Site Photos as Evidence: What to Capture and How to Keep It

Last updated

Photos are the cheapest insurance you’ll ever buy

A decent photo costs you ten seconds. It costs nothing in materials, doesn’t need anyone’s sign-off, and doesn’t rely on anyone else remembering the job the way you do. Compare that to what it costs when you don’t have one: a variation claim knocked back because you can’t prove what was there before you started, a progress claim disputed because nobody can confirm the stage was actually complete, or a defect callback two years after handover where it’s your word against a building owner’s.

Every trade already knows this in theory. Most still end up short a photo at the exact moment they need it, because “I’ll grab a photo” quietly becomes “I meant to grab a photo” on a day when the crane’s waiting and the supervisor’s chasing you for something else. The fix isn’t taking more photos. It’s taking the right ones, at the right moment, in a way you can actually find again.

What to photograph, and when

The rule that matters most: shoot before it gets covered up. Once a flashing’s behind cladding or a seal’s behind a reveal, the only proof it was ever done right is whatever you captured before it disappeared. A few points worth building into habit:

  • Before you start — existing damage and site condition. Photograph the opening, the surrounding finishes, and anything already damaged before your crew touches it. This is the single most argued-about gap on any job — someone finds a scratched sill or a cracked render six weeks later and it’s suddenly your problem, unless you’ve got a dated photo proving it was already there.
  • Delivered materials. Frames, glass, hardware — photograph what turns up on site against the delivery docket, especially if anything looks off. Damaged stock, wrong sizes, or short deliveries are much easier to argue with a photo than a memory.
  • Each stage, as it’s completed — not at the end of the day. Flashings and seals before they’re covered, fixings before linings go on, packers and shims in place before they’re hidden. If a builder or certifier would ever want to see it and won’t be able to once the next trade’s in, that’s your cue to shoot it now.
  • The finished opening, every stage complete. One clean set of photos showing the job done, matching whatever your inspection or sign-off process expects for that opening.
  • Anything unusual. Out-of-tolerance openings, site instructions given verbally, on-the-spot design changes — anything that isn’t standard scope. These are the photos that end up mattering most in a variation dispute, because they’re the ones nobody thought to write down at the time.

If you’re already running a staged sign-off process on installation — the kind of checkpoint structure covered in our guide to building an ITP for window and facade installation — photographing at each stage isn’t an extra job, it’s just part of marking the stage complete.

What makes a photo actually usable as evidence

Not every photo holds up. A photo that’s genuinely useful later — in a claim, a dispute, or a defects argument — usually has three things going for it:

  1. A timestamp you can see, not just metadata you have to explain. A phone’s EXIF data is fine until you need to forward the photo to someone, at which point most apps strip it out and you’re left arguing about when it was taken. A visible date stamp on the image itself survives being screenshotted, forwarded, and printed for a hearing.
  2. Context that shows where it is. A tight close-up of a flashing lap proves the flashing exists. It doesn’t prove which opening, which level, or which building. Pair every detail shot with a wide shot that shows the surrounding structure, or at minimum a way to identify the location — a marked-up opening number, a level sign, something in frame that ties it to a spot on the drawings.
  3. Both a wide shot and a detail shot. The wide shot proves what and where. The detail shot proves it was done right. One without the other leaves a gap someone else gets to fill with their own version of events.

A photo missing all three isn’t worthless — it’s still better than nothing — but it’s the kind of photo that leaves a question open, because it raises exactly the question it should have answered.

The WhatsApp problem: the photos exist, nobody can find them

Here’s the scenario that plays out on almost every job eventually. A defect turns up eighteen months after handover — a leak, a seal failure, a complaint from the building owner. The builder wants proof the stage was checked and signed off before it got covered up. You know you took the photo. You’re almost certain of it.

Then you go looking. It’s somewhere in a WhatsApp thread with forty other jobs mixed in, or buried in a camera roll with twelve thousand other photos, sorted by nothing more useful than the date you happened to take it. You know it’s Level 6, but you don’t know if it’s in the batch from March or the batch from April, and the site super who’d remember left the company eight months ago. Three hours later you’ve either found it, or you’re writing an email explaining why you can’t.

