skills/channel-mix-optimizer/examples/output.md

> Reference output. Numbers from `evals/fixtures/ecommerce-healthy.mjs` (30d).
> This example deliberately shows the *honest negative*: the RPE gap is below
> the threshold, so the answer is "do not reallocate yet", not a made-up shift.

**Paid channel scorecard, 30 days**

| Channel | Entrances | CR | AOV | RPE | vs site avg |
|---|---|---|---|---|---|
| Paid Search | 9,800 | 2.41% | €80.08 | €1.93 | +6% |
| Paid Social | 4,100 | 2.15% | €72.73 | €1.56 | −14% |

Site weighted average RPE: €1.82.

**No reallocation recommended this month.** Paid Search returns €1.93 per
entrance against Paid Social's €1.56 — a ratio of 1.24×, below the 2× gap that
would justify moving budget on this evidence alone. At these levels the
difference is well inside what a CPC difference between the two platforms could
explain in either direction.

**What would change the answer.** Pull CPC from both platforms and compare
`CPC ÷ CR × 1` against AOV on each. If Paid Social's CPC is more than 24% below
Paid Search's, it is already the better buy and the RPE gap is misleading.
Sealmetrics has no spend data, so this is the one number I cannot get for you.

**Scale candidates:** none. `brand-es` has the best economics (3.64% CR,
€83.10 AOV, RPE €3.03) but it is a brand campaign — its ceiling is your existing
demand, not your budget.

**Cut candidates:** none at the threshold. `retarget-es` is the weakest line
(2,700 entrances, 1.22% CR, RPE €1.07) but it has not run 30 days at that level
yet. Watch it; do not cut on one month.

Attribution is last non-direct click, which undervalues Paid Social if you use
it for awareness. Do not cut it on this table alone.