How do I know what got recorded actually got done?

The problem

An entry on a workflow comes back. Looks fine on the screen. The number is plausible, the tick is there, the text says what you would expect. You cannot tell from the entry alone whether the work behind it was actually done.

It can be a tyre pressure that has been a perfect 32 PSI every day for three weeks, written down at what looks acceptable rather than read off the gauge. It can be a twenty-step cleaning checklist that has been ticked through in ninety seconds rather than worked through. It can be a safety walk note that says "all fine" every shift, typed without anyone walking. It can be a condition choice that says "no issues" on a sixty-second inspection that wasn't really an inspection. Different entry types, same pattern: the worker entered what was expected without showing the work behind it.

The form check passes. The work check fails.

Why this matters

The whole point of an entry on a workflow is that it is evidence the work happened. If the entry alone does not prove the work, you have a record you cannot trust. An auditor reading it is reading a fiction. An insurer relying on it is relying on a hope. You looking for a trend are looking at a flat line that may or may not be real.

The cost is real. A tyre actually at 25 PSI on a summer afternoon while the log says 32 PSI is a blowout waiting to happen. A cleaning checklist ticked through in ninety seconds while the surfaces are still dirty means a customer complaint or an inspection failure waiting to happen. The lie in the entry masks the problem in the world. By the time you find out it was a lie, the problem has already cost you.

The fix is simple and takes about a minute to set up.

The fix: pair the input with a photo of the work

Add a photo input element directly after the entry that needs proof. Set both to mandatory. The worker has to capture the photo of the work at the moment they fill in the answer, before the next step opens.

The mechanic works regardless of input type. A number gets paired with a photo of the gauge or display. A tick gets paired with a photo of the area or item that was checked. A text note gets paired with a photo of what was observed. A choice gets paired with a photo that backs the choice up. In each case, the entry is the claim. The photo is the work being shown.

That is the whole setup. Once it is in, the worker cannot enter the answer without showing the work behind it. The temptation to flick through or guess what looks acceptable disappears, because the canvas will not let them.

A worked example: a tyre pressure check

Let us walk one through end to end. The setup is the same no matter what input type you are backing. We will use a pre-use tyre pressure check because the contrast is sharpest where the entry is a number that could plausibly be made up.

The check is one group with five elements. Four number inputs, one for each tyre: front-left, front-right, back-left, back-right. One photo input at the bottom that captures a photo of each gauge reading. Five steps in total, all mandatory.

What the driver sees on their phone before the shift: they walk around the vehicle, read the gauge on each tyre, and type the four numbers in. Then they capture a photo of each gauge with the reading visible. Four numbers, four photos, all taken at the moment.

What the depot manager sees in the dashboard afterwards: every reading on the check has a matching photo. The number is a claim. The photo is the work. Scrolling through the day's submissions becomes a fast visual check: does each gauge in the photos match the number written next to it?

The same pattern works anywhere the entry alone doesn't prove the work

A numeric reading is the example. The pattern works on any entry on any workflow where the worker could type, tick, or pick the answer without doing the work behind it.

For numeric entries:

• A fridge or freezer temperature check: the number plus a photo of the thermometer display.
• A pool or spa water chemistry log: the pH and chlorine readings plus a photo of the test strip or meter.
• A care home vital signs round: the blood pressure plus a photo of the cuff display.
• A till close at end of shift: the cash count plus a photo of the till total on the screen.

For checklist ticks:

• A cleaning checklist: the tick on each area plus a photo of the cleaned area.
• A safety walk: the tick on each zone plus a photo of the zone as you walked it.
• An equipment pre-use check: the tick on each item plus a photo of the item.

For text notes and choice answers:

• An incident report: the text description plus a photo of where it happened.
• A condition note: the choice (good / damaged / missing) plus a photo of what you saw.
• A handover note: the text plus a photo of the state being handed over.

The setup is the same in each case. Pick the entry where the worker could fill it in without doing the work. Add a photo input next to it. Mark both mandatory. Done.

Closing

The pairing costs the worker about five seconds at the moment of the work and gives you a record you can actually trust afterwards. The entry is the claim. The photo is the work being shown.

If you do not pair every input on every workflow, start with the entry where the worker is most likely to fill in what looks acceptable rather than do the work behind it. Usually that is the entry on a regulated, audited, or insurance-driven workflow. Lock that one to its photo evidence first, see how it feels, and roll it out from there.

→ Try this in the playground.