full set of their own buyer questions, assistants never named three in ten companies in our list.

    30.1%
  3. 3

    Winner takes the answer

    Across all 12 categories we scanned, the leading brand holds a median of 87.5% of available answer slots.

    87.5%
  4. 4

    Nobody can prove it worked

    In a Semrush survey of 481 marketers, 45% said they struggle to measure visibility in AI answers, and only 9% said they can measure all the metrics that matter.

    45%
  5. 5

    The tactics expire before you can copy them

    The average AI citation half-life is 4.5 weeks across 3.5 million citation events and more than 120,000 domains.

    4.5 weeks
  6. 6

    Everyone’s job, nobody’s budget

    In one survey of 599 SEO, content, and growth professionals, 52.8% said the team that already owns search runs AI visibility.

    52.8%

Seven practitioners

  • Mohammad FarooqDirector of SEO, G2
  • George BonaciVP of Growth and Demand, Ramp
  • Lauren KeudellDirector of SEO, Sprout Social
  • Debbie SoonHead of Marketing, Privy
  • Travis NorthupDirector of Demand Generation, QuotaPath
  • Lucy HoyleSenior Content Engineer, Carta
  • Josh SpilkerHead of Search Marketing, AirOps
The free audit

Scan your company against its category.

Catalyst helps B2B companies get named and chosen when buyers ask AI assistants for advice. We build the content that answers those buyer questions, and we measure who gets named in the answers.

We offer the scans behind this report as a standalone free audit. It reports voice share by buyer question across five AI platforms, identifies which of the two problems in this report applies to your company, and shows which questions a competitor owns and which questions remain unclaimed.

Get a personalized AEO reportRead the report