skills/weekly-health-check/examples/output.md

> Reference output. Numbers come from `evals/fixtures/ecommerce-healthy.mjs`,
> so this doubles as the expected shape for that eval case. Match this density
> and tone: verdict first, no padding, and nothing invented when nothing fired.

✅ **On track** — conversions and revenue grew slightly faster than traffic. No action this week.

| Metric | This week | Last week | Δ |
|---|---|---|---|
| Entrances | 9,850 | 9,610 | +2.5% |
| Conversion rate | 2.35% | 2.33% | +0.02 pp |
| Conversions | 231 | 224 | +3.1% |
| Revenue | €17,900 | €17,250 | +3.8% |
| AOV | €77.49 | €77.01 | +0.6% |

**Findings:** none. Every channel moved within ±3%, the widest campaign spread
(brand-es at 3.66% CR vs retarget-es at 1.09%) is unchanged from last week, and
bot share held at 6%. Padding this report with a "watch item" would be noise.

Attribution is last non-direct click, so these figures will not match GA4 or
your ad platform dashboards.

**Follow-up:** nothing was due for verification this week.

Want me to look at retarget-es? It has taken 645 entrances at 1.09% CR for
three weeks — the only line here that looks structurally weak rather than noisy.