Changelog
What changed in DiscoveredBy, by month.
Gemini API retired
Gemini is now collected from the app only: the answer gemini.google.com shows a person who is not signed in, fetched through a licensed data provider. The separate Gemini API engine was removed, and its stored answers were deleted. Brand studies now ask Gemini, with the study language line added after the question.
Read the docs
Plain engine names
Engines now go by their product names: ChatGPT (app) is now ChatGPT, Gemini (app) is Gemini, and Gemini (API) is Gemini API, the one name that keeps a marker because DiscoveredBy runs two Gemini engines. How each engine is collected has not changed.
Read the docsWeekly report: read through the Explorer
The weekly report now reads its numbers about AI answers with the Explorer's own definitions. Its tiles are share of voice, position, citation rate and citation share beside brand visibility, each engine and competitor row shows brand visibility, and the prompts that moved are the ones whose visibility changed most. Older reports keep their figures, named for what they counted.
Read the docsSite audit: a full SEO crawl of your site
Site audit now runs on a full technical SEO crawler: it checks every page it finds against 238 rules in 17 sections and shows a health score beside its measurement coverage, issue priorities that owners and editors can record decisions on, a site report and a report for every URL. Stopping an audit keeps the pages already fetched, and the page issue queue counts the rules that flag each page in your newest completed audit. It no longer runs the saved page diagnostics checks on every page, and results from the earlier version are not carried over.
Read the docsProject setup: a homepage check and your project limit
Typing your site's address during setup now reads your homepage once and shows its title, its description and where it redirects to, as a check rather than a gate. At your plan's project limit, setup says so before you type anything. If you already have a project, a Back to the app link leaves setup, and finishing opens Overview with a banner saying when your first results arrive.
Read the docsConversion value in your GA4 property's currency
AI traffic, Site traffic and the weekly report show conversion value in the currency your GA4 property reports, read by the daily import. Until it is known, or when two mapped properties report different currencies, a value shows as a plain number rather than being assumed to be dollars.
Read the docsTeam seats on Billing
Settings → Billing shows a Team seats meter against your plan's cap, counted on your fullest project, since the cap applies to each project separately.
Read the docsAn article's page: publish it, fill the gaps, track it
Each article's page now leads with the draft as it will read on your site. Beside it, Publish it lists what to check first and the address, page title, meta description and structured data to copy, plus a WordPress draft for an article; Fill the gaps lists what the brief wanted that nothing you have published supports; Track it takes the live URL and says whether anything measures the page; and What it read shows every claim check.
Read the docsArticles: choose what kind of page the writer drafts
The writer decides whether a gap calls for a new article, a comparison page, a product or service page or an update to one of your pages, and a "Write as" choice on the "Write your own topic" form and on Rewrite lets you overrule it. A product or service page or a page update is written as copy for your own site, not as an article.
Read the docsSearch plan: Google and AI answers, topic by topic
A new Plan tab under Prompts groups the Google searches that found your site and the prompts you track into topics, shows Google and AI answer numbers for each topic side by side, and names the next step, such as tracking a prompt or writing a page. It replaces the Opportunities tab, is rebuilt weekly or on demand, and shows how often each topic's head term is searched on Google a month.
Read the docsGoogle properties: suggested, then mapped by you
Connecting a Google account no longer maps anything to a project. Integrations instead suggests the project a Search Console or Analytics property belongs to, when its site matches the domain of exactly one project you can edit, with a Map to button beside the dropdown. A mapping belongs to the project, so every member sees its data whether or not they connected Google themselves.
Read the docsPick which prompts to classify, then review the changes
The Classify tab under Prompts groups your prompts as not, partly or fully classified, so you can filter them, tick a batch and ask for AI suggestions. The review lists suggested changes first, outlines each value that differs from the current one, and saves only the prompts you keep ticked.
Read the docsSuggestions: a review queue, theme by theme
Prompt suggestions are listed under their themes, with a search box and theme, buying stage and intent filters, and each row opens to show why it was suggested. Tick rows, or a whole theme, to track or dismiss several at once from a bar that says how many prompt slots the selection uses at most.
Read the docsSuggested prompts: more Search Console queries, in a buyer's words
The research run behind suggested prompts now reads up to 25 of your most-shown Search Console queries (it was 15), wherever your site ranks for them rather than only where it already ranked well, and is told to write suggestions the way buyers search: short searches, plain questions and a few casual sentences, naming a place only where a buyer would.
