- Split branded and non-branded performance in every report so brand demand stops masking weak organic growth.
- Pull Search Console data through the API into a warehouse to keep history beyond the interface limits.
- Treat multimodal data as a directional signal for image, visual and voice-style discovery, not a precise forecast.
- Build a monthly review that ties segmented data to specific page, content and brand actions.
For years, one of the most frustrating parts of SEO reporting has been the brand problem. A client's clicks go up, everyone celebrates, and then you discover most of the growth came from people who already knew the company and typed its name. With branded and multimodal data now available through the Search Console API, Google has handed analysts a cleaner way to separate demand you earned from demand you already had. Here is how we think about using it at Growth Asteroid.
Why branded data matters so much
Branded queries behave differently from everything else. They convert well, they rarely depend on content quality, and they rise with PR, advertising, word of mouth and offline activity. Non-branded queries reflect how well your pages answer problems for people who have not chosen you yet.
Mixing the two produces misleading stories:
- A brand campaign makes organic traffic look healthy while category rankings slip.
- A content push wins new non-branded visibility, but the total is flattened by a seasonal dip in brand searches.
- Average position improves simply because branded terms, which rank first, dominate impressions.
Previously, teams maintained regex filters of brand names and misspellings, which were always incomplete. Having Google classify branded traffic natively reduces that maintenance and makes comparisons more consistent across properties.
What multimodal data adds
Search is no longer only typed text. People search with images, camera input, voice and mixed prompts, and results increasingly blend formats. Multimodal reporting gives you a view into how visibility is distributed across those surfaces rather than treating everything as one web search bucket.
Treat it as directional. The useful questions are:
- Which pages earn visibility through image or visual-led discovery?
- Are product and local pages appearing where visual intent is strong?
- Does visual visibility turn into clicks, or only impressions?
Do not build forecasts off small volumes. Use the data to decide where better images, structured data and page layout deserve investment.
Setting up the API pull
The interface is fine for spot checks, but the API is where serious analysis happens. A sensible setup looks like this:
- Authenticate once with a service account added to each property, so pulls run on a schedule.
- Request by date, page and query dimensions in separate calls, since combining everything can hide rows through anonymization.
- Store daily results in a warehouse such as BigQuery so history survives beyond Google's retention window.
- Record the segment as a column (branded, non-branded, and the surface type) rather than as separate tables.
- Log your own annotations: launches, campaigns, migrations, so changes can be explained later.
Expect gaps. Anonymized queries are still excluded from query-level data, so totals by query will not match totals by page. Always reconcile and say so in your reporting.
Reporting that tells the truth
Once segmented, restructure your dashboards around three views.
1. Brand demand trend
Track branded clicks and impressions over time against marketing activity. A decline here is a brand or reputation signal, not an SEO failure, and should be routed to the right team.
2. Non-branded growth
This is your real SEO scoreboard. Measure clicks, impressions, click-through rate and position for non-branded queries by page group: product, category, blog, location. Ignore blended averages.
3. Brand-assisted overlap
Some queries combine brand and topic, such as a company name plus a product type. These often signal late-stage buyers. Watch whether your pages capture them or competitors and marketplaces do.
Practical examples
A local service firm sees clicks up 20 percent. Segmenting shows branded searches flat and non-branded up on city-plus-service pages, so the new location pages are working and should be extended.
An ecommerce store sees total clicks stable but non-branded clicks falling. A television campaign had inflated brand searches and hidden a category-page problem. Fixing internal links and product schema becomes the priority.
A publisher sees strong image-surface impressions on recipe-style guides with weak clicks. Better original images, descriptive file names and clearer page context are the next test.
A monthly workflow you can run
- Refresh the warehouse and confirm no data gaps.
- Compare branded and non-branded trends against last month and the same month a year earlier.
- List the top ten non-branded pages gaining and losing clicks, and explain each.
- Review multimodal visibility for pages where visuals matter.
- Assign three actions with owners and a date, then check results next cycle.
Conclusion
Branded and multimodal data will not rank your pages for you, but it removes a lot of guesswork about why traffic moves. Use it to give brand teams and SEO teams separate, honest scorecards, and to spot visual opportunities early. If you want help wiring the API into a warehouse and turning the output into decisions, Growth Asteroid builds exactly these reporting systems for growth-minded businesses.
Frequently asked questions
Does the API data replace my brand regex filters?
Largely, but check it against your own list for a few months. Differences between the two will show you where classification is incomplete for your brand names or misspellings.
Should I report branded and non-branded traffic separately to clients?
Yes. Branded traffic reflects existing awareness, while non-branded traffic reflects SEO effectiveness. Reporting them together hides problems and exaggerates wins.
How reliable is the multimodal data?
Use it as a directional signal. Volumes can be small and anonymization removes some rows, so prioritise testing visuals and structured data on pages where it shows clear patterns rather than making precise forecasts.
