Checklists for Field Technicians That Don't Get Faked

Checklists for field technicians work when three things are true: the list is under a dozen items, every item names a real failure it prevents, and at least one item demands a photo rather than a tick. Add a fourth and it holds: the right checklist attaches itself to the job automatically, by job type, at booking. A list that is forty items long, generic, tickable with a fingertip, and chosen by whoever remembers to grab the right form gets completed in the van at 16:55 from memory. Not because your technicians are dishonest, but because you designed a form that is faster to fake than to follow.

Why checklists for field technicians get faked: design, not discipline

A technician finishes a consumer unit change at 16:40. The paperwork is a two-page tick sheet with 38 lines, most of which do not apply, several unanswerable indoors, one asking for a value measured ninety minutes ago and never written down. The van is loaded. School pickup is at 17:15. Ticking all 38 boxes takes ninety seconds. Doing it properly takes twenty minutes spread across the day, and requires them to have known at 15:00 what the form would ask at 16:40. The incentive is arithmetic, not character.

Now change one variable. Make the list eight items, make item four require a photograph of the labelled board, and make the job impossible to close without it. The ninety-second fake no longer exists, because you cannot photograph a board you are not standing in front of. You do not police honesty. You remove the option to skip.

Rule one: under a dozen items, and mean it

Twelve is not a magic number from a study. It is a ceiling you can derive yourself. A technician runs five jobs a day. A 12-item checklist is 60 interactions, each costing five to eight seconds done honestly on site: five to eight minutes a day, survivable, so it gets done. A 40-item checklist is 200 interactions and over twenty minutes of form-filling. Nobody spends that. They batch it, and batching is faking. The list failed because it was long, not because it was wrong.

Aviation, and Atul Gawande's The Checklist Manifesto which carried the idea into surgery, land on the same range: five to nine items is the sweet spot. Under a dozen is the limit. So what happens to the other 28? Three options, in order of preference:

• Delete them. Most tick sheets carry items added after one bad job years ago that nobody has audited since. If an item has never once been answered "no", it is not a check. It is decoration.
• Move them upstream. "Confirm correct cable size" is not a site check. It is a design decision that belongs in the quote, made at a desk with the drawings.
• Split by job type. Twenty of those 28 apply to one kind of job only. That is not one checklist with 40 items. It is four checklists with eight items each, plus routing.

Rule two: every item names the failure it prevents

Take any item and ask what goes wrong if it is skipped, and what that costs. If you cannot answer in one sentence with a number in it, delete the item. Generic items produce generic ticking. "Work area left safe" gets ticked reflexively. "Photo of the isolator with the lock-off fitted" either exists or it does not.

Every failure in that right-hand column carries a price: a return visit, an unpaid line, a disputed invoice, an insurance excess. That is the argument that lands with technicians, not "compliance" and not "the office needs it". The photo of the pre-existing crack is the one that saves them from being blamed.

The "no" has to be usable

An item that can only be answered "yes" is not a check. Each needs a fail path that does something: flags the job, notifies the office, opens an extra-work line. If ticking "no" only lengthens the technician's evening, they will tick "yes".

Rule three: at least one item must require a photo

The photo is the anchor. It makes the rest of the list credible, because it proves presence. Taken through the camera in the job app it carries a time and a place, and cannot be produced from the driver's seat of a van in another postcode.

• Name the shot, not the subject. "Photo of the work" gets you a blurry ceiling. "Photo of the consumer unit, door open, all labels visible" gets you the shot you need.
• Capture it in the app, not the camera roll. A gallery upload could be from any job on any day. A photo taken inside the checklist item is tied to that item, that job, that moment.
• One to three photo items per list, not eight. Photo fatigue is real. Pick the shots that settle disputes: the before-state, the finished labelled result, and above all the thing about to be covered up. Cable runs before plastering, pipework before the floor, membrane before tiles. Once it is buried, the photograph is the only evidence it was done right.

