Skip to content

ResourcesCompare

Switching AI visibility platforms: what to check in citation share and prominence

Most switching mistakes are counting mistakes. Before you set an old figure beside a new one, find out what each tool counts and what it divides by.

Updated Jack F, Co-founder

Part of Answyn Compare.

Now readingKeep metric parity

To switch AI visibility platforms without losing your baseline, keep metric parity: the same questions, engines and market, counted the same way. Export answers, citations, the question list and history before you cancel. Find out what the new tool divides by, whether it counts a brand once per answer or every mention, and how it merges aliases. Then run both tools in parallel for at least four weeks, and compare trends, never absolute figures.

This is our page, and we sell one of the platforms on it, so read what follows as an argued position rather than a neutral audit. Where another tool is the better buy for a team leaving Rankscale or Brandlight, we say so, and the definitions in the tables are the vendors' own.

It is written for a team that already tracks AI visibility and wants to move without losing its baseline. If you are choosing a first platform, start with AEO software for UK marketing teams or our guide to the best AEO platforms for UK businesses.

Keep metric parity, or the old baseline stops meaning anything

Metric parity means two tools measure the same thing the same way: the same questions, engines and market, the same counting rules and the same denominator. Without it, a rise or fall after a switch can be nothing more than the new tool's arithmetic.

Rankscale's own help article shows how far two honest methods can sit apart. Its Visibility Score takes the share of runs that include your brand and discounts it for average position, so a brand found in 80 of 100 runs at an average position of 2 scores about 72.7%. A tool that reports a plain mention rate would show 80% for the same answers. Both are correct by their own definitions, and neither can be read against the other.

Definitions can differ inside one vendor, too. Ahrefs' Brand Radar page describes AI Share of Voice as the share of responses in your topic set that mention or cite your brand. Its help article on AI visibility metrics, dated 3 September 2026, describes it as a brand's share of impressions, summed from search volumes, against other tracked brands. Before you trust a new dashboard, ask which definition it uses and when it last changed.

The switching checklist

Work through these in order. The first two protect your history; the rest stop you comparing figures that were never alike.

  1. Export before you cancel. Take the answers with their citations, the question list, the competitor list with its aliases, and as much history as the old tool will release. Find out what happens to the data when the contract ends.
  2. Keep the question set, engines and market the same. Bring the exact questions across in the same wording, ask them of the same engines from the same country, and match the cadence or compare windows of the same length. A new question set, a dropped engine or a different market moves every figure by itself.
  3. Find out how the new tool counts. Ask what it divides by (every answer, every prompt, or only answers that name a brand), whether a brand counts once per answer or every time it is named, whether spellings of one company are merged, and whether names that are not rivals, such as the assistants themselves, can be excluded.
  4. Check what sits behind each figure. Can every metric be split by engine? Can you open the answers that produced a number? How long does the plan you are buying keep history? Published history runs from none on Profound's free Trial to all-time on its Enterprise plan.
  5. Run old and new in parallel. Keep the old tool running on the same questions for at least four weeks after the new one starts, so the two baselines overlap and you can see how far apart they sit.
  6. Never compare absolute figures across tools. Compare each tool with itself over the same weeks. If both move the same way, the trend is probably real. If only one moves, read the answers behind it before you report anything.

For the two tools this page is usually asked about, the exit terms are published. Rankscale's help article on exporting says an export can include every execution, with the answer text where the engine allows it, citations and sentiment, as CSV or straight to Google Sheets, and that its All option covers the complete history of the selected search terms. Brandlight's product page says customers export in CSV or JSON at contract end, and that deletion follows within thirty days.

Our own data shows why item 3 asks about spellings. In Answyn's own tracking of its category (UK market, ChatGPT, Google AI Overviews and Claude, 6 October to 4 November 2026), Consider-stage answers named Otterly as "Otterly.AI" in 262 answers, "OtterlyAI" in 71, "Otterly.ai" in 39, "Otterly AI" in 28 and "Otterly" in 20. Those counts cannot simply be added, because one answer can use two spellings. A tool that does not merge them splits one rival into five, and every share of voice beside it shifts.

