Compare screens
Compare sets your brand beside itself: the same read once for each language, place, or platform, with every reading next to the others. It answers the questions a single report cannot: how do I do in German against English, in France against Germany, on one assistant against another?
There are two, one in each channel:
- AI comparison, in the AI visibility group, reads assistant answers: visibility, , position, and sentiment.
- Search comparison, in the Search group, reads the results pages: click share, page one, top 3, and average position. See How search rankings are measured for what those mean.
Both screens are about your own brand, so there is no brand filter. The competitors are one click away: open any slice and you land on the full report for it.
Choosing what to compare
The tabs under the title choose the axis. Each thing on the axis is a slice.
| Tab | Each slice is |
|---|---|
| Languages | One language, across every place you ask in it |
| Locations | One country or city, in every language you ask there |
| Assistants | One AI platform (AI comparison only) |
| Search engines | One search engine (Search comparison only) |
A tab appears only when it has at least two slices. A workspace with one region has no Languages or Locations tab; add a second region in Settings to get one, or turn on another platform to compare assistants or engines. When there is nothing with two slices, the screen says so instead of showing a comparison of one.
The filters at the top still apply to every slice. The date range narrows them all, and narrowing language, place, or platform in the filters removes the slices it leaves out. Search comparison reads the same queries as Search rankings: non-branded by default.
At a glance
One row per slice, with your headline numbers in each.
- AI comparison shows visibility (with its movement since the previous period), share of voice, position, sentiment, your rank among the brands you track, your strongest rival there, and how many responses the slice holds.
- Search comparison shows click share, how many queries put you on page one and in the top 3, your average position, your strongest rival there, and how many queries the slice covers.
Click a row to open Visibility or Search rankings narrowed to that slice, with every competitor back in view.
Over time
A chart with one line per slice, for the measure you pick above it. Beside it the slices are ranked on that measure, headed by the Gap: the distance between your best slice and your worst, and which one leads. The gap is the number a comparison exists to find.
On AI comparison you can also change how the days are grouped and how far back the chart reaches.
Topics by language, location, or platform
Topics run down the side, slices run across the top, and each cell is your reading on that topic in that slice.
| Screen | Measures |
|---|---|
| AI comparison | Visibility, share of voice, or sentiment |
| Search comparison | The share of that topic's queries where you reach page one, or the top 3 |
The Gap column is the distance between your best and worst slice on that topic. The grid opens sorted on it, widest first, because the topics where your slices disagree most are usually the ones worth reading. Every heading sorts: the topic, each slice, the gap, and the count.
Three cell states are deliberately distinct:
| Cell | Meaning |
|---|---|
| A shaded number | Your reading on that topic in that slice |
A plain 0% | The topic was asked there, and you never appeared |
A hatched cell reading n/a | Nothing on that topic was asked in that slice |
Shading works as it does on the heatmap: percentages shade relative to the strongest cell on screen, and the number is always printed, so colour is only a scanning aid. Sentiment is a score rather than a share, so it keeps a fixed scale.
Click a cell to open the responses or search results behind it, narrowed to that topic and that slice.
Reading it well
Compare topics before you compare headlines. Each language and each place asks its own . When you add a region, Silktide writes your questions again for that language and place, and you may have chosen only some of them. So two languages' headline numbers can answer different question sets, and a gap between them may say more about the questions than about you. A is shared by every region, which makes the topic grid the like-for-like reading. Assistants and search engines are the exception: every platform is asked the same questions, so those headlines compare directly.
Check the counts. The Responses and Queries columns tell you how much each slice rests on. A region added this week has few measurements yet, so its trend line is short and its figures move more until a few batches have run.
Look for the pattern, not the cell. One topic trailing in one language is noise more often than not. The same slice trailing across most topics is a finding.
Related
- Visibility screen - the AI reading for one slice, with every competitor
- Search rankings screen - the search reading for one slice
- Regions screen - adding a language or place to compare
- Heatmap - how to read the grid and its shading
- Metrics at a glance - what every number means
- Measuring visibility - cadence, assistants, and sample size