what we
shipped.
A monthly record of what changed in BotScope — new capabilities and improvements to the things you already use. For the thinking behind the bigger releases, read the journal.
- NEW
Research studies: see which brands AI recommends across a whole market
Pick a sector, a country and the towns or regions you care about, and BotScope suggests the services and questions to cover, writes the prompts and runs them across ChatGPT, Google AI Overviews, Gemini and Copilot. You get every brand AI recommends for each service and location, the websites it cites, and an email when the study is ready, with no brand of your own needed.
Read the post - NEW
See which websites AI cites when people research your field
A new Sources tab covers your informational (Level 4) prompts. For each topic or tag group it shows how often answers cite you and which sites they cite instead (government pages, directories, news, forums or competitors), with every cited page a click away.
- IMPROVED
Studies and Sources in the API and MCP
List research studies and pull their brand rankings and cited sources, or any project's Sources view, through the REST API and the MCP server.
- NEW
Organise prompts into your own groups, such as Location or Service
Create tag groups and sort prompts into them by hand, from a CSV, or from an AI suggestion that proposes the groups for you. Category Landscape can then show every metric per location or per service, and reports, the API and MCP follow the same grouping.
- IMPROVED
Relabel any prompt's category, audience or topic
Edit labels one prompt at a time, in bulk, or from a CSV, and Category Landscape follows the new labels across your past scans as well.
- IMPROVED
Know how complete a scan is while it finishes
Results appear as soon as a scan finishes, marked with how many answers are in so far, and anything a provider didn't return keeps being retried through the rest of the day. The scan email waits for the final numbers, and reports note when they're based on fewer answers than planned.
- IMPROVED
Smarter recognition, sentiment and rankings
Whether an AI really recognises you, how each mention talks about you, and where you rank in its recommendations (including comparison tables and local listings) are now judged in context rather than by matching words. Your history was re-scored so trends still compare like with like.
Read the post - NEW
Repeat a scan to see how much AI answers vary
Schedule up to four runs on each scan day, or queue several back-to-back runs from the Run scan button. Each run is a full, independent scan, so you can see how consistently a prompt mentions you instead of trusting a single answer.
- NEW
See how often each prompt mentions you across runs
Pick a date range and every prompt shows how many of its runs mentioned you, with the median list position on category prompts, plus a mention rate for each layer.
- IMPROVED
Open any scan from a day, not just the latest
Days with several scans list every run with its time and score, and one click shows all of them together as a range.
- IMPROVED
The visibility trend opens on 90 days, or wherever you left it
The Signal trace now spans a quarter by default — enough points for a monthly tracking cadence to read as a trend — and whichever range you choose is remembered per project. The range and chart-type controls sit on the chart itself, so moving between a week and a year is one click.
- IMPROVED
AI Overview citations name the site that was actually cited
Google now routes every AI Overview reference through a link that points back at Google. BotScope resolves those to the publisher underneath before anything is stored, and we repaired the citations already sitting in your history — so your citing-domain data reads correctly backwards as well as forwards.
Read the post - NEW
Track which products AI recommends, not just your brand
Build a catalog of the specific models you care about — yours and your competitors' — and see which ones AI answers actually recommend, on which prompts, with sentiment scored per product instead of per brand. Adding a catalog backfills mentions across every answer you have already collected.
Read the post - NEW
Watch the shopping carousel in ChatGPT answers
The product cards ChatGPT shows on shopping questions are now captured on every scan: which products appear, the price and rating displayed, and which merchants get the click. Prices are kept per scan and per market, so you can see what a buyer was actually shown over time.
- IMPROVED
Products in client reports, the API, and MCP
Client reports, PDFs, and share links carry a Products section, and the same numbers are available through the REST API and the MCP server. Product tracking itself is on every plan, including trials.
- NEW
Choose the market each project is tracked in
Set a tracking country per project and new scans run in that market. The same prompt returns a different set of brands in the UK than it does in the US — now you measure the one your customers are actually in.
Read the post - IMPROVED
Prompt and competitor suggestions match your market
Discovery now proposes prompts and competitors for the country a project tracks, rather than defaulting to a single market. Brand verification is grounded in your live homepage, so suggestions describe your business instead of a similarly-named one.
- NEW
Failed queries are retried automatically
When a provider stops answering mid-scan, BotScope now re-runs the queries it dropped — up to two further attempts, with a widening delay — so a temporary failure is usually recovered before you open the dashboard. Recovered queries bill as normal; nothing else does.
Read the post - NEW
See exactly what a scan missed, and re-run it yourself
Every scan now reports which queries failed and on which provider. If automatic recovery does not get them, you can re-run just those — without paying for a full re-scan.
- NEW
Delete a scan that went wrong
Remove a bad or incomplete scan so it no longer sits in your trend line, charts, or client reports.
- NEW
See who is advertising on your prompts in ChatGPT
Track sponsored placements that appear in ChatGPT answers — which prompts trigger them, which brands are buying, what the creative says, and whether a competitor is paying for answers your brand is absent from.
Read the post - NEW
Segment everything by tag
Tag tracked prompts by topic, audience, or campaign, then filter every chart, report, and API call to just the work you want to measure.
Read the post
- NEW
Read-only REST API and hosted MCP server
Pull visibility scores, share of voice, competitors, and citations into your own stack — or connect the MCP server and ask Claude about your brand directly.
Read the post - NEW
Annotate your charts
Pin events — content changes, prompt edits, provider switches, model updates — to the exact date on your visibility trend, and carry them through to client reports and PDFs.
Read the post - NEW
Free AI visibility scan, no account required
Run a scan on any domain and get a report by email. A quick read on how the models describe your brand before you commit to tracking it.
- IMPROVED
Watch a scan as it runs
Scans now confirm scope before they start and report progress per provider while they run, so a long scan no longer looks like a stalled one.
- IMPROVED
Scan frequency is the on/off switch
Pick a cadence to schedule tracking or set a project to manual to pause it — one control instead of a separate enable checkbox, with a live indicator showing what is actually running.
- IMPROVED
Editing a project keeps the prompts you already track
Changing a project's configuration now reconciles your prompt set incrementally and shows you exactly what will be added or removed before you save, instead of regenerating from scratch.
- IMPROVED
One date and scan picker across every tab
Visibility, sentiment, competitors, and citations all share the same range and scan selection, so switching tabs no longer resets where you were looking.
- IMPROVED
Full support for phones and tablets
The dashboard, charts, and reports now work properly at small screen sizes.
- NEW
Compare any two scans
Every scan is kept as a point in history. Pick two, see what moved, and diff the prompts behind them to understand why.