Provisional: 1 test run. Every score below comes from a single run of our own test. See how we test.
The test
Write an SEO title and meta description for a fictional e-bike shop page, list 5 on-page fixes in priority order, and suggest 5 related keywords. No keyword data was supplied, so a tool must not quote search volumes.
Results
| Tool | Plan tested | Score |
|---|---|---|
| Claude | Opus, paid plan | 90/100 Provisional: 1 test run |
| ChatGPT | "Instant" model | 89/100 Provisional: 1 test run |
| Perplexity | Free plan | 75/100 Provisional: 1 test run |
Claude: 90/100 (Provisional: 1 test run)
Quoted no volumes and said its keywords were not measured demand. Title and meta were within limits, but it misreported the meta length and put the title fix last.
| Criterion | Points |
|---|---|
| Honesty | 30/30 |
| Technical correctness | 22/25 |
| Prioritisation | 15/20 |
| Writing | 13/15 |
| Friction | 10/10 |
ChatGPT: 89/100 (Provisional: 1 test run)
Stated plainly that it had no volume data. Title and meta were within limits, but its own character counts were wrong.
| Criterion | Points |
|---|---|
| Honesty | 30/30 |
| Technical correctness | 21/25 |
| Prioritisation | 17/20 |
| Writing | 13/15 |
| Friction | 8/10 |
Perplexity: 75/100 (Provisional: 1 test run)
Gave no volumes but did not flag their absence. Its title ran to 64 characters (over 60) and it pulled sources about Hamilton, Ontario instead of Hamilton, New Zealand.
| Criterion | Points |
|---|---|
| Honesty | 25/30 |
| Technical correctness | 16/25 |
| Prioritisation | 14/20 |
| Writing | 12/15 |
| Friction | 8/10 |
Caveats
Claude and ChatGPT are within likely run-to-run variation. We plan a second run of every tool before treating any ranking as settled.