Platform · Competitor Research
Your brand against every competitor, in every AI answer.
A workspace tracks your brand and a named list of competitors. We scan every answer we collect for all of them at once: your share of voice, who gets mentioned first, and who moved this month. Our team reads that and turns it into the plan.
What a competitor is here
A name, the aliases behind it, and the domains it owns
In the platform a competitor is a name, the aliases an engine might use for it, and the web properties it owns. Zenith, Zenith CRM, and ZenithHQ are one brand, and zenith.io is its domain.
A single pass over each incoming answer, from any engine, checks your brand and every tracked competitor at the same time: how many times each is named, where the first mention sits, and which cited pages belong to whom. That last one is why the definition matters. Each cited page is tagged You, the competitor whose domain it is, or Third-party.
How it is measured
Three numbers that each answer a different question
Share of voice is your slice of every brand mention in the window, across you and your tracked competitors. It counts occurrences, so an answer that names Zenith three times adds three, and it is zero-sum: a point one brand gains, another loses. Mention rate counts answers instead of namings, which makes it the share of collected answers that mention you at all.
Position score answers who gets mentioned first. Each mention with a known place in the text scores 0 to 100 by how early it appears, and we average those scores. A brand named in the opening words scores 100. If a brand is only cited in the sources and never named, it is left out rather than scored zero.
- The low sample rule keeps a competitor with too few mentions to trust below the qualified brands until more data comes in. Your own brand is never held back by it.
- You can filter every figure by engine, country, prompt, tag, or persona.
Where the list comes from
An agent proposes the competitors. A person approves them.
At onboarding we ask a research agent with live web access for eight competitors, under these rules: research, do not recall; never invent a URL; verify every site; never return you. A clean-up pass then strips comparison wording, merges duplicates, drops dead sites, and caps the list at ten. What is left goes to a review screen where you edit names, aliases, and up to five websites each before anything runs. Our strategist goes through it with you.
The list is never frozen. Add a competitor later, rename one, or change its aliases, and detection re-runs across every answer already collected, so history is re-scored against the new definition rather than only the answers from that day on. A renamed competitor keeps its old name as an alias for the same reason. That re-scan runs on a queue, so it is not instant.
- Since one pass covers the whole set, you get the split by engine from the first run: your share on ChatGPT beside your share on Perplexity.
In the reports
Who moved, and who owns the sources
The cards our team reads every month, and what each one is telling you.
Competitor Movers
This card ranks every competitor from this window and the one before it by change in share points. New means no mentions last period and some now, and Dropped is the reverse: named before, gone now. Rows that barely moved are left out, and your own brand sits pinned above the list rather than ranked inside it.
Ownership on every cited source
Each cited page and domain is tagged You, the competitor that owns it, or Third-party. The first two tags show whose sites the engines lean on, while Third-party marks where there is presence to earn: review sites, communities, and publications the engines already trust in your category.
Brands you did not list
We measure only tracked competitors, so the platform will not surface a name it was never told to look for. The nearest signal is a Third-party domain that keeps getting cited, which the reports flag as a rival worth tracking. When one surfaces, we add it and re-score the history.
Why it matters
The answer names two or three brands. The question is which.
A mention on its own tells you little. Share of voice says what the engines are handing to someone else. Movers name who is taking it, and ownership shows where they earned it.
5
AI engines scanned per answer
ChatGPT, Gemini, Perplexity, Copilot, and Google AI Mode. We check each answer for every tracked brand in one pass.
Daily
Every prompt, every engine
Each prompt runs once a day on every enabled engine, and we store every answer in full, with its sources.
8
Competitors researched at onboarding
The agent verifies that each site is reachable before proposing it. Up to ten survive clean-up, and you approve the list.
0 to 100
Position score
A brand named in the first words of the answer scores 100. One that is cited but never named is left out rather than scored zero.
Inside our GEO work
The benchmark behind the monthly report
Prompt tracking says whether you are in the answer. Competitor research says who else is.
01 · DISCOVERThis feature
We map your presence against the field rather than on its own: who is named instead of you, how often, how early, and on which engine, prompt by prompt.
02 · OPTIMIZE
The competitor gap strategy in content routines picks the prompt where competitors are most visible and you are least, then briefs the writer with the top five brands named there.
03 · DETECT
Movers, per-engine share, and citation ownership catch recommendations changing before that change reaches your pipeline. You see a newcomer with no mentions last month, or a rival domain climbing the cited list.
Related features
Where the competitor set shows up next
- Every competitor number is built from these daily runs, with each answer stored in full and every mention highlighted.
Prompt Tracking
- Share of voice, movers, and ownership sliced by engine, country, prompt, tag, or persona, saved as a report you can open any time.
Custom Reports
- The competitor gap strategy turns the prompt where you trail the most into a drafted article for your approval.
Content Routines
- How each engine describes your product against the competitors it names, and the pages that get it to repeat your positioning.
Product Marketing
Questions about competitor research
What clients ask about the benchmark
How do you pick which competitors to track?
At onboarding we ask a research agent with web access for eight competitors, each with a verified live website, and a clean-up pass removes duplicates, comparison phrasing, and any entry that is really you. You and our strategist edit that list before the first run: rename, add aliases, add or remove brands. After that we add competitors as they surface in the reports.
Can you see competitors I did not list?
Not by name. Detection matches the aliases and owned domains of the brands in your list, so a brand the platform was never told about is not counted. The closest signal is the Third-party tag on cited domains and business profiles, which marks a source that is neither yours nor a tracked competitor. When one keeps appearing, we add it as a competitor and re-scan every past answer.
What is the difference between share of voice and visibility?
Visibility, or mention rate, is the share of collected answers that name you at least once. Share of voice is your share of all brand mentions, yours and your competitors together, counted per occurrence. If competitors are named more often than you are, visibility can rise while share of voice falls. Both are in the report because they answer different questions.
How do I know if a competitor is gaining?
The Competitor Movers card compares the current window with the equal-length window before it and ranks every competitor by the change in share points. Grew and shrank mean what they say. New is a competitor with no mentions last period, and Dropped is one that had mentions last period and none now. Rows with too few observations are dimmed, and your own brand is shown on its own rather than ranked with the rest.
Does it tell me whether AI recommends a competitor or just names them?
No. Detection is matching, not sentiment analysis: it finds aliases at word boundaries and owned domains in the text and the citations, and records how often and how early. It does not judge whether a mention is a recommendation, a comparison, or a complaint. For that our team opens the answers themselves, which are stored in full with every mention highlighted.
See what AI is already saying about your brand.
A free audit shows you exactly where you stand across ChatGPT, Gemini, Perplexity, Copilot, Claude, Grok, and AI Overviews, and what it would take to close the gap.