CLAWSIGNAL RESEARCH / OWNED ACCOUNT
ClawSignal AI visibility baseline
A dated measurement with the zero result left intact. This is ClawSignal's own account, not a customer case study and not a claim of future lift.
Observed July 19, 2026 · Published July 22, 2026
What the baseline found
ClawSignal ran 10 fixed service prompts across ChatGPT, Claude, Gemini, Grok. The production records contain 40 checks, with 0 ClawSignal appearances and 0 citations. A tracking system is useful only if it can report an unfavorable result without smoothing it into a success metric.
| Assistant | Checks | Appearances | Citations |
|---|---|---|---|
| ChatGPT | 10 | 0 | 0 |
| Claude | 10 | 0 | 0 |
| Gemini | 10 | 0 | 0 |
| Grok | 10 | 0 | 0 |
Method and provenance
- The monitored entity was ClawSignal's owned production project.
- The prompt set contained 10 service-intent queries.
- Each prompt was checked on ChatGPT, Claude, Gemini, and Grok.
- Appearance and citation fields were read from production scan records.
- No customer data, estimated lift, or synthetic success result was added.
Source tables: ai_scan_runs, ai_visibility_checks, and ai_share_of_voice. The July 19 aggregate row is stored as partial even though its adapter counters record four attempted, four succeeded, and zero failed. We preserve that discrepancy as a data-quality limitation instead of silently relabeling the run.
What this proves, and what it does not
Supported
- The four-assistant scan path produced stored observations.
- The system retained a zero-visibility result.
- The baseline is specific enough to repeat after changes.
Not yet supported
- No causal lift can be inferred from a baseline.
- No customer outcome is represented here.
- No claim is made that one scan predicts future recommendations.
Start with a result you can repeat
Run your own scan, keep the first result intact, and measure changes against it.