Read the docsA prompt's page: are you in the answer, who is instead, what to do
A prompt's page now opens on a sentence saying whether its answers include you and, when others lead, which companies they recommend most, then how many answers cited your site and named your brand, split by engine with each engine's latest answer a click away. Cards for your page, gap analysis and companies you do not track yet follow, and each tab has its own address, so a link can open it.
Read the docsRun gap analysis for one prompt, on more plans
A prompt's page can now run gap analysis for that prompt alone: it compares the pages AI engines cite for it with yours and, when it finds nothing, says why. On-demand gap analysis, for one prompt or the whole project, is now included in more plans.
Read the docsEvery reference in an answer, products and businesses included
View answer on a prompt's Runs tab now lists all of an answer's references, each labelled Web source, Shopping or Local business, with a filter by type. For Gemini and ChatGPT that includes product and business cards with the names, prices and addresses the provider supplies; a missing brand or merchant reads Not supplied rather than being guessed. Shopping and local business entries do not count as web citations.
Read the docsDraft a project description and goal from your website
In project settings, owners can choose Generate from website: it reads your homepage and up to four relevant pages and drafts a description and a suggested business goal, with any caveats and the pages it read. Nothing is saved until you use a draft and save, and when the site does not give enough evidence it says so instead of guessing.
Read the docsGrowth Advisor: Brief me weekly
The project owner can turn on Brief me weekly beside New briefing, and a briefing is then written every Monday, skipped when one is running or was created in the last 7 days.
Read the docs
AI crawlers: which AI bots fetch your pages
Send your server or CDN logs through a webhook, a Cloudflare Worker, Cloudflare Logpush or a log file upload, and see which AI crawlers requested which pages, whether each request is verified against the addresses its operator publishes, the errors your site served them, and which fetched pages AI answers cite. Only bot requests are kept, and client IP addresses are never stored.
Read the docsTo do: one ranked list of what to do next
A new tab under Actions gathers open work from seven sources, from page fixes and citation gaps to alerts and fact check findings, into one list banded High, Medium or Low by a stated rule per kind. Each button changes the item on its own page, and the To do tab shows how many High-priority items are open.
Read the docsChatGPT replaces ChatGPT API
ChatGPT is now collected from the app only: the answer chatgpt.com shows a person who is not signed in, fetched through a licensed data provider. The separate ChatGPT API engine was removed, and its stored answers were deleted.
Read the docsGemini: what gemini.google.com shows
A new engine: the answer gemini.google.com shows a person who is not signed in, collected through a licensed data provider. It runs beside Gemini API, and the two are separate engines in every filter and breakdown.
Read the docsAnswer features: shopping, local, ads and more inside answers
See which engines put shopping, local businesses, ads, video, images or tables inside their answers, and which answers used web search, each as a rate per engine.
Read the docsGoogle AI Overviews and Google AI Mode
Two new engines: the AI Overview on Google's results page for your prompt, and Google AI Mode's answer. A search where Google shows no AI answer counts toward a shown rate, not toward any answer-based number.
Read the docsPrompt coverage: which prompts you could pause
Find running prompts you could pause while the sources and brands they regularly see are still seen by a prompt you keep, with protections for prompts other work depends on, a preview of any selection, and pausing from the same screen.
Read the docsFact check: where AI gets your facts wrong
Keep a list of approved facts about your brand. The answers of a weekly fact study and of up to ten tracked prompts you choose are checked against them, and each contradiction is quoted, linked to its sources and queued for your team to review.
Read the docsAttributes: what each brand is known for
A weekly study asks the chat engines on your plan, and ChatGPT, what your brand and each active tracked competitor are best known for, then which brands they name for each chosen quality, showing association and market prominence side by side with the quotes behind them.
Read the docsA new Overview
Overview now shows how your brand stands against the competitors you track: a headline, five headline numbers, a trend, a brand ranking and one answer in full, for the range and filters set in the filter bar.
Read the docsObjections: why buyers might not choose you
A weekly study asks the chat engines on your plan, and ChatGPT, directly why a buyer might not choose your brand and each active tracked competitor, then ranks the objections by prominence with the quotes behind them and shows what changed since the last comparable study.
