Google expands the Search Analytics API with branded and visual search data
Use three example requests to pull branded queries, visual search performance, and country and device cohorts.
Google brought three Search Console reporting features to the Search Analytics API on November 16. SEO teams can now use Google’s branded-query classification, pull web multimodal performance, and select several countries or devices in one request. The practical change is that these segments can become repeatable dashboard inputs instead of recurring manual exports. Google’s announcement
Start with a baseline you can reconcile
Take a completed seven-day period and save the date range, property, search type, and dimensions with every export. Run a plain Web request first. This gives you a control for checking access and date coverage before you add a filter. For visual search, use the same property and dates and keep the results in a separate series labeled Web multimodal.
Use the Search Console API’s read-only OAuth scope and a property your account can access. Submit each JSON body below to the Search Analytics query endpoint for that property. A Domain property such as sc-domain:example.com and a URL-prefix property such as https://www.example.com/ are separate inputs. Use the exact property identifier, encoded in the path. The example dates are fixed so you can reproduce the request structure. These are documented request examples. They were not executed against an authenticated property. Search Analytics query reference
Three requests to try
Group Web results by Google’s branded classification. Use the resulting labels to compare the two segments. A classification request cannot also use a query filter, so keep a legacy keyword-regex report in a separate request while you check how the two approaches differ.
{
"startDate": "2026-11-02",
"endDate": "2026-11-08",
"type": "web",
"dimensions": [
"brandedClassification"
]
}
Pull visual search results by landing page. Google’s announcement includes Lens, Circle to Search on Android, and image uploads in Google Search. Use the landing-page cut to see which URLs receive visual-search impressions and clicks. The type groups multiple experiences, so the returned rows should be labeled with the combined reporting category.
{
"startDate": "2026-11-02",
"endDate": "2026-11-08",
"type": "webMultimodal",
"dimensions": [
"page"
],
"rowLimit": 1000
}
Select two countries and two device types while preserving the breakdown. Within a list, the values are alternatives. Across the two dimensions, the conditions both apply. With country and device included as dimensions, a dashboard can retain the individual rows while using one request for the regional cohort.
{
"startDate": "2026-11-02",
"endDate": "2026-11-08",
"type": "web",
"dimensions": [
"country",
"device"
],
"dimensionFilterGroups": [
{
"groupType": "and",
"filterLists": [
{
"dimension": "country",
"expressions": [
"USA",
"CAN"
]
},
{
"dimension": "device",
"expressions": [
"MOBILE",
"DESKTOP"
]
}
]
}
]
}
Give an empty response its own status
If your baseline returns data and the branded request has no rows, check property eligibility and the date range. Google limits branded classification to eligible top-level properties with sufficient volume. Update the dashboard to distinguish a successful request with no returned rows from a measured zero. Use the same approach for an empty multimodal result and confirm the type and dates in the Search Console interface before diagnosing a traffic loss.
Capture row limits and aggregation settings in the export. A ranked page extract can omit rows beyond its limit, and a page-level export and property total may use different aggregation. In a worked example, two rows with 2 clicks from 20 impressions and 8 clicks from 400 impressions have a combined CTR of 10 divided by 420, or about 2.38%. Averaging their individual percentages would give 6%.
If your site already has bulk Search Console export enabled, Google also says multimodal data will be available in BigQuery. Check the export before moving a production dashboard and preserve the old series during the transition. For the separate Google AI impression report, see how Search Console counts generative AI impressions. For analytics and server-side attribution, see how Google Search traffic appears in referrers.
Keep learning
Continue this topic
Measure referrals
Set up the GA4 investigation
Check definitions
Keep each visibility event separate
Test the score
Compare visibility with actual visits
SEO
Ask a question or join the discussion