The difference between a norm, computed from the company's own customers of the same kind where the relationship is full, and a benchmark, a figure from outside describing an average company, why every gap on this site is against a norm, the three things a benchmark cannot do, be defended to a rep, be applied to a specific customer, or be versioned with the base, the two places an external figure is allowed, as a wallet fallback and as context, and the rule that it is always labelled.
Every analytics vendor offers benchmarks and every sales team wants to know how it compares. The gaps on this site are never against a benchmark, and this guide says why: what a norm is, what a benchmark cannot do, and the two places an external figure is allowed.
| Norm | Benchmark | |
|---|---|---|
| Source | The company's own customers of the same kind, full relationship | A survey, an index, a vendor's aggregate |
| Describes | What these customers demonstrably do with this company | What an average company's average customer does |
| Applies to | A specific customer in the cell | Nobody in particular |
| Defended by | Pointing at the population and the count | Citing the source |
| Versioned with | The base, quarterly | The publisher's schedule |
Be defended to a rep. "Customers like this one buy five lines from us" is a fact about the company. "Companies like ours see four lines per customer" is a fact about a survey.
Be applied to a specific customer. A benchmark of spend per employee describes the average. This customer, with its sector, size and history, is not the average, and the norm's cell is as close to it as data gets.
Be versioned with the base. The norm moves when the base does, on a stated cadence, with the movement on the movements page. A benchmark moves when its publisher updates it, and the gaps computed on it move without explanation.
| Place | Rule |
|---|---|
| Wallet fallback | When the customer will not state its spend and no cell in the base is large enough; scaled to the customer's size; labelled as an estimate; replaced by the norm as soon as a cell exists |
| Context | Beside the norm, never instead of it; to say whether the company's own base is above or below what is published; never producing a gap |
| Measure | Norm, mid manufacturing, 84 customers | Published benchmark | Used for gaps |
|---|---|---|---|
| Products per customer | 4 (75th pct) | 4.1 | The norm |
| Spend per site | $95,000 (median) | $120,000 | The norm |
| Share of wallet ceiling | 90% | not published | The norm |
The published spend per site is higher than the company's own. That is context: the company's full-relationship customers spend less per site than a survey's average, which may be the sector, the product or the price. It is not a gap at any customer.
Every gap is against a norm. Every external figure is labelled. No external figure produces a gap on its own.
Benchmark as the norm. The gap list is somebody else's customers.
Norm from too few. Six customers, a median that moves with one.
External figure unlabelled. An estimate read as a statement.
Context read as target. The 4.1 becomes a quota.
Covirage computes norms from the customer master and the ledger within stated cells, labels every external fallback, and shows the norm table with each report. The norm guide covers the method, and the norm table reading guide covers how to check one.
Because the rep's first question is 'says who, about whom', and the answer is a survey of other companies' customers. The norm's answer is 'our own customers of this size and sector, these eighty-four, last year'. The second ends the argument; the first starts one.
Then the cell is greyed and merged upward until it has enough members, and where even the merged cell is thin, an external figure scaled to the customer's size is the fallback, labelled as an estimate. It ranks; it does not claim.
As context, no: a company whose norm for products per customer is 2.8 in a category where a published figure is 4.1 has learned something about its own base. As a gap producer, yes: the 4.1 is not a target for any specific customer, and a list built on it will be wrong at the first call.