Find the AI Prompts Where Competitors Are Named and You Aren't

Updated

A prompt where AI answers name your competitors and not you is the clearest gap in AI search: the buyer asked, the answer picked products in your category, and yours wasn't one. Sightkick lists these prompts from its daily tracking, and your own AI agent can turn each into a page to fix or a cited page to pitch.

Sightkick makes this page. The worked example is sightkick.so's own tracking data, read on 2026-11-04.

Key takeaways

  • On sightkick.so, 58 tracked prompts named rivals and never named Sightkick in 60 answers each, from 2026-10-21 to 2026-11-04.

  • 42 of the 58 already had a sightkick.so page meant to answer them, and 8 of those 42 pointed at a page Google hadn't indexed.

  • One prompt and four read-only tools give your agent the gaps, their verdicts, rivals' new pages and the cited pages to pitch.

  • Judge over 14 days or more. 0 mentions in 60 answers means a rate likely under 6%, not zero.

  • Only ChatGPT, Gemini, Google AI Overviews and Google AI Mode are tracked.

Why does a prompt that names your rivals but not you matter?

A prompt that names your rivals but not you shows demand that already exists and a shortlist you're missing from. The buyer is comparing, the AI answer is picking products, and the pick goes elsewhere. Comparative questions name brands far more often than informational ones: across four AI models, comparative queries drew a brand mention 43.3% of the time against 18% for informational ones (Semrush with Kevin Indig, 2026-06-09).

The answer also shows what it was built from. Every cited source is a page someone can change: your own, if a page of yours should be the answer, or a third party's list, review or thread that names your rivals and could name you. That makes a competitor gap something to act on, which a single share-of-voice score isn't. How that score is built is in how to measure AI share of voice.

On sightkick.so the gap is wide. From 2026-10-21 to 2026-11-04, Sightkick was named in 23 of 5,960 answers to 85 unbranded prompts (0.4%), while Profound was named 2,043 times, Peec AI 1,988 times and Otterly 1,875 times in the same window's answers (Sightkick tracking for sightkick.so, get_visibility summary, read 2026-11-04). Method: each tracked prompt is asked once a day in each of the four AI models, and every answer is stored with the brands it names and the sources it cites.

What prompt do you give your AI agent?

Paste this prompt into Claude Code, Codex, ChatGPT or any AI agent connected to Sightkick. It calls four read-only tools, groups the gaps, and proposes one next step per prompt from what the tools return.

Use Sightkick to find the AI prompts where our competitors are named and we aren't.

1. get_visibility, view gap_candidates, limit 15: the prompts AI answers left us out of over the last 14 days, the rivals named instead, and the sources cited.
2. get_visibility, view gaps: last week's verdict on each gap (covered by one of our pages, planned, or offpage) and why.
3. get_visibility, view rival_pages: comparison, alternatives and pricing pages our rivals added this week.
4. get_outreach, view prospects: cited pages that name rivals and not us, by Opportunity.

Then give me one table: prompt, answers checked, times we were named, rivals named with counts, top 3 cited URLs, verdict, next step.
The next step is one of: improve or get indexed the page we have (name it), write a new page, pitch a cited page (name it), or leave it (say why).
Use only what the tools return. Never judge from a single answer: say how many answers each row rests on.

To connect an agent first, in Claude Code: claude mcp add --transport http sightkick https://app.sightkick.so/mcp. Other clients are on the MCP server page, and the CLI (npx -y sightkick) runs the same tools with an API key.

Which Sightkick tools does the agent call?

The agent calls three views of get_visibility and one of get_outreach. All four are reads and change nothing.