The photo did its job the moment you took it. It failed the moment nobody could retrieve it. That’s the actual failure mode of site photo documentation — it’s almost never “we didn’t take the photo.” It’s “we took the photo and then lost it in six months of camera roll.” A photo that can’t be found when it’s needed provides exactly as much protection as a photo that was never taken.

Organise by project, level, and opening — not by the date you happened to shoot it

Sorting photos by date feels natural because that’s how your camera roll already works. It’s also useless the moment you need to answer “show me Level 6, Opening 4, Stage 3” instead of “show me everything from the 14th of March.”

The fix is to organise around the structure of the job, not the calendar:

  • By project first. Obvious, but worth stating — mixing photos from two sites in one stream is how the wrong photo ends up in the wrong claim.
  • By level or zone within the project. Most disputes and claims get raised at level or zone granularity, so that’s the level your photo library needs to answer to.
  • By opening or item within the level. The detail that actually matters in a dispute is almost always “this specific opening, this specific stage” — not “somewhere on Level 6.”
  • By stage within the opening. So you can show progression — before, during, complete — without scrolling through everything else on the level to find it.

Structured this way, “find the photo for Level 6, Opening 4, flashing stage” takes seconds instead of an afternoon. Unstructured, in a shared chat thread or a phone that’s changed hands twice since the job started, it’s a lucky dip.

Where the photos actually pay off

This is the part that makes the ten seconds worth it. Photo evidence backs up three different fights, usually months or years apart from when the photo was taken:

  • Progress claims. A claim that says “Level 6 fixing stage complete” is a lot harder to knock back when it comes with a photo showing exactly that, dated the day you’re claiming for. Assessors and head contractors push back hardest on claims they can’t verify — a dated photo gives the assessor something to check.
  • Variation claims. If you had to deviate from the drawn scope — an out-of-tolerance opening, a site instruction given on the fly, extra work nobody wrote up — a photo taken at the time is worth more than any explanation given after the fact. We’ve covered this properly in how to document variations so you actually get paid, but the short version is the same principle as everywhere else on this page: the record made at the time beats the memory recalled later, every time.
  • Defects-liability defence. This is the one that catches people out, because it can be years after you’ve moved on to other jobs. A leak, a failed seal, a complaint traced back to your scope — if you can produce a photo showing the flashing, the seal, or the fixing was done correctly before it got covered up, you’ve got the evidence on hand. If you can’t, it’s your word against whoever’s making the claim, and you’re the one with more to lose.

None of this needs to be complicated. It needs to happen consistently, and it needs to be findable later — which is the part almost every crew gets wrong, not because they’re careless, but because nobody built a system for it beyond “we’ll find it in the group chat.”

Where WNDW fits

WNDW ties every photo to the opening and stage it belongs to at the moment your foreman marks that stage complete — no separate upload step, no sorting through a camera roll later. Two years on, “show me Level 6, Opening 4, flashing stage” is a search, not an afternoon of scrolling. See how it works on the foreman page.

FAQ

How many photos do I actually need per opening or stage?

Enough to prove the work happened and prove it was done right — usually a wide shot for context and one or two detail shots of the thing someone might argue about later, like a flashing lap or a fixing pattern. More isn't better if nobody can find them again.

Do timestamps on photos actually matter if a dispute goes to adjudication?

A visible date and some way to tie the photo to a location matters more than the file's metadata, because metadata is easy to fake or lose (screenshotting a photo strips it) and hard to explain to someone who isn't technical. A photo that shows the date and the location on its face is far more useful than one where you're asking someone to trust the EXIF data.

Should I photograph work that's already covered up if I missed it at the time?

If it's genuinely covered and you can't get access without undoing someone else's finished work, you're stuck relying on whatever record exists — which is exactly why shooting before cover-up matters so much the first time. Don't fabricate a photo of covered work; it won't hold up and it's not worth the risk.

Who owns the photos — me or the builder?

That depends on your subcontract, but in practice it rarely gets tested because nobody goes looking until there's a dispute. Keep your own copies regardless of what the contract says about ownership — you don't want your only record sitting on someone else's server when you need it.

Is a phone camera roll good enough, or do I need a proper system?

A camera roll works fine for the first few weeks. It stops working the moment you're running multiple levels, multiple openings, and more than a few months of history — that's when "which photo was level 6 opening 12" turns into an afternoon you don't get paid for.

See it in WNDW

Put this into practice on your next project.

See how it works