Google adds AI-powered configuration to Search Console Performance reports
Google has announced an experimental feature called AI-powered configuration for the Search results Performance report in Google Search Console. The idea is simple: you describe the report you want in natural language and Search Console automatically sets up the filters, comparisons, and metrics for you. Google’s announcement is on the Google Search Blog here: https://developers.google.com/search/blog/2025/12/ai-powered-configuration.
This feature is rolling out gradually and will only be visible to a limited set of properties at first. If you do not see it yet, that does not indicate a problem with your account or site.
What “AI-powered configuration” actually does
AI-powered configuration acts as a shortcut for building Performance report views that would normally take several clicks and careful selection across dimensions. You type a request such as “Show me queries on phone searches that contain the word ‘sports’ in the last 6 months”, and Search Console attempts to translate that into a configuration using the standard Performance report controls.
Which report it applies to
At launch, it is limited to the Search results Performance report. It is not available for the Discover report or the Google News report, so publishers relying heavily on Discover visibility will still need to configure those views manually.
Which dimensions you can ask for
The prompts are designed to map to the same dimensions you already use: queries, pages, countries, devices, search appearance, and date ranges. That means you can request combinations that are often useful but fiddly to assemble, such as filtering to one device while comparing two time periods and focusing on a specific directory of URLs.
Which metrics you can request
You can ask for common Search Console metrics such as clicks, impressions, average click-through rate, and average position. This matters because many site owners default to clicks only, while CTR and position frequently reveal the more actionable story.
Examples of prompts that translate well into SEO analysis
Natural language prompts are only helpful if they consistently produce reliable configurations. The strongest use cases tend to be questions you already ask during routine SEO checks, especially where you need to isolate a subset of traffic or understand a change over time.
Example: isolating intent terms on mobile
If you run a sports retailer, you might ask: “Show me queries on phone searches that contain ‘boots’ in the last 3 months.” That can quickly reveal whether mobile demand is rising while desktop remains flat, which may affect how you prioritise mobile page speed, product page UX, and snippet optimisation.
Example: comparing a site section year on year
A content-led business might ask: “Compare traffic for pages that contain ‘/blog/’ this quarter to the same quarter last year.” That sort of comparison is useful when diagnosing whether growth is seasonal, whether a content refresh programme is paying off, or whether a broad ranking shift has hit one directory more than others.
Example: market-specific CTR and ranking checks
If you serve multiple regions, you could ask: “Show me average CTR and average position for queries in Spain in the last 28 days.” A dip in CTR with stable position can point to a snippet competitiveness issue (titles and descriptions losing appeal), while a dip in position tends to suggest increased competition, relevance problems, or technical/indexing issues affecting that region’s URLs.
What this means for SEO teams and site owners
The announcement is not a ranking change. It is a workflow change. Even so, workflow changes influence outcomes because they alter what people measure, how quickly they spot issues, and how often they act.
Faster iteration on diagnosis
Search Console is often where an SEO starts when a stakeholder says, “Traffic is down,” or “This category is not growing.” If you can describe an analysis in one sentence and instantly see the configured report, you are more likely to test multiple angles in a shorter time. That can reduce the lag between problem detection and corrective action, especially for teams juggling many sites or large, complex properties.
A higher premium on asking the right questions
When configuration becomes easy, the bottleneck shifts to analytical thinking. The sites that benefit most will be those with a clear set of diagnostic prompts aligned to how the business works, such as prompts for branded versus non-branded queries, high-impression low-CTR queries, or URL groups tied to commercial priorities.
Better collaboration with non-technical stakeholders
Natural language input can lower the barrier for product managers, editors, and marketers who struggle with the interface. That can be positive for SEO if it leads to more shared visibility of performance drivers. It can also be risky if people interpret reports without understanding how filters, comparisons, and attribution limits work in Search Console.
Limitations to keep in mind before you trust the output
Google has been clear that this is experimental and may misinterpret prompts. Treat it as an assistant, not an authority.
Always verify the suggested configuration
Before you make decisions based on the report, check that the correct dimensions, filters, and date ranges were applied. For example, a prompt asking for “pages containing /services” may behave differently depending on whether it matches URL paths exactly, partial strings, or canonicalised URLs. A small mismatch can produce a misleading story.
It does not replace deeper analysis workflows
AI-powered configuration helps set up the view, but it does not handle everything you might do next, such as sorting tables to find outliers or exporting data for modelling, forecasting, or joining with analytics and conversion data. Expect it to speed up the front end of the process rather than replace your established reporting stack.
Not available everywhere yet
The gradual roll-out means some organisations will have it on one property but not another. If you manage multiple domains, document a consistent manual method alongside any prompt-based shortcuts so that your process remains repeatable.
Practical ways to use it without diluting rigour
To get value from faster configuration, pair it with a disciplined approach to interpretation and action.
Create a prompt library aligned to your SEO playbook
Develop a set of standard prompts that map to your recurring tasks, such as diagnosing a traffic drop, finding pages with rising impressions but falling CTR, or checking whether a new template affected mobile performance. Keeping language consistent should also reduce misinterpretation and make it easier to compare reports over time.
Use prompts to test hypotheses, not to generate them
If rankings fell after a release, you might already suspect that a site section or device type is affected. Prompts can rapidly confirm or reject that hypothesis by producing the relevant comparison view. Once the direction is clear, move to methodical checks: indexing coverage, template changes, internal linking, structured data, and content relevance.
Turn report findings into concrete on-page actions
If you uncover “high impressions, low CTR” query groups, prioritise rewriting titles to better match intent, tightening meta descriptions, and improving rich result eligibility where appropriate. If you see “stable CTR, declining position,” focus on content depth, topical coverage, internal linking, and competitive analysis to regain relevance and authority signals.
Questions and answers
Does AI-powered configuration change how Google ranks pages?
No. It changes how you build views in the Performance report. Rankings are not affected by whether you use prompts or manual controls.
Will it help me find quick wins for organic traffic?
It can, because quick wins often live inside very specific slices of data, such as one device type, one country, or one folder of URLs. If prompts make those slices easier to access, you can identify opportunities sooner, especially around CTR improvements and content refresh targets.
What should I do if the AI builds the wrong report?
Correct the configuration manually and then rewrite your prompt in a more explicit way, including the exact dimension and timeframe you want. Treat each misfire as a signal to tighten wording, then reuse the improved phrasing in future checks.
How can I use this for international SEO?
Ask for country-specific metrics and comparisons, then cross-check whether the landing pages shown match the intended locale. If performance diverges by country, follow up by checking hreflang implementation, localisation quality, and whether the correct regional URLs are being surfaced.
Is this useful for technical SEO, or only content SEO?
It supports both. Content teams can segment by query intent and CTR patterns. Technical teams can segment by device, search appearance, and URL patterns to spot issues tied to templates, structured data eligibility, or mobile usability changes that show up as performance shifts.
What to watch next
Google is positioning Search Console as a more assisted decision-support tool, similar to how other Google platforms have introduced AI help for routine configuration tasks. If adoption is strong, expect broader coverage across other reports and more refined interpretation of prompt language over time. Until then, the most effective approach is to use AI-powered configuration as a speed boost, while keeping careful validation and sound SEO judgement at the centre of every decision.
