The Internal Communications Proof Audit.
A five-part review of the evidence that exists after an important employee communication is published. It measures communication proof, not employee comprehension. Run it on a communication you have already sent; where you cannot answer, write down that you cannot. The unanswered question is the finding.
Nothing to download and nothing to fill in first. No email gate.
Each part depends on the one before it.
How to run it.
Audience, reach, acknowledgement, exceptions, evidence.
Who needed to receive this communication?
Proof begins with the denominator. Until you can say precisely who a communication was for, nothing that follows can be measured against anything. This is where most reviews come apart: the recipient list was assembled by hand, under time pressure, from a source nobody has checked in months.
What to write down
- The roles, locations and groups that needed to receive it.
- Anyone deliberately left out, and who approved leaving them out.
- The system the recipient list came from, and when it was last reconciled with your people records.
- Whether contractors, part-time staff and people who joined that week were in scope.
What this part does not prove
A correct list is a starting condition, not evidence. It describes who should have received the communication, not who did.
A 300-person logistics company publishes a change to its overtime policy to "all staff". The list came from a payroll export taken four months earlier: 18 people who joined since then were never in scope, and 6 who had already left still were.
What evidence exists that it reached them?
This is the part most often overstated. A communication that left your system is not a communication that arrived, and one that arrived is not one that anybody looked at. Separate the two kinds of signal, then record which of the two you actually hold.
What to write down
- Which channels carried the communication, and what each channel records.
- Delivery signals you hold: an accepted address, a queued notification, a bounce, a failed send.
- View signals you hold, and the inference each one rests on.
- What is recorded for a person with no work address, no device, or no access at all.
What this part does not prove
Delivery signals describe the transport, not the person. A view signal is an inference drawn by software; it is not a statement made by a recipient.
The same policy change goes out by email and as an in-app notification. The mail system records that 282 addresses accepted it. The in-app channel records that 191 people opened their notification list that week. Neither figure is the number of people who can tell you what changed.
Who confirmed it, explicitly?
An acknowledgement is the only signal on this list that comes from the recipient rather than from the infrastructure. A person took an action that says, in effect, "I have this." Record the action, the person and the time, then stop, because that is the whole of what it says.
What to write down
- The exact action a recipient took, described in their words as well as yours.
- The timestamp held against each confirmation, and the time zone it is stored in.
- Who is allowed to request a confirmation, and who is allowed to see the result.
- How the record is corrected when somebody confirms on another person's behalf or another person's device.
What this part does not prove
An acknowledgement establishes that a named person took a named action at a named time. It does not establish that they read the communication, understood it, or agreed with it.
By the end of the week, 214 of the 282 recipients have confirmed the overtime change. That is 214 confirmations. It is not 214 people who can correctly state the new rule.
Who is still outstanding, and why?
The outstanding list is where a review earns its keep. A single number, still pending: 68, hides four different problems that need four different responses, and only one of them is a nudge.
What to write down
- Invalid, missing or out-of-date contact details.
- People on leave, travelling, or away from work during the window.
- Access barriers: no device, no login, no membership of the space where it was published.
- Approved exclusions, with the name of the person who approved each one.
What this part does not prove
An exception is a person, not a rounding error. Every name left outstanding is somebody for whom the communication has not been shown to have landed.
Of the 68 outstanding, 9 addresses bounced, 21 people were on leave, 3 had never been given a login, and 35 simply have not confirmed yet. Only the last group is a chasing problem.
Can leadership see the result without rebuilding it by hand?
The last question is operational rather than philosophical. If answering "did this land?" takes one person two days and four exports, you do not have evidence. You have the raw material for evidence, plus a standing cost every time somebody asks.
What to write down
- Where the record lives, and who can reach it without asking someone for help.
- How long it is retained, and what happens to it when the person leaves.
- Whether it can be exported in a form a reviewer outside the tool can read.
- What is written down when a record is changed or removed, and who is able to do that.
What this part does not prove
Being able to retrieve a record is not the same as being able to rely on it. Ask each vendor, in writing, which write protections apply to that record and which of them have been independently verified.
Asked in a Monday leadership meeting how the overtime change went, the communications lead needs a mail export, a notification log, a spreadsheet of confirmations and a list of leavers. The answer arrives on Wednesday, and by then it is out of date.
Six words that get used interchangeably.
They describe six different states. The distance between the first and the last is the whole subject of this review.
An acknowledgement does not establish comprehension.
An acknowledgement records that a named person took a named action at a named time. That is useful, and it is a stronger record than an inference drawn by a channel, but it is not a measure of understanding, and no communication tool measures understanding. Where comprehension is what matters, a safety procedure or a regulated change, pair the record with something that tests it: a question set, a sign-off that restates the content, a conversation with a supervisor.
This review is not a certification and it is not a compliance audit. It issues no score, no badge and no assurance. It is a structured way to establish what you can currently show, and to name the gaps you would want closed before somebody outside your team asks.
- Part three: the acknowledgement, by name and time, on an announcement (Command and HQ).
- Part four: the people still outstanding, by department, with one reminder to all of them.
- Part five: required reading with a due date, late marked late, and an evidence pack as a printable report and a CSV with every download recorded (HQ).
- Part one: the recipient list and who approved the exclusions.
- Part two: the channels you use and what each records.
- Comprehension: a three-question quiz can be attached on Command and HQ; it tests recall, not understanding, and we do not describe it as more.
The worksheet.
Print this page: the worksheet prints without navigation or calls to action and leaves room to write. Where you cannot answer, say so on the page.
Who needed to receive this communication?
What evidence exists that it reached them?
Who confirmed it, explicitly?
Who is still outstanding, and why?
Can leadership see the result without rebuilding it by hand?
Your answers stay in this browser. They are not sent to Kayden Connect and they are not stored on our side. Clearing your site data removes them.
Run the review on a real communication.
If you would rather hold parts three, four and five as a record than assemble them from a spreadsheet afterwards, ask for an early-access invitation and tell us which communication you would test it on.