GapCheck Report
Where CiteGround stands in AI answers, and what is winnable.
Prepared for
CiteGround
Property
citeground.com
Prompts
12 buyer prompts
Runs
118 archived
Engines
ChatGPT, Google AI Overviews, Perplexity
Window
rolling 28-day · 1 pull
Generated
2026-07-04 13:29 UTC
Prepared by
Ali Varinlioglu, CiteGround · [email protected] / [email protected] / https://cal.com/varinli/30min
Version
v1 · 2026-07-04
This is our own category read. We share it as a work sample. A client's read is built for them and stays theirs. Any number in here can be re-run live on a call.
Executive summary
CiteGround is absent from the AI answer on all 12 measured buyer prompts; Profound and Otterly lead a tight pack at 28% and 26% share of voice.
Share of voice
0% vs Profound 28%
Answer presence
12 absent of 12 measured
Winnable
6 called takeable · 1 high / 4 med / 1 low confidence
Evidence
118 runs · 28-day window
First moveShip citeground.com/gapcheck as the answer to the two free-checker prompts. What's cited there right now is just rivals' own product pages, not big domains. That's an open slot.
Prepared by Ali Varinlioglu, CiteGround · every number is from archived runs; the archive appends, nothing is rewritten
The verdict
Share of voice: you vs Profound
0% vs 28%
share of answers that mention the brand, all runs and engines in the window
Prompts by verdict
12 / 0 / 0
absent / present / unstable of 12 measured · 118 runs
Winnable prompts (analyst call)
6
called takeable · 12/12 prompts assessed
Confidence on the takeable calls
each call in the map carries its tag; low means it could flip with more runs
Share of voice by engine
ChatGPT 46 runs
Google AI Overviews 36 runs
Perplexity 36 runs
Calibration: in the 3 category baskets this archive has measured, the leading brand holds 28-95% share of voice.
Share of voice, every tracked brand
share of the 118 archived answers naming each tracked brand; brands can co-appear, so shares need not sum to 100%
Share of voice over time
youProfoundFirst pull archived 2026-07-03 (118 runs). Every future pull appends to this axis; movement stays attributable because archived runs are never edited.
First moveShip citeground.com/gapcheck as the answer to the two free-checker prompts. What's cited there right now is just rivals' own product pages, not big domains. That's an open slot.
The map: prompt-by-prompt judgment
Who the AI names, on what grounding, and whether the prompt is worth going after. The numbers are the archive's; the call column is the analyst's. Run counts differ per prompt because the window holds every archived run (baskets append; nothing is trimmed to a neat grid) and failed calls are skipped, never archived as answers.
| Buyer prompt | CiteGround | Call | Why |
|---|---|---|---|
| “best AI visibility checker for my brand” | absent | long shot med confidence |
Answer won't sit still. 35 domains, no single lever to pull, and the funded rivals already own the roundup slots. |
| 12 runs · agreement 33% · 35 domains · top: youtube.com, semrush.com, ahrefs.com · named: Otterly 5/12, Frase 3/12, Profound 3/12, Peec 1/12 | |||
| “how do I check if ChatGPT recommends my product” | absent | takeable med confidence |
Nobody's named here. It's a how-to answer built on reddit and vendor docs, so a decent method page just walks in. |
| 12 runs · agreement 100% · 18 domains · top: openai.com, reddit.com, bazaarvoice.com · no tracked brand named | |||
| “free tool to see which brands AI answers mention” | absent | takeable high confidence |
This one's ours. The answer names little indie tools, not big brands, and GapCheck is literally the thing being asked for. |
| 12 runs · agreement 83% · 36 domains · top: visibrand.app, areyoumention.com, viametric.app · named: Peec 1/12 | |||