How each platform defines visibility, share of voice, position and citations

When ChatGPT answers these questions, the pages it cites most are vendors' own. In Answyn's own tracking of its category (UK market, 6 October to 4 November 2026), the four domains ChatGPT cited most for the five citation-share questions we track, which include both switching questions, were tryprofound.com (102 citations), otterly.ai (97), semrush.com (89) and peec.ai (86), and brandlight.ai was cited 41 times. Each vendor defines the metrics its own way, so the tables below set the published definitions side by side, as read on each vendor's pages on 4 November 2026. "Not published" means the pages we read do not say.

Visibility and share of voice, as each vendor defines them (read 4 November 2026)
PlatformVisibility, and what it divides byShare of voice
RankscaleDetection Rate is the share of runs whose answer includes you. Visibility Score discounts that rate for average position.Your portion of total mentions against competitors. One answer counts as one mention.
BrandlightA visibility score per engine and category, "built and weighted on thousands of real questions". Formula not published.Named on the homepage. Not defined.
AnswynMention rate: the share of collected answers that name you in their own words. Responses with no AI answer are left out.Your appearances over all brand appearances in the same answers, each brand counted once per answer.
ProfoundVisibility Score: responses that mention you, out of responses that mention at least one brand.How often you are mentioned compared with all brands. Counts mentions, not responses.
Peec AIThe percentage of AI responses in which your brand appears.Your mentions as a percentage of all tracked brands' mentions.
Semrush AI Visibility ToolkitAI Visibility: a 0 to 100 benchmark of how often you appear compared with competitors.Your percentage of the mentions in AI answers, against competitors in your market.
Otterly.AIBrand Coverage: prompts that mention you, as a percentage of all prompts.Your brand mentions divided by all brand mentions, times 100.
Ahrefs Brand RadarMentions: responses that name you at least once, each counted once.Product page: responses in your topic set that mention or cite you. Help article: your share of impressions.
SE Visible (SE Ranking)Mentions weighted by where they appear in the answer. Answers without you count as zero.Your share of mentions compared with the competitors you track.
ScrunchPresence: named explicitly in the written answer. A name that appears only in a cited URL does not count.Competitive Presence, normalised so every brand's slices sum to 1.
HubSpot AEOHow often you appear across your tracked prompts, as a percentage.How often you are mentioned compared with competitors across the same prompts.
AthenaHQMention rate: how often AI answers name your brand.Share of voice across competitors. Method not published.
AuthoritasNot published.AI Share of Voice for branded and unbranded searches. Definition not published.
ConductorEach response is marked Mentioned or Not Mentioned. A visibility rate is not defined.Your mentions compared with every brand in the responses.
FraseThe share of measured answers that named or cited you, so a citation alone counts.Of every brand named in your answers, how often it was you.
PromptwatchA Brand Visibility Score is named but not defined.Share of Model Voice is named but not defined.
SearchablePresence Rate: tracked prompts that mention you. Visibility Score, 0 to 100, blends frequency, position, context and sentiment.Your mentions as a percentage of total category mentions.
Similarweb AI Search IntelligenceResponses that mention you, out of all analysed answers.Not on the product page.
Position and citations, as each vendor defines them (read 4 November 2026)
PlatformPosition or prominenceCitation metric
RankscaleAverage Position: where you rank when you appear, lower is better. Top 3 Visibility counts only high placements.A citation is a source or URL in an answer. Share of Citations shows how citation presence splits between you and rivals.
BrandlightCategories from Lead source to Buried mention. Average position is shown but not defined.Citation Impact Score: "the share of your visibility that traces back to each source".
AnswynProminence: first, among others, or only in passing. Average position: order of first mention among the brands named.Citation share: how often answers cite a page, and how often that page is yours. Own and third-party citations split.
ProfoundAverage Position relative to competitors across responses. A lower value is a higher rank.Citation Share: the percentage of all citations that point to a domain, page or citation category.
Peec AIAverage ranking when you appear. Win rate: the share of answers that rank you first.Citation share: your domain's share of inline citations. Used as a source: chats that used one of your pages.
Semrush AI Visibility ToolkitAverage Position: where a citation of your domain typically appears in responses.Citations: the number of responses that cite your domain. Cited Pages lists your pages.
Otterly.AIAvg. Brand Position: the average position of your brand in AI answers.Domain Coverage: prompts that cite your domain, as a share of all prompts. Link positions tracked daily.
Ahrefs Brand RadarNot published.A page counts once for each response that cites it. A domain counts once if any of its pages is cited.
SE Visible (SE Ranking)Where you usually appear in the answers that mention you.Domain and URL citations, your own pages listed separately, and a Gap score for pages citing rivals but not you.
ScrunchPosition Score: 0 to 100 in three buckets, top 100, middle 50 and bottom 0.Citations Rate: responses that cite the domain or URL, divided by all responses.
HubSpot AEONot published.Owned domain citation rate against competitors. Top owned citations lists your most-cited pages.
AthenaHQAverage ranking position in AI answers.Citation rate: how often AI answers link to you.
AuthoritasTracks how models rank brands over time. Definition not published.Compares citations and mentions across models. Own-page report not published.
ConductorNot defined.Cited means your website appeared as a citation. Top cited pages lists yours.
FraseYour rank among tracked brands on Visibility and Share of voice, averaged.Which sites answers cite. An own-page report is not described.
PromptwatchNot published.Citation Share and Source Origin are named but not defined.
SearchablePart of Visibility Score. Not defined separately.Citation tracking by domain. An own-page report is not described.
Similarweb AI Search IntelligenceNot on the product page.Citation Analysis: which websites most influence AI brand visibility. Page level not described.

