Discussion starter: an AI search question, posted with an answer from Niraj Raut to open the thread. If you have dealt with this on a site, reply with what you saw, especially where it differs.
- AI Search
AI Overviews now lead into follow-up conversations. Does a citation in the third turn count the same as one in the first answer in your reporting?
Disclosure: Niraj Raut, who posted this answer, runs the SEO consultancy linked at the end of it.
1 reply
-
adminAdmin
I would not count them the same, and I would not try to force them into a single number. A citation in the first answer reaches everyone who saw the AI Overview; a citation in a third follow-up reaches only the smaller group who kept asking, but those users have narrowed their intent and are often closer to a decision. The two measure different things: reach and consideration. Report them separately where you can measure them, weight them by the purpose of the report, and be honest that most tracking tools and Google’s own reports do not yet separate first answers from later turns.
What changed
At I/O 2026 Google described users being able to ask follow-up questions directly from AI Overviews, continuing into a conversation in AI Mode. That turns what used to be a single answer into a sequence. Each turn can retrieve and cite different sources, and the later turns usually ask for more specific information: pricing, comparisons, how to do something, which option suits a particular situation.
What you can measure today
- Search Console’s “Search generative AI performance” reports show impressions for AI Overviews and AI Mode by page, country, device and date, without clicks, CTR or queries. They do not give you a turn-by-turn breakdown, so you cannot tell from them whether an impression came from a first answer or a follow-up.
- AI Mode clicks are counted within Search Console’s web search totals, without a separate breakdown, and arrive in analytics as ordinary Google organic traffic.
- Prompt-tracking tools mostly track single prompts. Some can run scripted conversations, but multi-turn results are even more variable than single answers, because each turn depends on the previous one.
Why the turns deserve different weights
First answer Later follow-up Audience size Everyone who triggered the AI Overview A subset who continued Intent Broad, often exploratory Narrower, often evaluative or practical Typical cited pages Overviews, definitions, category pages Specific guides, comparisons, pricing, product details What a citation shows Reach and brand exposure Relevance at a decision point A reporting approach
- Define conversation paths for your most important topics: a first question and two or three realistic follow-ups, based on what customers ask sales or support.
- Run them repeatedly, because answers vary, and record citations at each turn.
- Report two measures: first-turn citation rate (reach) and follow-up citation rate (consideration). Show them side by side rather than adding them.
- Connect to outcomes. Pages that tend to be cited in follow-ups, such as pricing and comparison pages, should also show up in Search Console’s AI impressions and in organic landing page data. Track conversions from those pages.
- State the limits. Scripted conversations are an approximation of real ones, and real users’ paths vary widely.
Hypothetical example:
A bank tracks the path “how do I choose a savings account?”, then “which accounts pay the most interest for easy access?”, then “is [bank]’s easy access account any good?”. Across repeated runs, its content is rarely cited in the first answer, which favours broad guides from comparison sites, but its product page and rate table are cited in most third-turn answers. Reporting only first-turn citations would say the bank is invisible; reporting only third-turn citations would overstate its reach. The report shows both, and the team focuses on keeping product and rate pages clear and current, because that is where it is winning.
What this means for content
If follow-ups tend to cite specific pages, those pages deserve attention: clear pricing, honest comparisons, precise product details and how-to content that answers the narrower questions. Broad guides compete for first-turn visibility against very strong sources; specific pages compete in later turns, where fewer sources have exactly the right answer. For many brands the later turns are where they can realistically be cited, and where a citation is most likely to influence a decision.
Where realistic follow-up paths come from
Scripted conversations are only as useful as the paths you script. Build them from evidence of how people actually narrow down: longer, more specific queries in Search Console for the topic; site search terms from people already on your pages; the questions sales and support teams hear after a prospect has done initial research; and the related questions search results show for the topic. A good path moves from a broad question to a comparison to a specific product or situation, because that is how real research tends to progress. Revisit the paths every few months. As AI answers change what users see first, the follow-up questions people ask change too.
Why a single “AI visibility score” hides this
A combined score that treats every citation as equal will rise if you win many low-value first-turn mentions and fall if you lose a few decisive follow-up citations, which is the opposite of what matters commercially for most businesses. Keeping the measures separate prevents that distortion and makes the report useful for deciding which content to invest in.
Watch for this:
Multi-turn tracking multiplies variability. A path of three prompts, each with variable answers, can produce very different third-turn results from run to run. Run each path enough times to see a stable rate before reporting it, and avoid drawing conclusions from week-to-week changes of a few points.
Count first-turn and follow-up citations separately, treat the first as reach and the later ones as consideration, connect both to page-level outcomes, and be explicit that neither Google’s reports nor most tools distinguish them yet.
Need help with this on your own site?
Niraj Raut works with businesses in Nepal, Australia, the UK, Europe and the US on technical SEO, ecommerce SEO, local SEO and AI search.
Add a reply Cancel reply
Replies are open to approved members. Log in or apply to join.