Interpret AI visibility trends
How do I interpret changes in AI visibility over time?
Potential customers may ask ChatGPT, Gemini, Perplexity or another AI assistant which local business to choose. Trends shows whether your business appears in those answers more often over time, and where it is still missed. Open AEO, choose a location and select Trends.
Work through the evidence in this order: first, what changed in the overall AI Visibility score. Second, which answer engine is strongest and which one still trails. Third, which prompt and citation evidence should guide the next improvement.
A strong overall gain does not mean every engine improved equally. Read the overall line first, compare the engines over the same months, then use the supporting evidence to choose where to focus.
Key facts
| Where it lives | AEO → Trends, for one location |
| One point means | One calendar month with a report. The chart draws the last twelve reports |
| Needs | At least two months with a report. With one, the tab shows the latest values and a note |
| Competitor lines | Up to the three highest-scoring competitors, in grey, with their names in the legend |
| More frequent updates | A Weekly schedule or a manual refresh inside a month replaces that month's report. It does not add a point |
| Spacing | Months are spaced evenly by report, not by calendar |
Steps
Open AEO, choose the client and location, and select Trends.
Read Visibility over time: your line against up to three competitors.
Read Per-engine trend and compare the engines over the same months.
Return to Overview and check the prompts behind any engine that trails.
Open Citations for the source evidence, and check Next update in the refresh clock before you expect another point.
Start with the overall direction
This location has three monthly reports. Overall AI Visibility moves from 21 in September to 43 in October and 49 in November. That is clear progress, but the grey lines show that the leading competitor is still ahead at 64. The change shown on the AI Visibility card on Overview, +6 here, compares the latest report with the previous stored report, not with a calendar month a fixed distance back.
Use the chart to establish direction and scale. Then compare the engines before deciding the work is done.
Compare the answer engines
The same three months look different by engine:
- ChatGPT improves from
27to52to61. - Perplexity improves from
24to46to50. - Gemini improves from
11to32to35.
All three engines improve, but Gemini trails ChatGPT by 26 points and has almost stopped moving between October and November. That makes Gemini the clearest remaining opportunity, not because the overall result is poor, but because the engine comparison shows where coverage is still available.
Confirm the gap with prompt coverage
Return to Overview and look at the Gemini column of Prompt coverage. In this report ChatGPT names the business on 12 of 16 prompts and Gemini on 8. Gemini misses, among others:
top rated family dentist near Lakeview, Chicagowhere should I go for teeth whitening in Lakeview, Chicagodentist open now near Lakeview, Chicagodentist with Saturday hours in Lakeview, Chicago
Those misses isolate recognisable customer needs: reputation, a specific service and availability. Start with those rows, because they turn a broad engine-level difference into an exact investigation. How do I read the prompt coverage grid? shows how to open a cell and read the answer.
The questions are chosen when the first report is generated and are reused on every later update, so each month is measured on the same prompts and the rows are comparable from one month to the next.
Use citation evidence as context
The Sources cited card on Overview rises from 17 to 20 to 23 over the same months. Open Citations to review the current source coverage and the missing directories alongside the Gemini gaps. That card is a coverage total, not a count of distinct sources; see Which sources do AI engines cite.
The increase is useful context, but it does not prove that citations caused the visibility gain. Treat prompts and citations as evidence to inspect before deciding whether content, listings or another factor needs attention.
How often a point is added
Each report month is one point. How many points you get depends on the location's schedule:
- Monthly (the default) adds one point a month.
- Weekly, or a manual Refresh… in the refresh clock, updates the latest month's report. The trend does not gain a point, because a same-month refresh replaces that month's report.
- Quarterly adds a point every three months, and Never adds none until someone refreshes.
The chart spaces the months that have a report evenly, not by calendar, so a January, April and July report look like three equal steps. Months with no report are not drawn, so a gap is not a zero. To see when the next point is due, check Next update in the refresh clock beside the page title, or choose Manage schedule. How AEO reports are made explains the options.
If this is the first monthly report
With only one stored point, Synup does not draw a trend. Each chart shows Latest report and a note that this is your first monthly data point. The note ends by saying a new one is added automatically on the 1st of each month. That sentence is generic: it describes the Monthly default, so on a Weekly, Quarterly or Never schedule the next point arrives as described above. The Automatic updates strip above the tabs, or the refresh clock, shows your own schedule and next update date.
See example previews the chart's shape. It is labelled Example · projected, not your data. Use it only to understand the layout. It is not a forecast or expected performance.
What you'll see when it's done
- Your overall line over the last twelve report months, with up to three competitors for context
- Each engine's line, so a trailing engine stands out
- The prompts and sources to open next, and the date the next point is due
Watch out for
- A point is a month, not an update. Refreshing more often than monthly does not add points.
- Quarterly and Never mean fewer points. A flat or sparse line may be the schedule, not a stall.
- The spacing is by report, not by date. Check the month labels before reading the slope.
- Competitor lines can change. Only the highest-scoring competitors for the latest month are drawn, so the legend can change when the report does.
- Sources cited is a coverage total. Do not read it as a count of pages.
Questions people ask
Why did a refresh not add a point?
A refresh in the same calendar month replaces that month's report. A new point appears only when a report is completed in a month that does not have one yet.
Why is there a gap in the months?
The chart only draws months that have a report. A Quarterly or Never schedule, or a month when no update completed, leaves no point for that month.
Why does the chart say a new point is added on the 1st of each month?
That note is the same for every location and describes the Monthly default. Your own schedule is in the refresh clock and the Automatic updates strip.
Where to go next
- What does my AI Visibility score actually measure? — understand what each point counts
- How do I compare AI visibility across all my locations? — read the same signal across a portfolio