Where the definitions differ most

Visibility figures divide by four different things: answers or runs (Answyn, Peec AI, Rankscale, Frase, Similarweb), prompts (Otterly.AI, HubSpot AEO and Searchable's Presence Rate), only the responses that name at least one brand (Profound), or nothing at all, because the figure is a score (Semrush, Searchable's Visibility Score, Brandlight). Two count a citation as presence: Frase's visibility counts answers that named or cited you, and the AI Share of Voice on Ahrefs' product page counts responses that mention or cite you. Scrunch and Answyn count only a name in the answer's own words.

Prominence comes in three forms. Profound, Peec AI, Otterly.AI, Rankscale, SE Visible and AthenaHQ report an average position, where 1 is first. Scrunch sorts mentions into top, middle and bottom buckets, Brandlight into categories from Lead source ("The answer is built mainly on you") to Buried mention, and Answyn into first, among others or only in passing, alongside an average position. Rankscale, SE Visible and Searchable also fold position into their visibility figure, and Frase's average position is something else again: your rank among the brands you track. A bucket cannot be turned into an average, so if the form changes, start a new series.

Citation metrics split the same way. Profound's Citation Share and Peec AI's citation share are your share of all citations. Otterly.AI's Domain Coverage, Scrunch's Citations Rate and Peec AI's "Used as a source" count the prompts or answers that cite you at least once. Brandlight's Citation Impact Score runs the other way, tracing your visibility back to each source, and Semrush's Average Position describes where a citation of your domain sits in the answer. A share of all citations and a share of answers citing you can move in opposite directions in the same month.

Answyn on the same footing

Answyn counts like this, in the same terms as the tables, so you can run the checklist on us too.

  • Mention rate, our visibility figure, is the share of collected answers in the window that name you in their own words. A name that appears only as a cited link is a citation, not a mention, and a response marked No AI answer, where the engine returned nothing, is left out of the base.
  • Share of voice is your appearances divided by all brand appearances in the same answers, with each brand counted once per answer, so a name repeated five times in one paragraph counts once.
  • Prominence records how clearly you are named: first, among others, or only in passing. Average position is the order of first mention among the brands an answer names, where 1 is first.
  • Citation share is how often answers cite a page, and how often that page is yours rather than a review site, a competitor or a third-party list. The Sources screen splits own-domain from third-party citations and lists your most-cited pages.
  • Aliases count as the same brand, and when you group a rival's spellings, the grouping applies to recorded answers as well as new ones. A name marked not a competitor stops counting towards share of voice, rankings and average position everywhere, and the AI engines and the companies that make them start on that excluded list.

Every headline figure in AEO Hub opens back to the answers that produced it, and most Analytics pages filter to a single engine. How much history is kept is set per plan, up to unlimited. Export answers downloads a CSV of collected answers with their mentions, citations and topics for the dates you choose, within that history. It is the export we would want from any tool we were leaving.