One caution. Do not require photographs inside occupied rooms, of documents carrying identity numbers, or of anything you would not want to hand over in a dispute. In the EU and UK, site photos showing identifiable people are personal data under GDPR. Photograph the installation, not the household.

Rule four: attach the right checklist to the right job type automatically

This is the rule most firms skip, and it quietly destroys the other three. A perfect eight-item list returns nothing if the technician must know it exists, work out which of your eleven lists applies, and choose it. They will pick the first, the one they used last time, or none. The checklist should already be on the job when they open it, chosen by job type at booking. The technician never picks. The office never remembers. It is a routing rule, written once.

Three rules for routing:

1. Job type is required at booking. If it can be blank it will be blank, and routing has nothing to route on.
2. One job type, one default list. If two could apply, split the job type rather than ask a person to choose in a cold plant room.
3. The office may add a second list, never remove the default. Additions are judgement. Removals are how a default quietly dies.

This is hard on paper and easy in software. In OdinTask you build the checklist once, tie it to a job type, and it lands on the job card before the technician arrives, photo items gated so the job cannot be closed without them. It works offline, which matters: plant rooms and risers are where there is no signal.

Gate the completion, do not police the person

A checklist with no gate is a suggestion. The gate: the job cannot be marked complete, and the invoice cannot be raised, until the mandatory items carry answers and the mandatory photos exist. Two details make that survivable rather than infuriating.

• Warn but allow, with a reason. Hard blocks breed workarounds. Let a technician finish without the photo if they type a reason, then put every reason on the office desk next morning. If the same reason appears fifteen times, your checklist is wrong, not your technician. Reasons dry up inside a fortnight, because typing an excuse is slower than taking the photo.
• Never gate on connectivity. If the app blocks completion because it cannot reach a server, you have taught the team that the checklist is the enemy. Queue it and replay it.

The regulatory checklists you cannot shorten

Some lists are not yours to design. In Sweden, an electrical firm registered with Elsäkerhetsverket must run an egenkontrollprogram, a documented self-inspection programme whose content follows the regulations, not your preferences. In the UK, an Electrical Installation Certificate under BS 7671 carries a defined schedule of test results. In New Zealand, a Certificate of Compliance under the Electricity (Safety) Regulations 2010 is likewise prescribed. Gas, asbestos and work at height all bring statutory paperwork, and Arbetsmiljöverket (or the HSE, or WorkSafe) does not care that your form is long. Do not shorten those. Do two other things.

• Separate the statutory record from the workflow checklist. The certificate is a legal document produced at the end. The eight-item checklist puts the technician in a position to produce it honestly. Confusing the two is how firms end up with 40-item hybrids serving neither purpose.
• Pre-fill everything the system already knows. Address, customer, technician, date, job number, serials scanned earlier. A 30-field statutory form with 18 pre-filled is a 12-field form in the hand. Twelve is where we started.

Rolling it out without a mutiny

Do it in that order, one job type at a time, over a month: delete, then justify each survivor, then add the photo items, then wire the routing, then turn the gate on as warn-but-allow. Prove the routing works by booking a fake job of each type before anyone real touches it.

Then tell the team what you deleted before you tell them what you added. "We cut the board-change sheet from 38 lines to 9, and one is a photo" is a conversation you win. "We are introducing a new compliance process" is one you lose.

How to tell whether it is working

• Time between the answers and the completion timestamp. If the checks all land in a 90-second burst at the end of the day, you are still being faked. Honest lists are answered across the visit.
• Distance between where the photo was taken and the site. If those diverge, you have your answer without asking anyone.
• Callbacks per hundred jobs, by job type. The only number that pays the bill. If it has not moved in a quarter, your items check the wrong things. Go back to rule two.

Short list, real failures, one photo, automatic routing, a gate easier to satisfy than to dodge. Get those right and nobody has to be trusted, which is a better foundation than hoping.

If you want the routing and the photo gate without building it, OdinTask does this: a checklist builder, attachment by job type, mandatory photo items, offline capture, and a signed PDF at the end. Start a 14-day free trial, or see what is included on the pricing page.