The four Sightkick tool calls behind a competitor gap review, as of November 2026

Tool and view

What it returns

What it's for

get_visibility gap_candidates

Prompts the answers left you out of over 14 days, worst first, with the rivals named instead, the most-cited URLs, and whether a page of yours could win it

The list of gaps

get_visibility gaps

Last week's gaps, each marked planned, covered (with the page) or offpage, with the reason

Deciding between your site and other sites

get_visibility rival_pages

What each tracked rival added to its sitemap, read every Monday, with new comparison, alternatives and pricing pages flagged

Spotting the page a rival just shipped

get_outreach prospects

Pages AI answers cite where rivals are named and you aren't, by Opportunity, with citations in the last 30 days, weeks cited out of the last 8, the contact and a drafted pitch

Choosing which third-party pages to pitch

Sources: Sightkick MCP server, Sightkick tool catalog

Two more help with the follow-up: get_prompts (view prompt) opens one prompt's answers per AI model, and get_pages (view list) says whether Google has indexed the page that should answer.

What did the agent find on sightkick.so?

On 2026-11-04 the gaps view listed 58 prompts for sightkick.so, each asked 60 times between 2026-10-21 and 2026-11-04 (once a day in each of the four AI models), with Sightkick named in none of the 60 answers. Last week's review marked 42 as covered by an existing sightkick.so page and 16 as off-page. Checking the covering pages against Google's index split the 42 further: 8 point at a page Google hadn't indexed.

sightkick.so's 58 competitor gaps by verdict, 2026-10-21 to 2026-11-04 (as of November 2026)

Verdict

Gaps

What it means

Next step

Covered, page indexed

34

A sightkick.so page answers the prompt, and Google has it

Improve that page with the facts the cited pages carry

Covered, page not indexed

8

The answering page exists, but Google hasn't indexed it

Get it indexed before writing anything

Off-page

16

Third-party pages lead, or the prompt asks for something the product facts don't back

Pitch the cited pages, or leave it

Planned

0

A new page is on the schedule

Write it

Sources: Sightkick get_visibility gaps and get_pages list for sightkick.so, read 2026-11-04, via the Sightkick tool catalog

The same rivals fill almost every gap. Of the 58 prompts, Profound was named in 55, Otterly in 54, Peec AI in 52, Scrunch AI in 39, Rankscale AI in 30 and Promptwatch in 15 (Sightkick tracking for sightkick.so, read 2026-11-04). The tracked prompt set changed on 2026-11-03, so this window mixes the old and new sets, but each of the three prompts below was asked on every day of it.

A page that exists but Google hasn't indexed

"What are the best alternatives to Peec AI for smaller startups on a budget?" was asked 60 times, and Sightkick was named in 0. Named instead: Peec AI 58 times, Otterly 55, Profound 24, Rankscale AI 19 and Scrunch AI 10 (Sightkick tracking, 2026-10-21 to 2026-11-04). The most-cited sources were a Peec AI alternatives post on frase.io (17 citations), one on scalenut.com (11) and rankscale.ai's homepage (11). The verdict was covered: sightkick.so already has a Peec AI alternatives page. But Google listed that page as "Discovered – currently not indexed" on 2026-11-04. The next step is getting it indexed, not writing a second page.

A page that exists but lacks the facts the answers want

"What's the pricing structure for most AI brand visibility tracking platforms these days?" was asked 60 times, and Sightkick was named in 0. Named instead: Otterly 43 times, Profound 36, Peec AI 33, Scrunch AI 20 and Rankscale AI 10 (Sightkick tracking, 2026-10-21 to 2026-11-04). The most-cited pages were Zapier's list of the best AI visibility tools (17 citations) and a Rankability post on what to pay for AI visibility tracking (17). The verdict was covered by 8 Best AI Search Visibility Tools for 2026, and its reason says dated prices belong there. So the move is to add each tool's price and its date to that page. On the off-page side, get_outreach lists a 2026 guide to AEO and GEO pricing on outreachbloom.com that names four of these rivals and not Sightkick, cited in 16 answers over the last 30 days across 6 tracked prompts.

A gap no page should close

"Which AI visibility platforms are best suited for agencies that need multi-client dashboards?" was asked 60 times, and Sightkick was named in 0. Named instead: Profound 42 times, Peec AI 40, Otterly 33 and Rankscale AI 14 (Sightkick tracking, 2026-10-21 to 2026-11-04). The verdict was offpage: the answers cite agency-specific vendor pages and comparisons, and the product facts don't establish a multi-client agency dashboard. A page claiming one would make a claim the product can't back. The right move is to leave it until the product changes.

