The method stays public.
Fixed buyer questions, fresh sessions, verbatim answers, model, and date. The same measurement repeats every week.
Eight questions, three assistants, 24 answers. Published unchanged.
One of the eight questions, read across three assistants.
“What’s the best help desk software for a support team of about 15 agents?”
Named most often across 24 answers
- 1Zendesk14 of 24
- 2Freshdesk13 of 24
- 3Help Scout12 of 24
Free, open source, also named: Zammad (avg. position 2.5) · osTicket (avg. position 3.5) · FreeScout (avg. position 4.2)
Help desk software category · measured July 29, 2026 · not a customer · read the raw file, or run the question yourself.
Eight questions a support manager actually asks, run across three assistants. The names above are what came back. One measured vendor was named in 2 of the 24 answers and is not identified on this public page.
- Free projects took the two best positions in the category.Zammad averaged position 2.5 and osTicket 3.5. Zendesk, the most-named vendor in the run, averaged 4.6.
- Named most often is not the same as named first.Zendesk led on mentions with 14 of 24 answers and still averaged position 4.6, behind Help Scout at 3.8 on 12.
- Paste the question above into any assistant right now.If your category works like this one, the gap is already there.
This is the document, not a description of it.
A real run from July 29, 2026: 8 buyer questions, three assistants, 24 answers. The raw file ships with this site, so every number below can be recounted from the run itself.
| Brand | Named | Avg. pos. |
|---|---|---|
| Zendesk | 14/24 | 4.6 |
| Freshdesk | 13/24 | 4.6 |
| Help Scout | 12/24 | 3.8 |
| Zoho Desk | 7/24 | 4.3 |
| Jira Service Management | 7/24 | 5.7 |
| HappyFox | 6/24 | 7.8 |
Read it like this: Help Scout is named in half the answers and sits at 3.8 in average position, ahead of both brands above it. Being named more often and being named first are two different problems, and a report that only counts mentions cannot tell you which one you have.
| Brand | ChatGPT | Grok | Perplexity |
|---|---|---|---|
| Zendesk | 4/8 | 7/8 | 3/8 |
| Freshdesk | 2/8 | 8/8 | 3/8 |
| Help Scout | 3/8 | 5/8 | 4/8 |
This is the half a single check never shows. Freshdesk is named in every answer Grok gave and in two of eight from ChatGPT. Asking one assistant once tells you almost nothing: the same category, asked the same way, produces a different shortlist per assistant.
Your own report carries every question with its verbatim answer per assistant, the model version and date, and the questions where you are absent. In this run Help Scout is missing from 12 of the 24. On Solo and Agency that list becomes the publishing queue.