Which Google Search Console Table Should You Use in Looker Studio?

When you add a GSC data source in Looker Studio, you pick your property, then choose a table (Site Impression or URL Impression), then a Search Type (web, image, video, news, discover, or googleNews). The screenshots below show what this looks like in the connector interface — Site Impression highlighted in one, URL Impression in the other.

Since it isn’t clear what the difference is, you may not know which choice is the right one or how it will impact your data.

Instead of just choosing one option, you could create multiple data sources from the same GSC property. I’ll come back to why this often makes sense once we’ve looked at what each table is actually counting.

Site Impression and URL Impression count impressions differently when your site shows up more than once in search results for the same query.

Site Impression counts impressions at the property level. If someone searches for a query and Google shows three pages from your tracked GSC property in the results, that counts as one impression for your property. This is how the GSC Performance report chart works when you look at it on the web. When you use Site Impression in Looker Studio, your numbers will be much closer to what you’ll see when you look at Search Console directly.

URL Impression counts impressions at the page level. That same search showing three of your pages counts as three impressions, one per URL. The totals will be higher than what you see in GSC, which is often the source of the “why don’t my numbers match?” confusion.

Here’s an outline of how the same three searches would be counted differently in each table:

Both tables pull from the same underlying data. The difference is in aggregation, not data access. The performance report chart always aggregates by property, which is what Site Impression mirrors. If you want to read more about this, Google’s documentation on what impressions, position, and clicks mean goes into it in more detail.

What can be confusing though is that the gap between the two totals isn’t predictable. The more topical authority your site has in a niche — meaning the more likely Google is to show multiple pages from your domain for the same query — the bigger the discrepancy between Site Impression and URL Impression totals will be.

There isn’t a universally right answer here. It depends on what question your report is trying to answer.

Use Site Impression when:

• You’re analyzing keyword performance across the whole site
• You’re reporting on overall search visibility trends over time
• You’re comparing performance across date ranges at the site level

Use URL Impression when:

• You need page-level analysis — which specific URLs are showing up for which queries
• You’re building a report that lets someone drill into individual page performance

What we will sometimes do for client dashboards is connect both tables as separate data sources. Then you can use Site Impression as the data source for the site-level summary charts and trend lines, and URL Impression as the data source for the page-level breakdown table. This way you get the best of both without the numbers confusingly colliding in the same chart.

Why bother with Site Impression at all when URL Impression is more granular? The short answer is alignment. If you’re looking at an analysis of the full property, you don’t want to show 10,000 impressions in GSC itself and then show 34,000 impressions in Looker Studio because you’re using the URL Impression data source. Building your site-level charts off Site Impression means that you’re reporting on the numbers that make sense for what the charts are reporting on when it comes to overall data. Then you can use URL Impression when you’re digging into the details.

The table choice isn’t the only decision you’re making at the data source level. Search type matters too, and “web” might be the default but it might not actually be right for your reporting needs.

Your options are: web, image, video, news, discover, and googleNews. But that doesn’t mean that you’re stuck with just picking one option. You can either toggle between types in the report, or set up multiple data sources.

Here is how that would work:

For toggling between search types in a single report, you can attach the search type parameter to a dropdown control — one data source handles it, and viewers can switch between web, image, and so on without you needing to duplicate anything. You only need separate data sources if you want to display two search types side-by-side in the same chart simultaneously, or blend data from two types into a single chart.

For Discover and Google News: these search types only work with URL Impression, not Site Impression. If you want a Discover data source, you need to select URL Impression as your table — Site Impression won’t return data for these channels.

Regardless of which table you choose, there’s AI Mode traffic bundled into your GSC data, and it behaves differently from regular web search data in a few ways worth understanding.

Google has confirmed that AI Mode traffic is included in the Performance report. There’s no separate AI Mode section; it’s mixed in with your standard web search data. Clicks from AI Mode do appear in GSC — including clicks on inline links, sidebar citations, and “show all” citation links — but the queries are almost exclusively anonymized and you won’t see the actual search terms driving those clicks.

Position data from AI Mode is also included, but it works differently from AI Overviews. With AI Overviews, all linked pages share a single position. With AI Mode, each link card, carousel item, and image block gets its own individual position using standard search element rules. This makes AI Mode position data harder to interpret, and Google Links within AI Mode responses factor into position numbers as well.

Generating more impressions in AI Mode doesn’t help surface those anonymized queries as the threshold for de-anonymization works differently here than in standard web search.

None of this is something you can change with a connector setting as it’s how GSC currently handles AI Mode data. If you want to go deeper on the mechanics, JC Chouinard published a detailed case study on this that I recommend checking out: Clicks, Impressions and Positions from AI Mode in Google Search Console.

If you’re trying to do something more actionable with AI-driven traffic on the GA4 side, I’ve also written about how to track traffic from AI Overviews in GA4 — that post covers the implementation side of things.

Site Impression and URL Impression aren’t interchangeable, and the connector gives you no guidance on the difference. Site Impression mirrors the aggregated, property-level view inside GSC and it’s the right choice for site-level reporting and for keeping your dashboard numbers consistent with what you see on the web version of GSC. URL Impression is the right choice when you need page-level data for detailed URL analysis.

In most dashboards we build, we connect both as separate data sources and use each one for what it’s actually suited for. The numbers look different because they’re counting different things, and once you understand that, you can make the right choice for what you’re building.

Have questions about this? Check out my related YouTube video, and drop your questions in the comments!