Optimizing for answer engines: who measures what, and who measures nothing
Two of the three major answer engines provide an official measurement of your citations: Google since June 2026, Microsoft since February 2026. Perplexity documents its crawlers but offers publishers no report. The first decision is therefore not what to optimize, it is where you will know whether it works.
- Microsoft came first, on 10 February 2026, with a citation report in the Bing Search Console, extended in June.
- Microsoft is also the only one to deliver grounding queries, the searches Copilot runs internally to build its answer. It is the most useful data of the lot.
- Google provides impressions by page, country, device and date, but no click and no query.
- Perplexity documents its crawlers and its address ranges, with no performance report. Measurement there stays a manual reading.
- Your robots.txt file decides whether these engines can even read your site. It is the least costly setting in the whole field, and the most often overlooked.
On this page
Grounding query
A grounding query is the search an answer engine runs itself, behind the scenes, to fetch the pages it will use to build its answer. It differs from what the person actually typed: the system rephrases the question, breaks it up and runs several searches. Microsoft is so far the only one to publish these queries to site owners, which makes it the most directly usable data in the whole field, because it reveals the real vocabulary that triggers a citation.
Who measures what
The market talks about this topic as if the three engines behaved the same way. They do not resemble each other, and the gap decides what you will be able to demonstrate at the end of an engagement.
| Engine | Official report | What it gives | What is missing |
|---|---|---|---|
| Microsoft Copilot and Bing summaries | Yes, since February 2026 | Citations, cited pages, grounding queries, intents, topics, citation share | No data on the other engines |
| Google, AI summaries and conversational mode | Yes, since June 2026 | Impressions by page, country, device, date | No click, no query |
| Perplexity | No | Documentation of crawlers and address ranges | Any citation measurement |
| Other assistants | No | Nothing standardized | Everything |
The surprise, for a management team used to Google's dominance, is that Microsoft offers more here. Its report delivers grounding queries and, in June 2026, added intents, topics, citation share and a comparison function. Google, for its part, isolates impressions without giving either click or query. For a budget decision, the gap is not trivial: you can demonstrate a result with one and only a presence with the other.
The budget consequence is direct and counter-intuitive. If you want to learn which vocabulary triggers a citation in your sector, the least costly source is a free account in the Bing tools, even if 95% of your traffic comes from Google. The grounding queries found there then serve to write for every surface, including those that measure nothing.
Using AI to understand your own buyers, rather than to be seen by it, is covered in our article on the synthetic persona.
Before deciding on a budget faced with an offer that stacks GEO, AEO and SXO, our article on the optimization acronyms separates real work from mere renaming.
What to settle before funding AI visibility tracking
The cost of a third-party dashboard often exceeds that of the work it measures. Before paying for a subscription, it is better to check what the engines themselves give for free, because the answer has changed twice in six months.
- Is our site verified in the Bing tools, which is free and gives the grounding queries?
- Do we have access to Google's dedicated report, and has anyone even looked at it?
- Which subscription are we paying for, and what does it add that these two sources do not give?
- On which surfaces do our customers actually search before they buy?
- If no citation is recorded within nine months, what do we stop?
The answer that holds up starts with the free sources and adds a paid tool only for what they do not cover. A weak answer sells a cross-platform dashboard before checking the official reports, or presents a sampled reading as an exhaustive measurement.
Deciding which part of this tracking deserves a subscription rather than a free account is settled on the surfaces where your customers really search. A 90-minute consultation settles it, with a written summary your team can execute.
The file that decides everything
Before any question of measurement comes a question of access, and it is settled in a text file at the root of your site. An engine that cannot read your pages will never cite you, whatever the budget committed to content.
Perplexity documents how it works: its crawlers collect information either automatically or in response to a person's request, and site owners manage each case independently through rules in the robots.txt file. The documentation states that a change can take up to 24 hours to be reflected, and publishes the address ranges of each crawler to allow verification.
Many businesses blocked AI crawlers in 2024 and 2025, often as a precaution and sometimes with no management decision. That block has a cost that shows up today: it removes the business from generated answers, with no compensation and no measure of the traffic it gives up. This setting deserves an explicit review, because it affects brand visibility and was decided at a time when no one could assess its cost.
