Product updates
What we changed, and why. Newest first.
What lands on this page
Every substantive change to tobecited is written down here: a new check in the audit, a change in how a score is calculated, a new fix we generate for you, a change to plans or limits. Bug fixes and interface polish are mentioned only when you would otherwise notice something behaving differently.
The reason this page exists is that we score websites, and a scoring method that quietly changes is worthless. If a number in your report moves without your site moving, the entry that moved it is here, dated. Where a change affects existing reports, the entry says so instead of leaving you to compare screenshots.
Entries are newest first, each one anchored to its date so you can link to a single change. We publish as changes ship — usually a few times a month — rather than batching them into a release note nobody reads.
Fixes are checked against what you already have
- Every fix is compared with your live site first and can say “yours is already fine” instead of handing you a snippet that replaces working content.
- What we generate builds on what you have done yourself: nothing you already publish is overwritten or dropped.
Two schedules, one place to set them
- Weekly check and monthly full audit are now separate schedules, each with its own day and hour, both in account settings.
- Projects got a Run tab — questions, pages, steps, reports. Visibility stays a read-only mirror: nothing edited, nothing spent.
- Before a run starts, the screen lists the steps it will take and the ones it skips.
We ask your server, not your robots.txt
- Crawler access is now measured, not predicted. A site that welcomes AI crawlers on paper while turning them away in practice no longer passes as open — and where the evidence is ambiguous, the report says so instead of guessing.
- For thin pages the report shows two pictures: what a crawler receives and what a person sees, with the difference spelled out.
- The readiness score is now mostly arithmetic over observable facts, with a breakdown that marks what was counted and what was judged.
- Every report carries a methodology version, and the trend chart compares only runs measured the same way.
The report shows instead of summarising
- New section: the passages an assistant could quote — the weakest blocks of your page, verbatim, with the reason and what to do about it. “Nothing to rewrite” is a valid outcome.
- Answers now come with sources — the domains assistants leaned on — and verbatim excerpts around the moment a brand is named.
- “Mentioned” no longer means “recommended”: mentions are read for tone.
- Brand presence looks past Wikipedia, and what is only declared is kept apart from what we confirmed.
- If an AI engine is unavailable, the run says so instead of quietly measuring less.
Weekly monitoring became a screen you want to open
- The Overview tab is now Visibility: a week summary, the trend chart, and a per-engine breakdown in one place.
- Every run now stores a deterministic diff of what changed since the previous one — no model call, no extra cost. It feeds the weekly email, the report, and the run list.
- Monitoring emails collapsed into one email per weekly check. Its tone follows the week: an alert when visibility moved, a calm digest when it did not. Two independent switches in settings decide whether it is sent.
- Weekly checks now cover every market on your plan, not just the primary one.
- Projects keep a floor of three buyer questions. Below that a measurement stops being a measurement, so the monitor pauses instead of quietly picking new questions and breaking your history.
- The visibility-score line got its own validated colour: in dark mode it used to read almost grey.
Public report links and CSV export
- Any finished report can be shared as a read-only link, and revoked just as easily.
- Full comparison data exports to CSV in long format, ready for a pivot table.