Measure and improve AI recommendations
Aions shows who AI recommends for the decision questions in your category, what the agent sees on your site and what you should change.
No account, no subscription, no cost.
14% (12 to 16%) after 21% (19 to 23%)
21 Sep to 2 Oct · 3 to 20 Oct · action 3 Oct · Intervals do not overlapWhere you lose
In a measurement for a Swiss online shop, it appeared in 18% of answers to questions about the manufacturer, a product feature or delivery time, and in 3% of answers to questions about the best shop, a shop comparison or returns. We formed the groups after the measurement.
Part 1 of our study, on car insurance, shows: when a question states conditions such as a price limit or cover, the providers AI systems mention change markedly. Search terms instead of full sentences change little.
Visa, Switzerland, survey summer 2026
FHNW online retailer survey 2026 (in German) (opens in a new tab)
Example from our work
A well-known Swiss online shop, share of AI answers that mention it; groups formed after the measurement
Do you know on which questions your shop drops out of the answer?
No account, no subscription, no cost.
The most important input
Buyers ask decision questions shortly before they buy, usually with conditions such as price, delivery time or returns. Measure the wrong questions and you improve the wrong answers.
| Question | Condition | Mention Rate | Chosen instead of you |
|---|---|---|---|
| Which office chair helps with back pain? | none | 41% · 49 of 12032 to 50% | You lead |
| Office chair under CHF 600, delivered by Friday? | price, delivery time | 6% · 7 of 1203 to 12% | Provider A |
| Where can I try a chair for 30 days and return it free? | returns | 0% · 0 of 1200 to 3% | Provider C |
| Ergonomic office chair made in Switzerland? | origin exploratory | 6% · 1 of 161 to 28% | Provider B |
Last 30 days · core questions daily (120 answers per question), exploratory questions weekly (16 answers)
Three different return periods on your pages
Finding CONFLICTING_RETURN_POLICY · action ready
The difference
Whether AI mentions you is one of four stages. Aions audits the whole chain from the agent's point of view.
Why now
Instead of ten links, there is one answer.
38% of respondents in Germany, France and the UK use AI tools to research products and services or to make purchase decisions. 63% of them use these tools to compare brands, models, prices and reviews.
McKinsey, Consumer AI Discovery Survey, December 2025, published 2 March 2026, 749 respondents · Europe’s agentic commerce moment
If you measure now, you have a starting point for every later comparison.
Examples from our work
Case A · Online shop, Switzerland
Rank 13 of 47 brands. Mentioned in 18% of answers to questions about the manufacturer, a product feature or delivery time, and in 3% for questions about the best shop, a shop comparison or returns. We formed the groups after the measurement. Finding on its own pages: three different statements on returns. Recommended: clean up the purchase conditions, then measure again with the same questions.
Case B · Manufacturer selling to businesses and consumers
The way the pages were delivered was changed; afterwards, 6 of 10 crawler user agents saw the content. Mention Rate after the change: 8.7% 1.
1 Re-test across all measured AI systems, including new systems that were not yet part of the first measurement. On the same systems as in the first measurement: 4.0% → 6.6%.
Findings catalogue
Our work with shops and manufacturers produced a catalogue of recurring findings. Each one with a plain-language description, action, evidence URL, acceptance criterion and responsible party.
The platform
What each crawler sees on your pages, page by page.
| Page type (pages) | Googlebot search index | GPTBot training | ChatGPT-User live fetch | Claude-User live fetch |
|---|---|---|---|---|
| Home page (1) | ✓ content | ✓ content | ✕ empty | ✕ empty |
| Category (4) | ✓ content | ✓ content | ✕ empty | ✕ empty |
| Product (9) | ✓ content | ✓ content | ✕ empty | ✓ content |
| Returns (2) | ✓ content | ✕ blocked | ✕ blocked | ✕ blocked |
| Purchasable | Preparation no information for agents (UCP, MCP) | |||
40 pages audited, a sample per template · shown: 4 of 7 page types
ChatGPT’s live fetch receives an empty page
Mention Rate, Win Rate, Share of Voice, sources and competitors. Every figure with its precision.
| Week | Answers | 95% interval |
|---|---|---|
| Week 39 | 98 of 700 | 12 to 17% |
| Week 40 | 110 of 700 | 13 to 19% |
| Week 41 | 146 of 700 | 18 to 24% |
| Week 42 | 145 of 700 | 18 to 24% |
| Week 43 | 41 of 200 | 15 to 27% |
Week 43: 19 and 20 Oct only, hence the wider interval. Scale 0 to 35%.
Intervals do not overlap: the difference holds. Marked: Price and availability added to the markup. The same 25 core questions and the same 4 systems before and after.
Decision questions for your category, with purchase conditions.
Concrete actions from the findings catalogue, with acceptance criteria.
| ID | Requirement | Classification | Benefit | Acceptance criterion | Responsible | Status |
|---|---|---|---|---|---|---|
| RND-01 | The home page, categories and product pages deliver their content in the initial HTML, without JavaScript. | binding | high | ChatGPT-User and Claude-User receive the text in the HTML on all 14 sample pages that are empty today. | Implementation partner | in progress |
| POL-01 | One return period for the whole shop, marked up on the shop operator in machine-readable form (MerchantReturnPolicy). | binding | high | The same period on /returns, /faq and /terms and in the markup. | Your team | open |
| POL-02 | Shipping cost and delivery time to Switzerland in the markup (ShippingService), the same as on the shipping page. | binding | medium | Markup, shipping page and checkout match on all 9 product pages. | Your agency | in progress |
| AVL-01 | Price, currency and availability are in the offer on every product page (Offer). | binding | high | Visible price and markup match, checked on 17 Oct. | Your agency | ✓ confirmed |
| AVL-02 | Every variant (colour, armrests) has its own offer with price, availability and variant URL. | recommended | medium | When switching variants and when calling the variant URL directly, page and markup show the same price. | Your agency | open |
| ORG-01 | The shop operator appears as one organisation with legal name, address and contact, without duplicates from the theme and apps. | recommended | supporting | The audit finds at most one entry for the shop operator on each page. | Implementation partner | open |
Binding: open items block acceptance · Recommended: does not block acceptance · Benefit: estimated lever for data quality and machine readability, not a promise of visibility · Aions checks each requirement against its acceptance criterion
Who it's for
Online shops and brands
Readiness, recommendations and monitoring as self-service. You or your agency implement.
From CHF 89 per month
Companies that want support
Question design, readiness with a test purchase by an agent, tailored recommendations. We coordinate implementation with implementation partners or your agency.
On quotation
SEO and GEO agencies
Up to 10 brands in one account, client reports with your logo. You implement, Aions measures.
From CHF 790 per month
Pricing
The free check and the readiness report are available now; access to the plans on request.
CHF 89 per month
1,200 credits (up to 4,800 answers), up to 3 AI systems, sources and sentiment.
CHF 290 per month
4,500 credits (up to 18,000 answers), up to 6 AI systems, monthly readiness audit, all recommendations.
from CHF 790 per month
Up to 10 brands, 13,500 credits (up to 54,000 answers), client reports with your logo.
Insights
All articlesAions study 2026 · Part 1: Car insurance
Part 1, car insurance: conditions in the question markedly change which insurers AI systems mention. Part 2, with another category, will follow.
Share of the five most-mentioned insurers, average of ChatGPT, Gemini, Copilot
Who we are
Currently running projects with Swiss companies, with a focus on e-commerce.
The free check shows up to three of the most serious findings on your website, by email.
No account, no subscription, no cost.