Blog · Board and management reporting
The difference between a ratio weighted by count, accounts, deals, inquiries, and one weighted by value, revenue, notional, premium, why the two disagree whenever the large items behave differently from the small ones, the six measures where the difference is largest, hit ratio, coverage, win rate, share of wallet, renewal rate and retention, the rule that a report shows both and ranks by value, and the case where count is the honest one.
Every ratio can be computed two ways and the two disagree exactly when it matters: when the large items behave differently from the small ones. This guide sets out count against value weighting, the six measures where the gap is largest, the rule, and the cases where count is right.
| Count-weighted | Value-weighted | |
|---|---|---|
| Treats | Every item alike | Large items as larger |
| Answers | How many? | How much? |
| Hides | The large item that behaves differently | The many small items |
| Dominated by | The numerous | The big |
| Measure | Count version | Value version | Typical disagreement |
|---|---|---|---|
| Hit ratio | Trades ÷ inquiries | Notional traded ÷ notional inquired | Client takes small tickets, leaves large: 41% vs 15% |
| Coverage | Accounts touched ÷ assigned | Revenue of touched ÷ revenue assigned | Rep touches the tail: 70% vs 40% |
| Win rate | Won ÷ outcomes | Won value ÷ outcome value | Large deals lost: 28% vs 21% |
| Share of wallet, rolled up | Mean of customer shares | Σ sales ÷ Σ wallets | Small customers at extreme shares: 38% vs 30% |
| Renewal rate | Renewed ÷ due | Renewed premium ÷ premium due | Large renewals lost: 90% vs 84% |
| Retention | Customers retained ÷ start | Revenue retained ÷ start revenue | Large customers left: 93% vs 89% |
Every ratio states its weighting. Where count and value differ by more than a stated amount, both are shown. The ranking column is value.
| Measure | Why count |
|---|---|
| Count to half | The question is how many relationships |
| Facilities compliant | Each facility is a decision |
| Reps above the data quality floor | Each rep is a person |
| Controls tested on cadence | Each control is an obligation |
Stated as count, on purpose.
| Rep | Accounts | Touched | Coverage by count | Revenue assigned | Revenue touched | Coverage by value |
|---|---|---|---|---|---|---|
| R-04 | 62 | 43 | 69% | $4.0m | $1.6m | 40% |
| R-11 | 44 | 30 | 68% | $3.6m | $3.2m | 89% |
Same count coverage. R-04 touched the tail and not the top; R-11 the reverse. The value column is the finding, and the count column would have called them equal.
| Region roll-up | Method | Result |
|---|---|---|
| Average of rep percentages | Count of reps | 71% |
| Σ touched ÷ Σ assigned | Accounts | 63% |
| Σ revenue touched ÷ Σ revenue assigned | Value | 58% |
Three numbers for the same region. The first is wrong; the second and third are the two honest ones, and the report shows both.
Weighting unstated. Two people, two ratios, one argument.
Count as the headline. The large item that left, invisible.
Averaged across reps. The eighteen-account rep offsets the sixty-two.
Value where count is the question. Count to half by revenue is not a count.
Covirage states the weighting on every ratio, shows both where they differ, and never averages across reps. The size-weighted hit ratio guide covers the sharpest case, and the renewal, retention and churn guide covers three measures on both bases.
Value, because it is what is at stake, with count beside it. A renewal rate of 90 percent by count and 84 by premium says the renewals lost were the larger ones, and the second number is the one the board should see first.
When the items are alike, or when the question is about breadth: how many relationships carry the business, how many facilities are compliant, how many reps are above the floor. Count to half is a count measure by design. The report says which.
An average of rep percentages is a count-weighted ratio of reps, and it is almost always wrong: the rep with eighteen accounts at 94 percent offsets the one with sixty-two at 53. Roll-ups are weighted by the underlying accounts or value, never averaged across reps.