Going paperless without losing anything

Verify before you destroy. The failure in going paperless is almost never a bad scan — it’s throwing away the original before confirming the copy is legible, complete and findable. Everything else about the transition is straightforward.

Which makes the whole project a sequencing problem rather than a technology one.

The order that matters

Capture, verify, then destroy. Three steps, and the middle one is the one people compress.

Verification is not “the app said it worked”. It’s looking at the image and confirming the fields you’ll need later are readable — the total, the date, the vendor — and that the whole document is in frame with no cropped edges.

For a receipt you’re capturing at a till, that’s a two-second glance. For a backlog you’re working through in one sitting, it’s worth being more systematic, because a batch mistake repeats across everything you did that afternoon.

Deal with the backlog separately from the flow

Two different problems, and mixing them is why paperless projects stall.

The flow is receipts arriving from now on. Fixing this is a habit change: capture at the point of spend, one route, verified before the paper goes. It’s the part with ongoing value and it can start today.

The backlog is the box, the drawer, the bag of accumulated slips. It’s a finite project with a definite end, and it doesn’t need doing before the flow is fixed. In fact doing it first is a common mistake — you spend a weekend on history while continuing to generate new paper.

Fix the flow first. Then work the backlog in sittings, oldest-and-most-important first, and accept that some of it isn’t worth capturing at all.

What in the backlog is worth capturing

Not everything. Be honest about it:

Worth it: anything supporting a tax position you might need to defend, warranty documentation for things you still own, anything relating to an ongoing dispute or contract, and records for periods still within your retention obligations.

Probably not: receipts for consumed items outside any retention window, duplicates of documents you already have digitally, and anything whose amount is small enough that its absence changes nothing.

A backlog triaged this way is usually a fraction of the volume it appeared to be.

The faded-paper problem

Thermal receipts fade, and fading is progressive and irreversible. In a backlog, the oldest slips are the ones most likely to be already unreadable.

So work oldest-first, and accept that some are gone. For a slip that’s partly legible, capture it anyway with the best light and contrast you can manage, and add a note recording what you can still read. A poor image plus a note beats nothing.

This is also the argument for not delaying the flow fix: every thermal receipt sitting in a drawer is depreciating.

Verification for a batch

When you’re processing many at once, a few checks catch the systematic errors:

Count. Know how many documents went in and how many records came out. A mismatch means something was skipped or doubled, and it’s the only check that catches a missing document.

Spot-check the images, not the extracted data. Pick a handful and look at the actual photos. Consistent problems — a crop that cuts the bottom line, glare in the same place — come from your setup and will affect everything in the batch.

Check retrievability, not just storage. Search for three documents you know are in there, by vendor and by amount. Finding them proves the whole chain worked; a file that exists and can’t be found isn’t much better than one that doesn’t.

Then destroy, in batches, once each batch has passed. Not before, and not all at the end either — batch verification keeps the exposure small.

Keep some paper

Going paperless doesn’t mean holding nothing. A small set of documents is worth keeping physically regardless: anything where an original signature or seal matters, anything you’d need to produce as a physical document, and anything legally required in original form.

That’s a short, stable list. Knowing what’s on it means the rest can go without hesitation.

Formats and durability

Whatever you capture into, the retained artefact needs to be readable without the software that created it. A widely supported image or document format, stored somewhere you control, exportable in bulk.

The test is the same one you’d apply to a backup: export everything, open a sample with a completely different application, and confirm you can tell what each document is. If that works, your records will outlive your choice of tool. If it doesn’t, you’ve moved a dependency rather than removed one.

And backups still apply. Paper’s one virtue was that it existed in exactly one place and you knew where. Digital copies need at least two locations, and at least one of them somewhere a mistake on your part can’t reach.

What you actually gain

Retrieval, mostly. A digital record is searchable by vendor, amount and date, which paper never was — the practical difference between finding a three-year-old invoice in a minute and not finding it at all.

Also: it stops degrading. A photograph of a thermal receipt taken today is exactly as legible in ten years. The paper won’t be.

Both of those are contingent on the copies being verified. Which is why the whole exercise comes back to sequencing: capture, verify, and only then destroy.