| “track my brand mentions in ChatGPT and Perplexity answers” | absent | long shot med confidence |
Rotates across 27 domains with the funded names in it. Nothing concentrated to grab first. |
| 10 runs · agreement 33% · 27 domains · top: youtube.com, keyword.com, adobe.com · named: Otterly 3/10, Peec 3/10, Profound 2/10 | |||
| “best AEO tools for B2B SaaS” | absent | long shot med confidence |
Hubspot and youtube brand-mass run this. We could get into a roundup eventually. Slow and crowded though. |
| 9 runs · agreement 67% · 38 domains · top: hubspot.com, youtube.com, reddit.com · named: Profound 5/9, Otterly 2/9 | |||
| “Otterly alternative for AI search monitoring” | absent | takeable med confidence |
Built on small alternatives lists (ziptie, nightwatch, wellows) that refresh often and take submissions. Gettable. |
| 9 runs · agreement 67% · 24 domains · top: otterly.ai, ziptie.dev, nightwatch.io · named: Otterly 9/9, Peec 5/9, Scrunch 3/9, Profound 3/9, Trakkr 1/9 | |||
| “Profound alternative for startups” | absent | takeable low confidence |
Small third-party lists, different ones every run. You can land fast here. It just won't stick well. |
| 9 runs · agreement 33% · 22 domains · top: trakkr.ai, seranking.com, writesonic.com · named: Profound 9/9, Trakkr 5/9, Otterly 4/9, Scrunch 4/9, Peec 3/9 | |||
| “how to measure share of voice in AI answers” | absent | long shot med confidence |
Semrush and youtube own this and it's stable. A methods page can get cited, but it won't push them out any time soon. |
| 9 runs · agreement 100% · 20 domains · top: youtube.com, netranks.ai, semrush.com · named: Trakkr 3/9, Scrunch 3/9 | |||
| “tool that shows which sources AI engines cite” | absent | takeable med confidence |
The query describes exactly what GapCheck does. Grounding's shaky across small sites and nobody owns it. Open. |
| 9 runs · agreement 33% · 38 domains · top: therankmasters.com, siftly.ai, otterly.ai · named: Profound 4/9, Otterly 3/9, Scrunch 3/9, Peec 2/9 | |||
| “check my website AI search visibility” | absent | takeable med confidence |
Tightest prompt in the set, 14 domains, and free-checker pages like amivisibleonai already get cited here. The pattern's proven. |
| 9 runs · agreement 100% · 14 domains · top: semrush.com, webaloha.co, seranking.com · no tracked brand named | |||
| “generative engine optimization tools comparison” | absent | not worth it med confidence |
Too crowded. Rivals own the comparison hubs here. Put the effort on the alternative prompts instead. |
| 9 runs · agreement 33% · 39 domains · top: youtube.com, tryprofound.com, yotpo.com · named: Profound 6/9, Frase 4/9, Otterly 3/9, Peec 3/9, Scrunch 3/9 | |||
| “AI search rank tracker that is actually reliable” | absent | not worth it high confidence |
Skip it. This one wants per-query rank tracking, and we're on record that that's noise. Answering it argues against our own position. |
| 9 runs · agreement 67% · 41 domains · top: rankscale.ai, reddit.com, nightwatch.io · named: Otterly 2/9, Scrunch 1/9, Profound 1/9 | |||
The mechanism: one target, worked end to end
“free tool to see which brands AI answers mention”
takeable high confidence
This one's ours. The answer names little indie tools, not big brands, and GapCheck is literally the thing being asked for.
Target: citeground.com/gapcheck own product page
The move: ship the free-checker page structured to answer this exact prompt; the cited rivals here are self-listed product pages, which reads as an open slot
Archive receipt: citeground.com cited in 0 of 12 runs on this prompt.
Prompt grounding: 36 distinct domains across 12 runs; top: visibrand.app, areyoumention.com, viametric.app.
The roadmap is $1.5-2.5k, one-time. Half of it credits against your first month if you move to a retainer (those run $3.5-5k a month).
