Platform How-Tos
How to Read Your Sightline Visibility Report
Published October 3, 2026
Your monthly Sightline report is a record of one month: how the AI engines we test answered a fixed panel of buyer prompts (questions a customer might ask), what changed since your previous report, and what work is done and next. Read it in your client portal or as a downloaded PDF; the numbers match. This walkthrough follows the PDF top to bottom, saying what each part means and whether to act.
What is in the report, and in what order?
Up to six sections, in this order in the PDF: a masthead with your business name and month, your AI visibility score with up to four numbers beside it, a tile grid, a plain-language summary, the work, and category scores near the end. Its printed title reads like "October 2026 Performance Report".
| Section | What it tells you | What to do |
|---|---|---|
| Masthead | Whose report (one per location), which month | Check the business and month |
| Score and up to four numbers | How AI answers treat you | Track the score; tell us your rivals if none are listed |
| Tile grid | Ads, plan progress, site health | A plus on Cost / lead is bad news |
| Summary | The why behind the numbers | Note questions for your review call |
| The work | Done and next, up to five titles each | Bring Next up to your review call |
| Category scores | Rough guides to six areas | Use as a map of possible gaps |
The web report has the same numbers and notes but no PDF masthead or score ring: the score and the numbers beside it are ordinary tiles (Ad spend first for Google Ads clients), and category scores come before the work lists.
What do the score and the numbers beside it mean?
The ring is your AI visibility score, out of 100 and not a percentage, blended from mention rate (40%), share of voice (25%), the tone of mentions (15%), answer accuracy (10%) and how early in answers you appear (10%). Tone and position never print. One report is a data point; several moving the same way are the signal. What a good score looks like covers score ranges.
Beside it, up to four numbers can print:
- Share of voice: your slice of the mentions split between you and the competitors listed for your business; unlisted businesses don't count. "vs. 0 tracked competitors" means nobody was listed, so it reads 100% if you were named at all, 0% if not. Tell your account team who you lose jobs to. Adding them changes the measurement: expect share of voice to drop, with its comparison skipped that month, and your score and mention rate to move, with their changes still shown.
- Mention rate: the share of answered checks (one buyer prompt put to one engine) that named you. Some prompts name you directly, so it is not a rate of unprompted recommendations.
- Answer accuracy: of answers that stated facts about you, the share matching the industry, area and description on your record, not the wider web. Tell your account team when those details change or AI repeats something outdated.
- AI citations: a count, not a list, of distinct outside websites that appeared as sources in answers naming you, excluding your own site. A source may have been retrieved without being used, or inferred from the answer text, so these are not verified citations. It is context, not a target or a trend, and not part of the score.
If fewer than 80% of checks return an answer, a note under the score (under the tiles on the web) reads "Partial panel this month: only N of M AI checks returned an answer." The AI numbers, category scores included, still print, but don't compare them with the month before or after: the report pauses those comparisons too, and nothing needs fixing on your side.
How should you read the tiles and the changes?
Which tiles appear depends on your services and what was measured that month. A change line compares a number with your previous report, in one neutral color with no arrows, so a plus is not always good: a rising Cost / lead looks like rising Leads.
- Ad spend, Leads, Cost / lead: Google Ads totals for the 30 days before our latest data pull, not the calendar month. Leads are the actions your Google Ads account is set up to count (Google calls them conversions).
- Action plan complete (to date): the share of your action plan done so far, not this month's work. Upgrades you haven't bought count too, so don't expect 100%.
- Search health: for clients with search engine optimization (SEO) work, a 0–100 site score built on Google's PageSpeed tests (95 at most), if that month's site scan finished before the numbers were set. A "held" note means the score is capped because the site isn't drawing enough measurable traffic yet.
- Search fixes complete (to date): the same running share for your search fix list.
No line prints when a number didn't change, either report lacks it, or the comparison is paused. Ad spend, AI citations, Action plan complete and the category scores never show one. Your first report has nothing to compare against, though its "vs. last month" label still prints.
What are the summary and the work lists for?
The summary explains the numbers; the work lists show what's done and what's next. The summary ("This month, in plain language" in the PDF, "What happened & what's next" on the web) is the part a person edits by hand: your account team reviews it before publishing.
"Completed recently" reaches back to the start of the previous month, since audits usually run early in the month. "Next up" holds open work, never upgrades you haven't bought. Each shows up to five titles, without instructions or dates. Your live dashboard has the full action plan and, once listed competitors are measured, a competitor chart.
Bring Next up and your questions to your monthly review call, or reply to the report email.
How much weight should the category scores carry?
Treat them as rough guides, not audited facts: the report says they are read from how the tested AI engines answered this month's prompt panel, "not a site crawl." None feeds your score, and none shows a monthly change.
They cover Entity authority (whether AI recognizes and correctly describes you when asked by name), Structured data, Citation footprint (outside sources in answers), E-E-A-T signals (experience, expertise, authoritativeness and trustworthiness), GBP optimization (your Google Business Profile) and Review platforms. Structured data comes from AI answers; it does not inspect the machine-readable labels in your site's code.
Rows marked as estimates (an asterisk and lighter bar in the PDF, a card note on the web) were not independently verified: always Structured data, Citation footprint and E-E-A-T signals; GBP optimization unless that month's audit confirmed your Google listing; Review platforms unless measured from it. When measured, that row becomes Google review standing, the only measured row, with a note naming what was read.
Unmarked does not mean measured: Entity authority is never marked yet comes only from AI answers, and an unmarked GBP row is still partly answer-based.
If a note says Google lists you as permanently closed and you are open, correct the listing's status. After AI suggests you, many buyers read your reviews before getting in touch; our reviews guide covers what they look for.
The free Visibility Check is a one-time snapshot, not the monthly audit, so don't compare the two scores. Not a client yet? Every retainer includes the monthly report; book a demo to talk it through.
Questions people ask
Not while it is published. A published report's numbers, comparisons and caveat notes are frozen and are not recalculated. Your account team can still correct the title or the summary: the web report shows the change, and the PDF on the report page is normally re-made to match, without a new email. A copy you already downloaded does not change.
Only in parts. A report is labeled with the month its audit ran, and the monthly audit starts on the 1st, so the October report usually reflects AI answers gathered in the first days of October (your first report uses the audit run right after you joined, which can fall later in the month). Some parts reach back: Completed recently includes the previous month, and the Google Ads tiles cover the 30 days before our latest data pull, which early in a month is mostly the month before.
Usually because it wasn't available or reliable when the numbers were set. Answer accuracy is left out when no answer named you or nothing could be judged. Search health appears only if that month's site scan finished before the numbers were set, and can be omitted when the scan was incomplete or hit a serious problem, such as a site blocked from search engines; the Google Ads tiles need a recent data pull. A missing tile is not a zero, so ask your account team if you expected one.
One address gets it: your account's main contact. The email links to the report in your client portal, which signs you in with a one-time link instead of a password, and it carries no attachment. To share the report, download the PDF from the report page (its "Download full PDF report" button) and send the file itself, since each download link is short-lived.