When a shared spreadsheet stops being good enough

A row is missing. The administrator is certain the referral was entered on Thursday, because she entered it. On Monday it is not there. Nobody deleted it on purpose, and nobody can prove anything either way, because the sheet keeps no record of what it used to say. The practice spends twenty minutes reconstructing it from the fax log, and then everybody gets on with their day.

That is how this usually starts. Not with the spreadsheet breaking, but with a small question it cannot answer.

A shared sheet is often the right tool. Four people, thirty referrals a month, everyone in the same room — a spreadsheet beats software, and anyone who says otherwise is selling. What changes is not the sheet’s quality but how much the practice has come to depend on it.

Three questions mark the line.

01What did this record say yesterday?

This is the audit trail question, and it usually arrives first, because it arrives as an argument.

A spreadsheet stores the current state of things. Someone types over a phone number, clears a cell to tidy it up, or sorts a column with a row half-selected, and the previous value is gone. Version history helps a little, but it answers “what did the file look like” rather than “who changed this record, when, and what was in it before” — and not at the speed a conversation moves.

Three places where this matters: a call attempt that may or may not have been logged, an authorisation number that was right and now is not, and a referral that is no longer there. The cost is not really the data. It is that staff start to distrust the sheet, and once they do, several of them quietly keep their own copies — a worse problem than the one they are solving.

02Who should be allowed to do that?

In a shared sheet, everyone who can read it can also change it or delete from it. There is no way to give the biller the authorisation columns and not the clinical notes, or to let a provider see the follow-up queue without being able to reorder it.

Most practices do not notice until someone new arrives. A temporary front-desk hire on her second day has the same access as the administrator who built the thing. Nobody intends that; the sheet has no way to express an intention about it.

Thinking about roles is useful even if you never buy anything. Write down who actually needs to do each of these: enter a new referral, change an authorisation, record a call attempt, see clinical notes, delete a record, and change the structure of the sheet itself. Most practices find four distinct answers — and find that one or two people currently hold all six, because nobody ever separated them.

03Is this record finished?

The third question is the one that quietly breaks reporting.

A spreadsheet will accept anything. A referral with no referring provider saves as happily as a complete one. A call attempt written as “4/8 LVM, will try again” is clear to the person who wrote it and uncountable by a computer, so a five-call rule cannot be verified without a human reading every cell. And the same referring office arrives as four different strings depending on who typed it, which makes every count you run wrong in a way you cannot see.

Two things fix this, neither of them exotic. Validation: a record cannot save until the fields that matter are filled in. And a controlled vocabulary: the referring office is chosen from a list rather than typed, so there is one spelling and counting becomes possible. The build described on our work page, for an independent practice here in Houston, took six weeks, and most of what it changed was this.

04Four checks to run on your own sheet

Do these before deciding anything. They take about an hour and they are free.

05Then decide — and sometimes decide no

If those checks come back clean — one spelling per office, changes you can account for, access that matches intention, nobody keeping a shadow copy — your spreadsheet is doing its job and you should leave it alone. That is a real outcome, and the most common one in a genuinely small practice.

If they come back the way they usually do, you are not looking for a bigger spreadsheet. You are looking for the spreadsheet to stop being the interface — for referrals to live in something that can refuse a half-finished record, remember who changed what, and be read by one person without being editable by all of them.

We run a free one-hour referral audit for independent practices in Houston. It covers the four checks above on your actual sheet, you get written findings, and there is no charge or obligation — including when the finding is that the sheet is fine and the money is better spent elsewhere.