A new GEO metric from Microsoft Clarity: the AI scrape to referral ratio
Microsoft Clarity published a metric comparing how much AI bots extract from your site against the visits they send back. The question is simple and has gone unanswered for a long time: are these platforms taking your content away, or are they sending visitors in return?
Two directions in one metric
Per the Microsoft Clarity blog post dated August 13, 2026, the new feature compares AI scrape activity against referral traffic in a single metric. The aim is to determine whether AI platforms are extracting content or actually driving visits back to your site.
The metric appears as an AI Scrape-to-Referral Ratio card in the Bot Analytics section of the Clarity AI Visibility Dashboard.
The post also mentions an operator level breakdown offering a ranked list of referring operators. It does not name those operators though, so we do not list which bots are tracked either.
The calculation, and what the source does not say
The ratio is computed exclusively across correctly mapped domains. The reasoning is to prevent misleading results when CDN coverage and Clarity coverage do not align. Where domain coverage gaps exist between bot data and referral traffic, Clarity provides context explaining how that affects the calculation.
What the source does not say matters too. The post gives no threshold or benchmark for what counts as a good or bad ratio. It only stresses evaluating the tradeoff between content extraction and traffic return.
Likewise the post contains no numerical examples, percentages or case data. We therefore do not offer a benchmark figure here. Inventing a threshold absent from the source would defeat the purpose of the metric.
A cost that could not be measured now can be
This metric turns a long voiced discomfort into a number: AI platforms use the text of content producers, but how many visits they send back was unclear. Uncertainty means being the weaker party in a negotiation. That assessment is ours.
Second, the ratio alone does not produce a decision. High scraping with low referrals is not always bad. If your content appears in AI answers there may be a gain in brand awareness, but that is not something Clarity measures. That distinction has to be kept in mind when interpreting.
Third, blocking decisions can now rest on data. If a platform extracts your content heavily and sends no visits, limiting or blocking it becomes a discussable option. Making that call with a ratio is healthier than making it on instinct.
Fourth, the practical consequence of there being no threshold. Since the source offers no benchmark, the first job is establishing your own baseline. Without a month of data there is no way to call a figure high or low. That note is ours.
What it means for businesses in Türkiye
The assessment below is not in the sources, it is our reading. The post contains no Türkiye specific information.
Businesses here that produce content, especially corporate sites investing in blog and guide material, can benefit directly. The question asked for a long time was whether the content we write gets used in AI answers, and if so whether anything comes back. At least the second half is now measurable.
Because Clarity is free, the barrier to entry is low. For a business already running Clarity on its site, there is no additional cost to start tracking this metric.
Three practical steps. First, open the Bot Analytics dashboard and build a one month baseline. Second, place the ratio next to your content calendar and see how newly published guide content behaves on the scrape and referral sides. Third, evaluate the results alongside your bot and access rules, because your blocking settings already shape part of the data in this view.
The UNALSOFT take
On the agentic AI and web design side, what we have been telling clients for a while is this: having your content appear in AI interfaces is a goal, but it is not a strategy until you can measure it. This metric closes one of the largest gaps on the measurement side. Our advice is not to treat it as a standalone success measure. The ratio between scraping and referrals only becomes meaningful when read alongside your blocking rules and your content calendar. Blocking a platform after looking at a single number usually loses visibility rather than winning it.
Is your content extracted, or does it bring visits?
Let us read your bot data and access rules together.