Which indoor cycling brands does AI recommend in the UK?

On 19 October 2026 Manual Focus asked six AI engines 309 UK shopping questions about indoor cycling, three times each. Across 5,562 answers, AI named Wahoo in 83% of smart trainer answers, Zwift in 88% of training app answers and Wahoo in 64% of smart bike answers, while cycling media supplied over half the citations.

Brands in the study can request a free one-page AI Shortlist card.

Key takeaways

• Wahoo is AI’s default smart trainer, named in 83% of unbranded trainer answers and placed first by all six engines.
• Zwift leads training apps at 88%, but Rouvy wins scenery questions (97%), MyWhoosh wins value questions (87%) and TrainerRoad leads performance training (84%).
• Smart bikes are close, with Wahoo named in 64% of answers and Wattbike in 58%.
• Challenger brands depend on the engine, as JetBlack ranged from 47% of Claude’s trainer answers to 11% of Google AI Mode’s.
• Editorial sites supplied 51% to 56% of citations, while Reddit, YouTube and forums supplied 3% to 5%.

Why does AI’s answer matter now?

More shopping journeys now start with an AI answer at the top of the results page. Sundar Pichai said at Google I/O on 19 May 2026 that AI Overviews has over 2.5 billion monthly users and AI Mode has passed one billion. In the UK, Ofcom found that about 30% of searches show an AI Overview, and ChatGPT’s UK visits rose from 368 million to 1.8 billion in the first eight months of 2025.

The answer also keeps the click. Pew Research Center found that people clicked a search result in 8% of visits when an AI summary appeared, against 15% without one, and clicked a link inside the summary in only 1%. Google now shows an AI Overview on 82% of the brand searches DemandSphere tracks, and it moved links higher inside its AI answers in October, as we covered in what Google’s new AI Mode links mean for brand websites.

For a shopper, the brands named in that answer are increasingly the shortlist. Our study measured that shortlist for one category, and shopping questions triggered an AI Overview far more often than Ofcom’s all-search average, in 64% of our searches.

What did we measure?

We wrote 309 questions the way UK riders ask them, such as “quietest smart trainer for a flat”. The 260 that name no brand make up every headline number, and each carries a tag for the buying moment behind it, from value and beginner to performance and problem questions.

The six engines were ChatGPT and Gemini as the public sees them, Google AI Overviews, Google AI Mode, Perplexity and Claude with web search, all set to a UK location and collected through DataForSEO. Each question ran three times per engine, because AI answers change between runs. An engine named exactly the same brands across all three runs only 20% of the time, from 13% for ChatGPT to 48% for Perplexity, which matches a SparkToro study from January 2026 advising brands to track how often they appear rather than their position.

So the headline measure is the shortlist rate, the share of answers that name a brand, averaged across the six engines, with a 95% range on every figure.

The AI shortlist by category

Wahoo’s trainer lead is the clearest result, with a 95% range of 79% to 87%. Rouvy and MyWhoosh are effectively tied for second among apps, and Wahoo’s smart bike lead over Wattbike is narrow.

Who wins which buying moment?

For a brand team, the more useful finding sits below the headline table, because the leader changes with the question behind the search.

A brand that wins performance questions can still be missing from the beginner shortlist. Van Rysel, Decathlon’s own brand, appears in roughly half of value answers for trainers but in 9% of performance answers. Peloton, a fitness brand more than a cycling one, leads the beginner moment for smart bikes.

The category entry points playbook is the method for mapping these moments in your own category before you measure them.

Do the AI engines agree?

At the top, yes. Every engine put Wahoo first for trainers and Zwift first for apps. Further down the list the engines behave like different shops with different buyers.

Claude names more brands per answer than any other engine, 4.2 on average against 2.9 to 3.3 for the rest. Saris appears in 40% of Claude’s trainer answers and under 10% of every other engine’s. Van Rysel appears in 43% of Claude’s trainer answers and 37% of ChatGPT’s, but in only 5% of Google AI Mode’s.

Because AI Overviews appeared for only 64% of these searches, they give the lowest rate for most leading brands, including Wahoo, Tacx, Zwift and Wattbike.

Our view at Manual Focus is that a brand team should treat each engine as a separate channel with its own shortlist, because a single “AI visibility” score averages away the gaps a challenger brand can actually close.

Which sites decide the answer?