What this changes in execution
Three practical consequences, in order of return.
Check access before producing
Reread the robots.txt file and decide, in writing, which crawlers can read the site. It is one hour of work, no recurring cost, and it conditions everything else. Producing content while a block is active amounts to paying for production no one will read.
Open an account in the Bing tools
Free, and it is the only source of grounding queries available. These queries reveal the real vocabulary that triggers a citation, which no keyword tool gives. They then steer production for every surface, Google included.
Record by hand what is not measured
On surfaces with no official report, set a list of questions from your salespeople and ask those questions every month. It is sampled, so directional, and it is better than nothing. A paid tool automates this reading without making it exhaustive: the cost is added, the reliability does not change in nature.
What stays in-house: the decision whether to open the site to AI crawlers, the list of questions your customers ask before buying, and the proprietary data you agree to publish. What gets delegated: configuring the robots.txt file, verifying the properties in the Google and Bing tools, the monthly reading of the reports, the recording on unmeasured surfaces and the rework of content. A business that opens access and verifies its two properties gets more citations than a business that buys a dashboard before doing either.
The general framework and Google's documented position are in from SEO to GEO, the comparison of the two disciplines in SEO and GEO, and the structure that makes a page citable in the format that gets cited. The technical checks on the site are detailed in what blocks a site in search.
Going further on AI visibility
Understanding the ground. SEO and GEO: what differs between the two, GEO, AEO, AIO, SXO, SGE: several acronyms for a single job, Generative AI and content marketing: what changed, Google's AI Overviews: what the browsing data shows, AI traffic in ecommerce: the percentage lies, the conversion speaks.
Getting cited. Answer-first: structuring content to get cited by answer engines, From SEO to GEO: what large companies change, Manufacturers and B2B vendors: getting kept in an AI-generated shortlist.
Controlling and measuring. Business owners: should you let AI crawlers access your content?, ChatGPT advertising: what is in place and what it decides, Synthetic persona: what research validates and what it does not, 360Brew: what LinkedIn's new ranking changes for your visibility.
Taking a position in answer engines before your competitors is at the heart of the Strengthen your visibility in AI answers goal.
Already running a marketing team? See how we plug in as reinforcement on answer engine optimization.
Frequently asked questions about answer engine optimization
Which engine gives businesses the most data?
Microsoft. Its report in the Bing tools, launched on 10 February 2026 and extended in June, gives the citations, the cited pages, the grounding queries, the intents, the topics and the citation share. Google provides impressions with no click or query. Perplexity provides no report.
What is a grounding query?
It is the search the engine runs itself to fetch the pages it will use to build its answer. It differs from what the person typed, because the system rephrases and breaks up the question. Microsoft is the only one to publish these queries, which makes it the most usable data in the field.
Should you block AI crawlers in your robots.txt file?
It is a management decision, not a technical setting. Blocking removes the business from generated answers, with no compensation and no measure of the traffic given up. Many blocks were set as a precaution in 2024 and 2025 and deserve a review now that measurement partly exists.
Should you pay for a cross-platform tracking tool?
Only for what the official reports do not cover, that is, the surfaces with no publisher report. Checking your properties first in the Google and Bing tools, which are free, avoids paying for data already available. On the rest, a tool automates a reading that stays sampled.
- Bing Webmaster Blog, AI Performance in Bing Webmaster Tools, public preview announcement, 10 February 2026.
- Bing, New AI Visibility Insights in Bing Webmaster Tools, June 2026.
- Google Search Console, help center, Generative AI performance report (Search), accessed July 2026.
- Perplexity, Perplexity Crawlers, official documentation, accessed July 2026.
Gabriel almost always takes your first call and carries out your audit. He builds the strategy starting from your growth goal: where to put your budget, which market to test and how to connect each lead to a real sale in your CRM. He mainly leads engagements for three goals: Optimize the profitability of your digital campaigns, Develop a new market, and Generate demand and growth. With Geneviève, he also works on organic search (SEO), AI visibility (GEO) and conversion rate optimization (CRO). The sales a Google Ads or Meta Ads campaign brings in depend on the page that receives the click. He writes mainly about marketing strategy, paid advertising and measurement.
About Falia →
SEO
Paid advertising
Conversion
AI visibility
Web design