This week I moved a wellness clinic off software it is about to cancel and onto a new system. One of the jobs was plain manual data entry: 22 health-screening questions that have to match the old form word for word, because clients answer them before a treatment.
I started clicking them in by hand. Roughly thirty attempts later I had one question — not one section, one question — and it still had not saved.
That is the part worth writing about. Not the fix. The thirty attempts I spent before I admitted the route was wrong.
The half hour that should have been a warning
Earlier the same day I had made two folders to file those questions into. The first took about fifteen attempts. The second one simply refused to open its dialog three times in a row.
I had the evidence in front of me and I kept going anyway. Two folders in, zero questions created, with 22 questions, six pipeline stages and six forms still to build.
Doing the arithmetic afterwards, the route I was on came to several hundred interactions. Nobody sits down and agrees to that. You agree to it one click at a time.
What does one entry actually cost you?
Most owners never measure this. We estimate the whole job — "an afternoon" — and then discover on Thursday that it was three afternoons.
The honest version is smaller and much easier to answer. Do one. Just one. Then ask:
- How long did the first one take, from start to actually saved?
- Did it work the first time, or did I have to redo it?
- Is the second one likely to be faster, or exactly the same?
If the first item took five minutes and there are forty of them, you have not booked "an afternoon". You have booked most of a working day, and that is before anything goes wrong.
A three-step test before you commit to the whole list
Time one
Do a single item properly and note what it cost. Not the version you imagine — the one you actually just did.
Multiply it honestly
Multiply by the real count, then add a third for the ones that fight back. If the number makes you wince, that is useful information, not a reason to look away.
Look for another door
Ask what else can put this data in. Most business software has more than one entrance: a bulk import, a spreadsheet upload, a template you can copy, a supplier who will do it for you, or a person who does this daily and charges less than your own hour.
In our case there was another door. Instead of clicking each question into place, we sent the whole list to the system at once. All 22 went in together and every one came back confirmed.
The real risk of manual data entry isn't the hours
Slowness is annoying. It is not the thing that hurts you.
These were medical screening questions — bleeding disorders, blood thinners, pregnancy — that must read exactly the same as the original. When you are on your thirtieth attempt at a fiddly screen, tired and copying wording between two windows, that is precisely when a word goes missing.
Nobody notices a missing word. There is no error message for it. You find out months later when someone answers a question you did not actually ask.
The cost of grinding is not the hours. It is the mistakes those hours produce in the exact material that has to be right.
Read it back before you call it done
Whichever route you use, finish by checking what actually landed. We pulled the list of all 22 questions back out of the new system and compared names, types and answer options against the old form.
That takes minutes and it is the only step that turns "I entered it" into "it is correct". Entering data and verifying data are two different jobs, and most people only ever do the first one.
Not everything had a shortcut, by the way. The forms themselves still have to be built by hand. Knowing which parts genuinely need your fingers — and which parts you were doing by hand out of habit — is the whole skill.
Do one, time it, then decide. The cheapest hour you will spend this month is the one where you find out what the job really costs before you commit to all of it.