Cycling media supplies most of the sources. Editorial sites supplied 56% of citations for trainers, 53% for smart bikes and 51% for apps, led by BikeRadar, Cyclingnews and Cycling Weekly. Reddit, YouTube and forums supplied 3% to 5%. Brand websites mattered most for apps, at 30% of citations, where Rouvy’s and MyWhoosh’s own pages were cited in a large share of answers.

So the work for a brand team is mostly earned media, in line with our earlier argument that the citation moved off the page. For an endurance brand, the review in a buyer’s guide on BikeRadar now does double duty, as the article a rider reads and as the source an AI engine quotes. The earned media pitch generator and the SEO playbook for premium endurance brands cover how to earn those placements.

What to do with this

1. Find your buying moments. List the five questions a buyer in your category asks most and check which brands AI names for each one, starting with the moment you most need to win.
2. Check every engine separately. Run your top questions through ChatGPT, Gemini, Google and Claude, because your position on one says little about the others.
3. Audit the cited sources. Note which buyer’s guides the engines cite for your category, then check whether your products appear in them and whether the details are current.
4. Fix your own pages for the questions brand sites win. App and software brands get cited directly, so answer value, compatibility and beginner questions plainly on your own site.
5. Measure again in three months, using the same questions, because a single snapshot can’t show whether your work moved anything.

What this study can’t tell you

The study counts whether AI names a brand, not whether it praises the brand, and not whether anyone buys. Logged-in users with chat history may see different answers. Perplexity and Claude were measured through their APIs, which can differ from their apps. This is a snapshot of October 2026.

Disclosure, James Vickers has previously worked with Wahoo and Wattbike and was on the founding team at RGT Cycling, which Wahoo acquired. He currently works with Ribble Cycles, which isn’t ranked but appears as a cited retailer in 19 of the 5,562 answers. No brand paid for or saw this research before publication.

Further reading

• The interactive report, with every brand and question.
• The citation moved off the page, on why rankings stopped predicting citations.
• SEO playbook for premium endurance brands, the search groundwork behind AI visibility.
• Owned, paid and earned media explained, for the earned side that AI engines lean on.
• Category entry points, for mapping the buying moments in your own market.

If you want to know where your brand sits on the AI shortlist, talk to us.

I run marketing for [brand], which sells [category] in the UK.

I want the questions buyers ask before they choose a brand in my category,
grouped by the buying moment behind them. Use these moments: value,
beginner, performance, problem (something isn't working), and any others the
data shows.

Attached are:
- a Search Console export of queries for the last 3 months
- our customer service tags or ticket subjects for the same period

From these, write the 20 questions a buyer in [category] asks most, phrased
the way a UK buyer would type them into an AI assistant, with no brand names
in the question. Tag each with its buying moment.

Give me a table with the columns: Question | Buying moment | Evidence (the
queries or tags it came from) | Why it matters for us.

Then name the one buying moment we most need to win and say why. Don't
invent search volumes; if a figure isn't in the files, say so.

Set this up as a routine that runs on the first Monday of each month.

Here are 10 unbranded buyer questions for [category] in the UK:
[paste questions]

For each question, get the answer from each AI engine we track: [list the
engines you can reach, for example your own assistant plus Google's AI
answers]. Ask each question three times per engine, in a fresh
conversation each time, set to a UK location where possible. If you can't
reach an engine yourself, tell me and give me a simple form so a person can
paste the answers in.

Record which brands each answer names. Then give me a table with the
columns: Engine | Question | Brands named (all three runs) | Times [brand]
was named (0 to 3) | Times each of [competitor 1], [competitor 2] and
[competitor 3] was named.

Finish with our shortlist rate per engine (answers naming us divided by all
answers), the change since last month, and the engine where we're weakest.
Say plainly which runs you couldn't complete.

I run marketing for [brand] in [category], UK.

Search for the buyer's guides, reviews and round-ups most likely to be cited
when someone asks an AI engine to recommend a [category product] in the UK.
Start with the specialist titles for [category], then any others that rank for
these questions: [paste 5 of your buyer questions].

For each guide, give me a table with the columns: Site | Guide title | URL |
Date last updated | Is [brand] included? | Which of our products, and is the
detail current (price, specs, model name)? | Who is listed instead of us.

Then give me a short hit list: the five guides where getting included or
corrected would matter most, and why. Don't guess at a guide's contents;
if you couldn't open it, mark it "not checked".