Edition 1
Live race tracking: what AI recommends, November 2026
RaceJoy was the brand the four AIs recommended most often: in 20.4% of decision answers (90% interval 15.4–26.0), followed by Racemap (18.1%) and Garmin LiveTrack (11.1%). The AIs disagreed: ChatGPT recommended RaceJoy in 25.9% of its decision answers, Perplexity in 13.0%; the widest gap among the top 15 was Racemap, 20.4 points between Claude and ChatGPT. The figures come from 240 valid answers of 240 planned, to 60 questions asked once of each AI on 29 October 2026, which named 411 brands in all.
1Leaderboard
| # | Brand | Recommendation rate % (90% interval) | ChatGPT | Claude | Gemini | Perplexity | |
|---|---|---|---|---|---|---|---|
| 1 | RaceJoy | 20.4 (15.4–26.0) | 25.9 | 22.2 | 20.4 | 13.0 | |
| 2 | Racemap | 18.1 (12.7–23.5) | 7.4 | 27.8 | 24.1 | 13.0 | |
| 3 | Garmin LiveTrack | 11.1 (7.2–15.5) | 5.6 | 9.3 | 20.4 | 9.3 | |
| 4 | Open Tracking | 10.7 (6.3–15.3) | 9.3 | 13.0 | 18.5 | 1.8 | |
| 5 | Strava Beacon | 10.7 (7.1–14.5) | 5.6 | 9.3 | 16.7 | 11.1 | |
| 6 | Google Maps | 9.3 (5.0–14.0) | 1.8 | 7.4 | 20.4 | 7.4 | |
| 7 | Owaka | 8.3 (4.8–12.1) | 9.3 | 5.6 | 3.7 | 14.8 | |
| 8 | Garmin inReach | 6.5 (2.8–11.0) | 5.6 | 7.4 | 7.4 | 5.6 | |
| 9 | RTRT.me | 6.5 (3.4–9.9) | 3.7 | 5.6 | 13.0 | 3.7 | |
| 10 | Tracktherace | 6.5 (3.7–9.4) | 0.0 | 3.7 | 20.4 | 1.8 | |
| 11 | LiveTrail | 6.0 (2.9–9.5) | 1.8 | 1.8 | 18.5 | 1.8 | |
| 12 | RaceMyRace | 6.0 (2.8–9.5) | 7.4 | 9.3 | 0.0 | 7.4 | |
| 13 | WHIP LIVE | 6.0 (3.1–9.4) | 1.8 | 0.0 | 18.5 | 3.7 | |
| 14 | Find My | 5.6 (2.6–8.9) | 1.8 | 5.6 | 11.1 | 3.7 | |
| 15 | Map My Tracks | 5.6 (2.7–8.9) | 7.4 | 3.7 | 1.8 | 9.3 |
Percent of decision answers in which each AI recommended the brand, top 15 of 411. Overall rate with its 90% interval. Shown to one decimal; the data files carry four.
2Where the AIs disagree
Among the top 15, the brands with the largest gap between the AI that recommended them most and the AI that recommended them least.
| Brand | Highest | Lowest | Gap (points) |
|---|---|---|---|
| Racemap | Claude 27.8% | ChatGPT 7.4% | 20.4 |
| Tracktherace | Gemini 20.4% | ChatGPT 0.0% | 20.4 |
| Google Maps | Gemini 20.4% | ChatGPT 1.8% | 18.5 |
| WHIP LIVE | Gemini 18.5% | Claude 0.0% | 18.5 |
| Open Tracking | Gemini 18.5% | Perplexity 1.8% | 16.7 |
3How each AI answers
| Per answer, median | ChatGPT | Claude | Gemini | Perplexity |
|---|---|---|---|---|
| Searched the web | 60 of 60 | 60 of 60 | 44 of 60 | 60 of 60 |
| Searches per answer | 3 | 2 | 3 | 3 |
| Pages found | 36 | 18 | 9 | 15 |
| Pages cited | 2 | 6 | 9 | 4 |
| Brands named | 2 | 7 | 6 | 2.5 |
| Brands recommended | 2 | 5 | 6 | 2 |
| Words per answer | 153.5 | 405.5 | 489 | 182 |
Gemini answered 16 of 60 questions without running any web search. That is the model Citewake asked, gemini-3.1-flash-lite, through its API with search available. Across all four AIs, 652 searches were run; 13.8% of them included a year. Every page Gemini reported was one it cited, so its pages found and cited match.