The last two reads complete the picture. rival_pages, read on 2026-11-02, showed no new comparison, alternatives or pricing page since 2026-10-28 from the six rivals whose sitemaps it read. prospects listed 44 open prospects and 70 in progress, none won yet. The top open one is a 2026 list of AI citation tracking tools on therankmasters.com: Opportunity 99, cited in 5 answers over the last 30 days and in each of the last 8 weeks, in all four AI models, naming Otterly, Profound, Peec AI and Scrunch AI (get_outreach prospects, read 2026-11-04).

What do you do with each gap?

Each gap gets one of two moves: make a page of your own the best answer, or get named on the pages the answers already cite. The verdict tells you which.

  • Covered: improve the page you have. Check that Google has indexed it, that it answers the prompt in its first lines, and that it carries the facts the cited pages carry: prices with dates, the rivals the answer names, what each is best for. Don't write a second page for the same question.

  • No page yet: write one. One page per buyer question, the answer in the first sentences under each question heading, and something the cited pages don't have, such as your own data. A page your agent writes is scored and scheduled with the same tools.

  • Offpage: pitch the cited pages. Lists, reviews, Reddit threads and YouTube videos that name your rivals are where the answer gets its shortlist. Pitch the ones cited week after week, and only with claims your product backs. The platform finds the contact, drafts the pitch and sends it from one inbox, by itself or on your say-so.

  • Not your fight: leave it. If the prompt asks for something you don't do, no page or pitch should pretend you do.

To compare your pages with your rivals' in Google as well, see SEO competitor analysis. To show the gaps to a team or a client, see how to create an AI visibility report.

What are the limits of competitor gap tracking?

  • One answer is a sample. Two answers to the same prompt share the same brand list less than 1 time in 100 (SparkToro and Gumshoe, 2026-01-28). Judge over 14 days or more. Even 0 mentions in 60 answers only says the true rate is likely under 6% (95% Wilson interval: 3.84 ÷ 63.84).

  • Four AI models. Sightkick tracks ChatGPT, Gemini, Google AI Overviews and Google AI Mode (3 included, the 4th as an add-on). Claude and Perplexity are not tracked, so a gap there won't show.

  • The prompt set decides what you see. Gaps exist only among the prompts you track. Lead with unbranded prompts: across 3,989 consumer-app brands, the median brand was named in 95% of answers to branded prompts and 7% of answers to unbranded ones (Searchable, 2026-10). A change to the set breaks the trend line, as sightkick.so's did on 2026-11-03.

  • Sources rotate. The median cited page loses half its citation share within 11 days of its peak, across 883,000 pages (Profound, 2026-09-30). Weeks cited out of the last 8 tells a steady source from a passing one.

  • Verdicts are judgments. Covered, planned and offpage come from the weekly review against your product facts. Read the reason before you act.

Frequently asked questions

How is a competitor gap different from AI share of voice?

Share of voice is one number across all prompts. A competitor gap is one prompt, with the rivals named and the sources cited, so it points at a page to change. The two move together, but only the gap says where to start.

How many answers do you need before acting on a gap?

At least 14 days of daily answers per AI model. Sightkick's gap_candidates view uses 14 days, which came to 60 answers per prompt for sightkick.so. A single answer, or a single day, can show a gap that isn't there.

Can you find competitor gaps without a tool?

Partly. Ask the same buyer question in each AI model every day, in a chat with no memory, log the brands and sources, and count over two weeks. For 85 prompts in four AI models that is 340 answers a day, which is why it's usually automated.

Does Sightkick track Claude or Perplexity?

No. Tracking covers ChatGPT, Gemini, Google AI Overviews and Google AI Mode.

What does Sightkick cost for this?

$39 per month on Standard or $99 per month on Pro, as of 2026-11-04, each with a 7-day trial that needs a card. Both include agent access and 50 tracked prompts. Details are on the pricing page.

Sources