1. Metrics are language, not goals
Metric system·A causal chain from process metrics (reach, clicks, visits) to outcome metrics (orders, repeat purchase, profit). Process metrics explain outcome metrics; skip a link and you are guessing.
The trap is vanity metrics: impressions, followers, gross sales. The test is simple — if this number improves, do profit and repeat purchase follow? If not, it is decoration.
2. The funnel and the two ledgers
| Layer | Question | Typical metrics |
|---|---|---|
| Reach | Did the right people see it | Impressions, CPM, audience coverage |
| Interest | Did content filter the right ones | CTR, completion rate, branded search |
| Conversion | Do visits become orders | Conversion rate, CPO, order value |
| Retention | Is the money long-term | Repeat rate, LTV, member activity |
Keep two ledgers. The performance ledger tracks this month's ROI; the asset ledger tracks branded search, repeat purchase and reputation. Optimizing only the first one quietly raises next year's acquisition cost.
3. The four-step review
Compare against the goal
Put actuals next to targets and locate the gap in the funnel before debating causes.
Attribute the gap
Ask in order: did the market move, was the strategy wrong, or did execution slip. Reversing the order misdiagnoses everything.
Write conclusions as actions
Who does what, which metric should move, and by when. A review without actions is theater.
Verify next cycle
Watch only the metric you tried to change. Unverified change equals no change.
4. What AI changes
- 1Live dashboards turn weekly reporting into continuous monitoring, with anomaly alerts enabling mid-course correction.
- 2Attribution models turn credit fights into calculations, so budget allocation finally has evidence behind it.
- 3AI drafts review summaries, but accountability stays human. Models confuse correlation with causation; managers still sign the call.