Does a high Google ranking mean AI will recommend you?

Not reliably. In our research, Google search traffic explained only about a third of the difference in who AI assistants recommend to buyers. The other two thirds come from something else, and your rankings report doesn’t show it.

The study measured one category in one market: twenty price monitoring services in the US, forty questions buyers actually ask, and four answer engines (ChatGPT, Google AI Overviews, Google AI Mode and Copilot). That is 160 answers, collected in September 2026. The numbers belong to that category and that market. The lesson travels further, including to businesses in Montenegro.

How much does your Google ranking predict?

Only part of the picture. The rank correlation (Spearman) between search traffic and mention share in AI answers is 0.58: a real link, but one that explains only about a third of the variation between services.

Mention share is the proportion of buyer questions in which an AI system names a service, weighted by how often each question is asked.

The names make it concrete. Priceva has the most US search traffic in the sample, 5,244 visits a month, yet ranks thirteenth of twenty for visibility in AI answers. Minderest gets 557 visits and ranks fifth. A tenth of the traffic, and a better result where the choice is made.

Three sites with almost identical traffic make the point more sharply:

• Prisync64.2%5,020 visits
• Minderest24.8%557 visits
• Pricefy24.6%5,013 visits
• Priceva6.9%5,244 visits

Source: our research · Ahrefs Brand Radar and Site Explorer · USA · September 2026

Prisync, Pricefy and Priceva are within five per cent of each other on traffic. Their visibility in answers differs ninefold.

An answer gives three names, not ten links

An AI answer doesn’t hand the buyer a list to choose from; it hands them a choice already made. The average answer in our study names about three of the twenty services.

On a results page the buyer picks from ten links, so fourth place is still on page one. In an AI answer there is no fourth place: you are on the shortlist or you aren’t. The question that matters is not who comes first, but who makes the shortlist.

The shift is quiet. Rankings hold, traffic holds, reports look as they always have, and at the moment the buyer forms a consideration set your name may simply be missing. No standard report will tell you.

Buyers ask in four different ways

The overall mention share hides the most useful finding: visibility varies several-fold depending on how the question is asked. Before collecting any answers, we split the forty questions into four buying modes, then analysed every answer individually.

Buying mode	What the buyer does	Answers naming at least one service	Best single service
Short list	asks for recommended tools	100%	70%
Two-way comparison	compares two products by name	100%	42%
Cost	asks what a solution costs	93%	64%
Task description	describes the problem without naming the category	59%	19%

Share of answers within each mode, 160 answers, unweighted. Source: our analysis of answers from Ahrefs Brand Radar.

When buyers describe the task in their own words, only 59% of answers name any service at all. The best-performing service appears in 19% of them, against 70% in short-list answers. Three services that are visible in the other modes (Intelligence Node, DataWeave and Pricefx) are never named here: only buyers who already know what the category is called will come across them.

This is the earliest moment of choice, and two companies out of three aren’t in it.

The losses already show up inside Google

You don’t have to care about ChatGPT for AI answers to cost you. For phrases where the services already rank, Google shows an AI Overview above the results, and the share of those phrases where a service is cited inside the box ranges from 45.8% (Pricefy) to 2.0% (Priceva).

That is a 23-fold gap in a channel the category already pays for. Priceva is the telling case again: the phrases where it does get into the box are consumer questions about store policies, asked by people who don’t buy price monitoring software. Ranking beneath an overview that names your competitors and not you is a lost click on a query you had already won.

What to measure alongside rankings

Rankings and traffic are still useful, but they can’t tell you whether you’ll be on the list the buyer sees first. For that you need mention share by buying mode, read side by side with your rankings.

In practice, that means five steps:

1. Freeze your question list. Write down the questions your buyers really ask and keep them unchanged between measurements. Otherwise you can’t tell whether your visibility moved or the list did.
2. Split the questions into four modes before the first measurement: short list, comparison, cost and task description.
3. Measure mention share for each mode separately, and note who appears alongside you. Those are your real competitors in answers.
4. Check Google AI Overviews for phrases you already rank for: on what share of them are you cited inside the box, not just listed below it?
5. Repeat on the same list. The same question can return different names, so a single measurement can’t separate change from chance. Base decisions on a series of at least three.

What this means for a business in Montenegro

The figures in this study don’t transfer to the Montenegrin market, and we have no Montenegrin data. What transfers is the mechanism: an AI answer names a handful of businesses, and the buyer chooses among them.

Take a hotel in Budva that ranks well for the town name and type of stay. A guest opens ChatGPT and describes what they need instead, say a quiet place near the beach for a family with small children. That is a task-description question, the mode in which the category in our study nearly disappeared. Whether the same holds for your business, only measurement will show.

Where to start:

• Take the AI search readiness quiz: six questions, three minutes, and a clear view of where your brand stands today.
• Read the full research, including the methodology and limitations.
• If you want this measurement for your own category, our AI Search Strategy Programme includes a query portfolio grouped by decision stage and a measurement plan with a dashboard.