Exhibit A: one archived answer, verbatim
There is no widely known free tool specifically designed to see which brands AI answers mention, but WriteSonic (formerly WriteSonic) tracks brand visibility across AI chatbots like ChatGPT, Claude, Gemini, and Perplexity, though it is a paid platform with content optimization features [2]. For free brand mention tracking, Brand24 offers an AI Brand Assistant that summarizes brand mentions across social media, blogs, and news, with a 14-day free trial [5]. Another option is Perplexity.ai, a free AI search engine that links answers to sources, allowing you to manually check which brands are cited in its responses [6].
Tool Free Access Tracks AI Answer Brand Mentions Notes WriteSonic ❌ (Paid) ✅ Tracks brand visibility across AI chatbots [2] Brand24 ✅ (14-day trial) ⚠️ (General mentions) AI Brand Assistant for sentiment and context [5] Perplexity.ai ✅ ⚠️ (Manual check) Free AI search with source links [6] Glean AI ❌ ❌ Enterprise productivity assistant [1] If you need to monitor brand mentions in AI-generated content without paying, the best free workaround is to use Perplexity.ai or Google Assistant and manually review the cited sources for brand names [6]. For automated tracking, paid tools like WriteSonic are currently the only dedicated solution [2].
Cited in this run: glean.com, builtin.com, simular.ai, reddit.com, semrush.com, s-pro.io, spotzerdigital.com, youtube.com, soundhound.com · tracked brands named: none
Rendered from the archived answer's own markdown, nothing added or removed; the archive keeps the raw text, including the engine's citation markers. Every row in the map summarizes runs like this one; any of them can be re-run live.
The honesty block
Variance, in one breath. Of 12 measured prompts: 3 stable, 4 mixed, 5 unstable. Run agreement is shown per prompt in the map. AI answers are noisy; we count runs and show the noise instead of reporting one lucky run. One word does two jobs here: in the verdict column, unstable counts prompts where CiteGround flickers in and out of the answer (0 of 12); in the agreement bands, it grades run-to-run churn of the whole named set (5 of 12 churn hard). Different measurements, same word, both shown on purpose. Sensitivity: dropping the 5 unstable prompts, CiteGround holds 0% and Profound 13% share of voice (headline: 0% vs 28%).
Instrument conditions. Every run is a stateless API call: no user account, no chat history, no personalization. Model and geo were not recorded for these runs; recording is now part of every pull, so future reads carry them per run.
Confidence tags. Every judgment call above carries the analyst's confidence: high, med, or low. Low means the call could flip with more runs or a corpus change. The tag is stated so you can weight the call, never to soften it.
What this report does not measure
- Traffic, conversions, or revenue impact. This report measures who AI answers name and cite, not what that drives downstream.
- Anything outside the engines and window stated above. Models change; the archive keeps re-measuring so movement is attributable.
- Page-level citation detail beyond the registrable domain. Named target pages in the mechanism layer are verified by hand, not by an API field.
- Whether these 12 prompts are how buyers actually talk. This basket is our draft. We revise it with you on the call.
- Whether an open slot stays open. Sources shift and a rival can ship the same page. Takeable means it's open at pull time, not that it's reserved for us.
Analyst margin notes
ProvenanceJudgment calls were drafted with LLM assistance and human-verified by the analyst (A.V.). Margin notes are initialed.
A.V.Ali Varinlioglu · analyst of record[email protected] / [email protected] / https://cal.com/varinli/30minCheck our work. Every number above comes from archived runs, appended from run one and never edited: this report's archive is 118 rows, digest sha256 477262f2dd6d6022 (re-serialize, re-hash, compare). Any prompt in this report can be re-run live on a call; methods: https://citeground.com/methods.
Method: each prompt runs multiple times per engine; brands and cited domains are extracted per run and archived append-only from run #1. Run counts differ per prompt because every archived run in the window counts and failed calls are skipped, never archived as answers. No confidence intervals on purpose: runs within a prompt are correlated, so a binomial interval would overstate precision; the per-engine spread and the sensitivity line carry the uncertainty instead. Counts of runs, not certainty theater. Per-query AI rank is never reported here; it is statistical noise.