On parity, plainly: Answyn collects Gemini, ChatGPT, Google AI Overviews, and Google AI Mode every day, and can add Microsoft Copilot, Perplexity, and Claude by agreement. Each plan includes the engines that business needs. Answyn tracks the United Kingdom in English by default, with other markets and native languages set per plan. The standard engines are collected daily rather than hourly, and Answyn is not SOC 2 certified. If your old tool ran hourly, Answyn cannot keep parity on cadence, and an engine or market beyond the defaults keeps parity only if the plan includes it from the first overlap week.

Who should not switch

Some teams are better off where they are, and the published facts say which.

Rankscale is inexpensive and flexible. Essentials is $17 a month, billed yearly; scheduling runs from hourly to monthly on every plan; regions are listed as All; its pricing page counts 17+ engines; and Scout recommendations come with every plan. It also publishes a formula for its Visibility Score, which several platforms in the tables above do not. If price, an hourly cadence or a wide roster is why you chose it, a daily tool with a shorter roster trades those away.

Brandlight covers a wide roster. Its product page names twelve engines, Naver among them, and runs every question against every engine weekly, with each brand and market in its own instance. Its blog (30 September 2026) describes impact tracking per URL, down to the query, and says it is SOC 2 Type II and GDPR compliant, with SSO available. If you report on several markets, need Korean coverage or face a security review that requires SOC 2, weigh that before you move.

And if the real reason for leaving is that nobody acts on the reports, a new dashboard will not change it. Someone still has to turn each gap into a page, a correction or a placement on another site.

Where another platform is the better buy

If Perplexity or Claude must stay in your set at a published price, Otterly.AI includes Perplexity and Microsoft Copilot on its Lite, Standard and Premium plans, with Claude as a paid add-on, and Scrunch lists Claude and Perplexity on its pricing page.

If you need several countries on a published plan, Peec AI's Pro and Advanced plans cover three each, and every Peec plan includes CSV exports. If you need all-time history with CSV and JSON exports, Profound lists both on Enterprise, at a custom price.

If you already pay for Semrush, its AI Visibility Toolkit (from £80.22 a month per domain, billed annually) may be the simpler move. Note that its Average Position describes where a citation of your domain appears, which is a different measure from where your brand is named.

If Answyn is on your list, access is by explicit invitation and there is no list price: each plan is quoted for the business using it. Start with a 30-minute call about running it beside your current tool on the same questions for the overlap weeks, and see how Answyn pricing works for what sets the price.

Questions people ask

Can I compare citation share between two AI visibility tools?
Not as absolute figures. Profound's Citation Share is your share of all citations, while Otterly.AI's Domain Coverage is the share of prompts that cite your domain at all, so the same answers produce different numbers. Compare each tool's trend over the same weeks, on the same questions, engines and market.
What should I export before I cancel Rankscale or Brandlight?
The answers with their citations, the question list, the competitor list with its aliases, and the full history. Rankscale's help article says an export can include every execution, with answer text where the engine allows it, citations and sentiment, and an All option for complete history. Brandlight's product page says customers export in CSV or JSON at contract end, with deletion within thirty days.
Why did our share of voice change when we switched tools?
Usually because the counting changed. Some tools count every mention and others count each brand once per answer; Ahrefs' help article weights share of voice by search-volume impressions; and a tool that does not merge spellings splits one rival into several. Check those three before you read anything into the change.
How long should we run the old and new tools side by side?
At least four weeks, on the same questions, engines and market, so both tools cover a full monthly window. Note the gap between the two baselines, then report on the new tool's own trend from the day the old one stops.
What is the difference between prominence and position?
Position is a rank: 1 means named first among the brands in an answer. Prominence describes how clearly a brand is named, which tools record as buckets, such as Answyn's first, among others or only in passing, or Brandlight's categories from Lead source to Buried mention. Ask a new tool which one its dashboard reports.
Can Answyn import our history from another tool?
Plan on Answyn's history starting with its own first collection, because every figure opens back to answers Answyn collected itself. Your question list can come across by CSV import, which is why the parallel weeks matter.

Sources

Written by Answyn, which sells one of the platforms on this page. Every vendor definition, plan detail and export term was read on the vendor's own pages on 4 November 2026 and is linked under Sources. Vendors revise their definitions; if one has moved, tell us at team@answyn.com and we will correct it and date the correction.

See the answers for your own brand.

Answyn puts the questions your buyers actually ask to every major engine, daily, and records who gets named and who gets cited.