Read the docsPerplexity answers through Perplexity's Sonar API
Perplexity now writes a generated answer through its Sonar API. Only a source one of its numbered markers points at becomes a citation, and every source it returns is recorded as a returned search result.
Read the docsAnalyst: ask your data in plain words
Ask a question about your tracked AI answers in plain words and get a short answer built from Explorer queries, with every number filled in from the results shown under it.
Read the docsLocal: city answers against the whole country
Prompts can now be tracked in a city. The Local page compares those answers with the same country country-wide: brand visibility, citation rate, share of voice and position, the gap for each city, and the prompts where a city differs most.
Read the docsExplorer: build your own chart
Pick metrics, break them down, filter them and choose a window, a comparison and optional smoothing, with the population rules and limits behind every number.
Read the docsLanguages, engine choice and templates for prompts
Track a prompt in one or more languages; chat engines get the words you typed unchanged, plus an instruction to answer in that language. Choose which engines run, preview the new prompt slots and monthly runs before you save, and save a combination as a reusable template.
Read the docsPersonas: run a prompt as a named buyer
Create named buyer profiles and run a tracked prompt as one, alongside or instead of the General audience every prompt already runs as.
Read the docsCompare vs. competitor
A controlled comparison against one active tracked competitor over only the answers that name both of you: coverage, named-first, sentiment and reasons.
Read the docsBrand-signal alerts, and team reactions to any alert
The daily watch now also flags brand share shifts, emerging brands, sentiment shifts and competitor-only social citations. Owners and editors can acknowledge, dismiss or snooze any alert with a shared team note.
Read the docsWordPress drafts
Send a finished article into a connected WordPress site as a new draft post. DiscoveredBy never publishes, edits or deletes a post there.
Read the docsBrand reasons: why AI favours or cautions against a brand
See why AI answers favour or caution against your brand and each tracked competitor, each reason counted as a strength or a weakness with the quote behind it.
Read the docsBrands and sub-brands
Add aliases, and sub-brands under your own brand and active tracked competitors. A sub-brand rolls up into its brand family, and a family counts once per answer.
Read the docsSite audit: crawl your site as a bot
Crawl your own site as DiscoveredByBot from robots.txt, sitemaps and internal links, and compare the raw HTML of each page with a browser render.
Read the docsRetrieved vs cited
See which sources each AI engine returned while answering your prompts and which ones the answer text actually attributed, one collection channel at a time.
Read the docsSource-mention history
Save the brand evidence found in cited third-party pages today, and compare it with a later observation for gains and losses.
Read the docsSaved page diagnostics
Run and save policy and HTML checks for a selected page on your site, then recheck after your edits and compare with the saved history.
Read the docsPage issue queue
Prioritize your own pages in one queue using saved gaps, draft status, diagnostic findings, linked evidence and recorded outcomes.
Read the docsAlert email preferences for project owners
A project owner chooses daily, weekly or no alert emails, and which alert categories to receive.
Read the docsInvite teammates by link
Owners can invite someone by link with an owner, editor or viewer role before that person has an account. The link expires after seven days; the recipient signs in with the invited address and accepts.
Read the docsSign in with Google
Continue with Google, including a personal Gmail account, or use a one-time email link. Signing in with Google on an address that already has an account joins it to that account.
Read the docsClient report links and templates
Share a fixed client digest of a week by link, with no sign-in needed, and control access with expiry and revocation. Save report templates, optionally with citation examples.
Read the docsClient reports across projects
Compare saved client readings for one completed week, and share reviewed, revocable combined agency snapshots.
Read the docsSource comparison and watchlists
Compare domain citation rates across equal periods and read owned-source share. Save domain or exact-URL watchlists and subscribe to observed citation gains and losses.
Read the docsEarned sources: an outreach pipeline
Save cited third-party pages or add one by hand, assign outreach work, report placements and compare later citations of the page. It sends no messages and finds no contacts.
Read the docsHand off page fixes
Download a fix as a Markdown or PDF task packet for whoever implements it, and approve up to 25 open drafts for handoff at once.
Read the docsSentiment trends
Compare sentiment across your brand and tracked competitors by prompt segment and collection channel, review negative or warned-against brand evidence, and inspect citation framing separately.
