Why internal benchmarks come first
An external benchmark answers the question “How are others doing?” – by their definitions, in their market, at their time. An internal benchmark answers the question “How are we doing compared to ourselves?” – same definition, same segment, just a different period. The second question is often the more useful one, because it leads to action.
Peak Atlas currently publishes no external market benchmarks, because we have no data basis that meets the criteria below. What we offer is the methodology – and tools for collecting your own values cleanly.
Seven details every comparison value needs
- Definition
- Numerator, denominator, source. “Closing rate” without a denominator is not a definition.
- Population
- Which cases count? All inquiries? Only those via the form? Only new customers?
- Period
- Start period of the cohort and cutoff date of the measurement. “Q1” isn’t enough; “inquiries January 1 to March 31, measured on June 30” is.
- Segment
- Industry, customer size, product, region, channel. A value across all segments usually hides more than it shows.
- Data quality
- Completeness (how many records have the field filled?), origin (system or manual entry?), known gaps.
- Median and range
- The median is robust against outliers; the range (smallest and largest value, or the 25th–75th percentile) shows how wide the distribution is. An average alone hides both.
- Comparability
- Two values are comparable if the six details above match – except for the period. As soon as another one differs, it’s not a comparison but a side-by-side with caveats.
Four steps to your own benchmark
- 01Define the metric: fill in the definition sheet for each metric – numerator, denominator, period, segment, source, type (activity or impact), owner. What you can’t fill in, you aren’t measuring yet.
- 02Form cohorts: choose a cohort length (month or quarter) and a fixed cutoff interval (e.g., 90 days after the cohort ends). From now on, every cohort is measured the same way.
- 03Collect first, then compare: after three to four cohorts you have a first median and range. Before that, every comparison is noise.
- 04Separate segments when they differ: if two segments consistently show different values, they get their own benchmarks. If not, they stay together – smaller groups fluctuate more.
Common mistakes
- Changing the definition midway and still using the old values.
- Using the mean instead of the median and reading a single large order as a trend.
- Measuring cohorts too early – before the decisions have been made – and then comparing them against a mature cohort.
- Setting an external value (“industry average”) without a definition as a target.
- Leaving out data quality: 60% filled fields make no benchmark, but a pointer to the next data task.
The KPI definition sheet
The definition sheet below is an empty CSV file with the columns metric, type, numerator, denominator, period, segment, source, data quality, owner and note. It deliberately contains no prefilled values – your definitions are yours. There is also the benchmark capture sheet, where you enter the measured values per cohort.
Downloads
To take away
Markdown opens in any text editor or note-taking tool. CSV files are semicolon-separated and open directly in Excel, Numbers, or LibreOffice.