4The sites the AIs cite
| # | Domain | Answers citing it | Share |
|---|---|---|---|
| 1 | apps.apple.com | 29 | 12.1% |
| 2 | racedirectorshq.com | 29 | 12.1% |
| 3 | adventureenablers.com | 26 | 10.8% |
| 4 | reddit.com | 22 | 9.2% |
| 5 | opentracking.co.uk | 21 | 8.8% |
| 6 | racejoy.net | 21 | 8.8% |
| 7 | play.google.com | 20 | 8.3% |
| 8 | owaka.live | 19 | 7.9% |
| 9 | go.racemap.com | 17 | 7.1% |
| 10 | raceid.com | 16 | 6.7% |
ChatGPT
- livedot.me11.7%
- racejoy.net11.7%
- owaka.live10.0%
- support.garmin.com10.0%
- usatriathlon.org10.0%
Claude
- apps.apple.com23.3%
- play.google.com21.7%
- go.racemap.com18.3%
- racedirectorshq.com16.7%
- adventureenablers.com15.0%
Gemini
- reddit.com36.7%
- racedirectorshq.com26.7%
- cobbr.app23.3%
- racemap.com23.3%
- whip.live21.7%
Perplexity
- apps.apple.com21.7%
- owaka.live15.0%
- docs.racemap.com13.3%
- adventureenablers.com11.7%
- racejoy.net11.7%
Share of the 240 answers whose cited pages include the domain: top ten overall, then the top five within each AI's own 60 answers.
5Agreement
On 36 of the 60 questions all four AIs recommended a brand first; they named the same brand first on 1.
6Method in brief
- Questions. 60 scored questions in three tiers (head, mid, tail), written as a buyer would ask them and never naming a brand: 54 decision questions and 6 informational. See the list. A further 15 validation questions are kept private.
- Asking. Each question was asked once of each AI through its official API, with its own web search on and no location sent. Models: ChatGPT: gpt-6-luna; Claude: claude-haiku-5-5; Gemini: gemini-3.1-flash-lite; Perplexity: perplexity/sonar.
- Collection. 29 October 2026, 09:07–09:29 UTC. 240 valid answers of 240 planned: 206 named at least one brand, 34 named none.
- Reading. One grand master model (Claude Opus 5.5) read the four answers to each question side by side and marked each brand as recommended, mentioned or negative. A brand counts only if its name appears word for word in the answer.
- Measure. Recommendation rate is the share of decision answers in which the AI recommended the brand. The overall rate weights each AI and each tier equally, so it can differ slightly from a raw count. Intervals are 90%, from a cluster bootstrap that resamples whole questions. Here the base is 216 decision answers (54 decision questions, 54 answers from each AI).
- Method. v0.8, benchmark v1, scoring run 20261029T100947098Z. The full method: citewake.com/methodology.
7Limitations
- One run. Each question was asked once of each AI, in one 22-minute window. AI answers vary from run to run; the intervals cover variation across questions, not repeat runs.
- English, no market. Questions were in English and no location was sent, so the results describe no particular country.
- Spellings merged. owakalive was counted as Owaka; whiplive was counted as WHIP LIVE; trackmelivecomau was counted as Track Me Live.
- Benchmark figures, not market share. These are benchmark visibility figures across a fixed bank of questions with equal weight per tier, not estimates of market share.
- Some how-to questions tagged as decisions. In benchmark v1, 7 how-to questions in this category were tagged as decision questions. Recommendation rates count every decision question, so these slightly lower all brands' rates alike.
- Top picks are rare. Most answers do not name a single best choice. Even RaceJoy was the top pick in only 0.9% of decision answers, so this report ranks by recommendation rate.
8Disclosure
Citewake is operated by the team that also builds aipozi, Biotree and livedot. Biotree and livedot appear in these categories. They were measured by the same method as every other brand, with no adjustment. In this category, livedot ranks 33rd of 411, recommended in 2.3% of decision answers (0.9–4.1).
9The questions
Each question links to its page, with the four answers in full, the brands read from them and the pages they cited.
All 60 scored questions, with tier and type
The 15 validation questions are never published.
10Data
- Every answer: all 240 in full, with the searches each AI ran, the sites it found and cited, and the brands readCSVJSON
- All 411 brands, every measureCSVJSON
- The reportPDF
Licence: CC BY 4.0. Use and share with credit to Citewake.
11How to cite
Citewake. (2026). Live race tracking: what AI recommends, November 2026 (Edition 1) [Data set]. https://doi.org/10.5281/zenodo.23763271
@techreport{citewake202611liveracetracking,
author = {{Citewake}},
title = {{Live race tracking: what AI recommends, November 2026}},
institution = {Citewake},
type = {Edition 1, data set},
year = {2026},
month = oct,
version = {1},
doi = {10.5281/zenodo.23763271},
url = {https://citewake.com/editions/2026-11/live-race-tracking},
note = {Licensed CC BY 4.0}
}Version 1, published 29 October 2026. Published editions never change; corrections appear as new, dated versions.