The metrics that reassure and do nothing
Almost every internal communication report that reaches a board is full of numbers that measure the department's effort, not the effect on the workforce.
- Number of posts per month. Measures how hard the person publishing is working.
- Impressions or views. Counts the times something appeared on a screen, not the people who found out.
- Likes and reactions. They measure goodwill, and mostly they measure the same twenty people every time.
- Email open rate. Useful if your whole workforce uses email. If it does not, it measures the office and nothing else.
- Total visits to the intranet. One person going in forty times covers for forty who never go in at all.
None of them is false. The problem is that none of them answers the only question a managing director asks: did people find out or not?
The measure that does matter: coverage
Coverage is the percentage of people in a group who have seen a particular piece of content. Distinct people over people in the group. That is all.
It is a boring number, which is why it works. If the Derby plant has 210 people and 128 of them have seen the notice about the summer shutdown, coverage is 61 %. With that figure you can already make a decision: either distribution at Derby is failing, or the notice was of no interest, or there is a shift that nothing reaches.
Three rules to make coverage useful. One: always broken down, by site, by department and by shift. The company average hides precisely the thing that needs fixing. That is the rule of the trade, not a description of any particular tool: Blastack, for one, breaks it down by department and stops there. Two: sorted lowest first, so you look at the worst group before anything else. Three: compared against itself month by month, not against another company.
Six usable measures, and the decision each one supports
| Measure | How it is worked out | What decision it supports |
|---|---|---|
| Coverage per announcement | Distinct people who have seen it over people in the target group | Repeating the notice on the site that came up short |
| Coverage of the worst site | The lowest value in the list above | Where you need a correspondent or a fresh round of sign-up codes |
| Workforce active in 30 days | People who have gone in at least once, over total headcount | Whether the channel is alive or only the usual faces look at it |
| Outstanding read receipts | Documents needing a receipt that are still unconfirmed, by role | Who to chase before an audit or an HSE inspection |
| Survey response rate | Responses over people invited, by site | Whether the engagement survey represents anything, or only the offices |
| Time to 50 % coverage | Hours from publication until half the group has seen it | Tuning the time you publish to the shift cycle |
Six is a ceiling, not a target. With the first two and the fourth you could run a whole internal communications function for a year.
Where the line is: aggregated yes, named only when it has to be
This is where an internal communications tool turns, without meaning to, into a monitoring system. The difference fits in one sentence: an aggregated figure says what percentage of a group has seen something; a named figure says that Jane Smith saw it at 11.14pm.
For running internal communication, aggregated is enough. Named data is only justified when the company itself needs the proof, a safety procedure or a policy that has to be evidenced, and at that point it stops being a metric and becomes a document record with a purpose of its own.
Sets out the principles of purpose limitation and data minimisation: personal data must be adequate, relevant and limited to what is necessary for the purposes it is processed for.
Require people to be told about the processing of their data, including the purposes and the lawful basis, at the point the data is obtained.
Supplements the UK GDPR in the employment context. The ICO's employment practices guidance on monitoring workers sets out what the regulator expects: be clear about the purpose, tell workers what is being monitored, and carry out a data protection impact assessment where monitoring is likely to result in a high risk.
This is not legal advice. What your company specifically has to do, whether an impact assessment is needed, which lawful basis applies, what you tell people and how, is for your data protection officer and your employment solicitor to settle.
What to take to the reps before you start measuring
The question you are going to be asked is 'is this here to keep tabs on us?', and it will arrive in the first five minutes. It is worth turning up with the answer already written down.
01
What is measured, and at what grain
Show them the screen, not a description of it. If what is there is a percentage by department, it takes two seconds to see there is no list of who read what.
02
What is named, and why
Read receipts carry a name, a date and a time, and that is deliberate: it is the proof that replaces the signature sheet. Saying so up front is worth more than justifying it later.
03
Which optional modules you are switching on
Access and activity logs, which device each person signs in from, which sections they visit, are usually optional. If you do not need them, do not switch them on, and put it on the record that they are off.
04
How long it is kept
A specific period, in months, and what happens afterwards. 'Indefinitely' is not an answer you can defend.
05
What it will not be used for
In writing: that this data does not go into performance reviews or into anybody's file. That is the sentence that unblocks the meeting.
The special case: engagement surveys
An anonymous engagement survey is only worth running if the anonymity is real, and 'real' means that not even an administrator with database access can reconstruct who answered what.
It is worth asking how it is built rather than settling for 'it's anonymous'. Two specific questions: is the timestamp of each response stored to the second? And if so, can it be cross-referenced with the session log? If both answers are yes, the survey is not anonymous whatever the screen says.
The price of genuine anonymity is that you lose the departmental breakdown when the group is small. It is a price worth paying: a breakdown for a department of four people is not anonymous however many warnings you put around it.
How to present this upstairs
One page. Three numbers and a sentence for each of them.
- Average coverage of the quarter's official announcements, with the worst site next to it.
- Percentage of the workforce that has gone in at least once this month.
- Outstanding read receipts, if you have documents that legally need a record.
Underneath, two lines: what you are going to do about the worst site, and by when. An internal communication report that does not end in a decision is a presentation, and presentations stop being asked for.