Lorixen.
EN

Job-Site Photos Without Signal: Designing an Upload Queue That Survives a Dead Zone

Short answer: Photos taken on site should be saved to the phone first and sent later by a queue that the app owns, not by the screen the person is looking at. Each photo gets a local record, a retry rule and a visible status, so a plumber in a plant room can tap Done and leave. The upload then happens in the background when signal returns, and the office sees which photos have arrived and which have not.

The failure we are designing against

Picture a roofer in a rural area. They photograph the damaged ridge tiles, fill in the job form and tap Submit. The spinner turns, the signal drops, the app shows an error, and they drive away. The photos exist only on the phone, nobody knows, and the customer is told the report is "coming". Two days later someone asks where the evidence is.

A well-built queue prevents this by separating three questions: did the photo get taken, is it safely stored on the device, and has the server confirmed receipt?

A phased plan

Phase 1: Save locally first

When the shutter fires, the app writes the image file to its own storage and creates a small record: job ID, photo ID, time taken, file path, status. Only then does the screen say the photo is saved. Everything after this point can fail and be retried, because the original is safe.

Use a durable store, such as a small database, not memory. If the app is killed, the queue must still be there on reopening.

Phase 2: Shrink before sending

A modern phone camera produces files of several megabytes. On a weak connection that is the difference between a minute and an hour. Create an upload copy at a sensible size, perhaps a long edge of around 2000 pixels at moderate quality, and keep the original only if you truly need it. Ask your client what the photos are for: a damage claim needs more detail than a tick-box proof of completion.

Phase 3: Upload in the background

On Android, scheduled work through WorkManager can run uploads with a "network required" constraint, and Android handles retry and backoff. On iPhone, a background URL session lets the system continue transfers even after the app is left. In both cases the transfer is managed by the operating system, which is more reliable than a timer in the app.

Rules to give your developer:

  • Upload one photo at a time or a small number in parallel, not all at once
  • Retry with growing delays, for example 30 seconds, then a few minutes, then longer
  • Pause on mobile data if staff prefer, using a setting with a sensible default
  • Resume partial uploads where the platform supports it

Phase 4: Confirm and clean up

The server should reply with a receipt, and the app marks the photo as sent. Only after confirmation should the upload copy be removed. Keep a short history so the worker can see "Sent 14:32" for each photo.

Phase 5: Handle the awkward cases

Plan these before launch, not after a complaint:

  • Duplicate sends. If the network drops after the server received a photo but before the app heard back, the retry would upload it twice. Give each photo a unique ID created on the phone, and have the server ignore a second copy.
  • Job closed before upload. If the office closes the job while photos are pending, accept the late photos and attach them anyway, or show a clear message.
  • Storage running low. Warn staff when the phone has little space, since a full phone cannot take photos.
  • Logged out. Keep unsent photos when someone signs out, or block sign-out while uploads are pending. Never silently delete evidence.
  • Corrupt or zero-byte files. Check the file before queueing.

What the screen should say

People need a simple, honest status. Three states are usually enough:

Status Meaning for the worker
Saved on phone Safe, will send when there is signal
Sending In progress
Sent The office has it

Add a banner such as "4 photos waiting to send" on the job list, and let staff tap it to see detail. A failed state with a reason, for example "File too large", gives a way forward. A vague red icon does not.

Let staff leave the app without fear. If the phone is full of signal bars in the van park, the queue will do its work. If they need to be sure before they leave a site, give them a "Check all sent" button that shows the real count.

Specify these points when you commission the app

  • [ ] Photos saved on the device before any network call
  • [ ] A durable queue that survives app restarts and phone reboots
  • [ ] A resized upload copy, with the maximum dimension agreed
  • [ ] Background upload through the operating system's own tools
  • [ ] A unique ID per photo to prevent duplicates
  • [ ] Visible per-photo status and a count of pending items
  • [ ] Rules for retries, mobile data and low storage
  • [ ] Behaviour at sign-out and when the job is already closed
  • [ ] A way for the office to see "photos expected versus received" for each job

Test it in a way that mirrors the field

Do not test on office wifi. Put the phone in aeroplane mode, take five photos, close the app, reopen it, then turn the signal back on and watch. Repeat with a flaky connection, using the developer tools that throttle the network. Try a phone with low storage. Keep a short log of results the way you would for any acceptance test.

An illustrative result might read: "Five photos taken offline, app force-closed, phone restarted. After signal returned, all five sent within four minutes, no duplicates." That is what you want to see written down.

Costs and trade-offs

A good queue is not the cheapest build. It adds database work, background tasks and a status screen, and it adds testing time on real devices. But it protects the evidence your business relies on, and it saves the office from phone calls asking for photos to be sent again. If your staff rarely work without signal, a simpler approach may be enough, as long as it still saves locally first.

FAQ

Can the app upload only on wifi?

Yes, but staff in remote places may never reach wifi. A better default is any connection, with a setting to restrict to wifi for people on limited data plans.

How many photos can a queue hold?

That depends on the phone's free storage. Plan for a day's work, and warn when space is low.

Do we need to keep the full-size originals?

Only if the use case demands it. Smaller upload copies reduce cost and speed up sending.

What about video?

The same pattern works, but files are far larger, so set tighter limits and expect longer upload times.

If your team loses photos to bad signal, describe a typical site visit and we will sketch a queue that fits it.

Sources

WhatsAppBook a call