Read the docsMCP connector
Connect a compatible assistant to one project with read-only tools for its prompts, answers, citations, daily metrics, weekly digests and more, using the same keys as the customer API.
Read the docsCustomer API and keys
Create scoped, expiring read keys for a project's prompts, collected answers, citations, daily metrics, saved weekly report digests and more.
Read the docsSource types
Organize cited pages with source categories, page formats, rule evidence and project-specific overrides.
Read the docsRobots diagnostics
Check the robots.txt rules declared for five AI crawlers on a URL of your site, with the rules that matched.
Read the docsSource reach, consistency and brand mentions in cited pages
Domains shows how far each source reaches across your prompts and how consistently. A Mentions view checks your brand and active tracked competitors against the saved text of third-party pages AI engines cite.
Read the docsGoogle demand for each prompt
With Search Console connected, each prompt shows the impressions Google recorded in the last 28 days for the searches it covers. It counts Google searches, not questions put to AI assistants.
Read the docsOrganize and classify prompts
Give prompts an intent, buyer stage, theme and branding, set them together on a selection, and ask for AI classification suggestions to review.
Read the docsImport prompts from CSV or Excel
Import a whole list of prompts at once from a CSV file or the first worksheet of an Excel file, with columns matched by their headers.
Read the docsJSON exports and an Answers export
Every dataset now downloads as JSON as well as CSV, with the same rows in the same order. A new Answers dataset carries the response text of each completed answer.
Read the docsWeekly article suggestions on the day you choose
With weekly suggestions switched on, Articles generates that week's topics on the day you pick, at most once a week per project.
Read the docs
Fan-out: what each engine searched for
See what each engine actually searched for behind your prompt, where the engine reports it, and the eight categories those searches fall into.
Read the docsCSV exports
Download prompts, daily metrics, brand mentions, citations, cited URLs and domains as CSV files. A download over its row cap is refused rather than cut short.
Read the docsEvery answer records how it was collected
Each answer stamps its platform, surface and collection method when it runs, so an older answer keeps saying how it was actually collected even after an engine changes.
Read the docs
Brand visibility counts answers that name you
Brand visibility is now the share of analysed answers that name your brand, beside separate numbers for brand position, share of voice, domain coverage and citation rate. A number with no data reads as no data, never zero.
Read the docsWeekly report
A dated report for each week, built from your own numbers with no model in the loop, readable on screen, by email and on paper.
Read the docsllms.txt Advisor
A weekly audit of your live llms.txt against the spec, with a compliant file written out for you to publish.
Read the docsGrowth Advisor
A weekly written briefing that reads your search and traffic numbers alongside outside research, and says what to do next.
Read the docsA business goal for each project
Write down what the business is trying to achieve, in plain English. Agents such as Growth Advisor read it as steering context, not as an instruction to follow literally.
Read the docsSearch performance and Traffic
Search Console and Google Analytics get separate screens. Search performance adds quick wins and movers; Traffic groups sessions by where they came from, with AI counted first, and lists pages where attention is not converting.
Read the docsSuggested prompts from research, with auto refill
Suggested prompts come from a research run that reads your site, with your best Search Console queries as extra context, and the queue tops itself up once it runs low.
Read the docsOptimizations: page fixes grounded in cited pages
For a gap with a page of yours behind it, DiscoveredBy reads the pages AI engines cite instead, works out what yours is missing and writes a draft, keeping a change only when one of its evidence quotes appears in the source page.
Read the docsCitation gaps
Prompts where a competitor is cited and you are not become a ranked queue of specific changes, based on what the cited page has that yours does not.
Read the docsAn Alerts page, with why each alert fired
Alerts from the daily watch get a list of their own, and each one opens to show why it fired: the before and after figures, saved when the alert was detected.
Read the docsGoogle properties mapped to projects automatically
In the ordinary case, connecting a Google account became the only step: a Search Console or Analytics property whose hostname matched the domain of exactly one project you could edit was mapped to it for you. Automatic mapping was removed in October 2026.
Replaced in October 2026: Google properties: suggested, then mapped by you
Read the docs
Search Console and Google Analytics in the app
Your Search Console queries and pages and your Google Analytics sessions show inside DiscoveredBy. A project holds one Search Console property at a time; mapping another moves the project to it.
Read the docs