DEV Community

Dan Liang
Dan Liang

Posted on

A structural disconnect can cost you business

Do you really understand your customers? Maybe not.

The customer paid and tried the tool with you on real work. They gave feedback, you made changes. More feedback, more changes. The step that got stuck last time actually works this time. After all that work together, the tool is getting easier to use. But when it's time to get their day-to-day work done, the customer still does it the old way.

The money was real, the feedback was real, and the fixes worked. So why did the trial move forward while the customer kept working the old way?

I've seen some founders get completely stumped by this.

Here's the disconnect: your team can check whether each feature has been fixed, but the customer has to get the whole job done. You can fix how the tool reads a spreadsheet separately from how it sends a draft for review. But to get a weekly report out, reading the spreadsheet, drafting the report, and reviewing it all have to fit together. If one step gets stuck, you don't get to try the next. Even if everyone's clear up front about what needs to be delivered, you may not see where something that used to get done without a second thought fits into the new process until you get that far. Confuse progress on the fixes with progress on handing over the work, and it's easy to think you're further along than you are.

That's what makes the disconnect structural. You're looking at it from the product side, measuring progress feature by feature. The customer judges the tool by whether they can finish a whole piece of work with it. A list of fixed features doesn't tell you which pieces of work the tool can now handle from start to finish.

Take drafting a weekly report. In the first round, the tool couldn't even read the project spreadsheet correctly, so naturally the feedback was to get that right first. In the second round, it read the spreadsheet correctly and the draft made it to the reviewer. Only then did someone ask about a date in the draft: "Is this the original delivery date, or the one we reconfirmed this week?" Whoever used to write the report had to check that date before sending it, too. That requirement wasn't new in the second round. The first round just never got as far as review. Fixing the spreadsheet step wasn't wasted effort. Getting it to work was exactly what let you see what the draft still needed.

Similar requirements came up in the training sessions and workshops on electronic ambulance records that Hansen and Simonsen (2012) describe: the record had to follow the patient when care changed hands, and different versions had to match. The discussion was about electronic records, but both requirements were already there when the records were on paper.

Not everything that comes up later was there all along, though. If the tool isn't connected to the approval system and a colleague has to paste the draft in by hand, that's extra work the pilot created. If a weekly report becomes a quarterly report and the standards for signing off change, that's a different task. To tell whether a fix revealed a requirement that was already there, compare the two rounds. Were they the same kind of work? Would you have had to meet that requirement doing it the old way, too? Did the first round get stuck before that step, and did the fix let you reach it this time?

So when we review a product's direction, we don't just count how many fixes worked. We also look at who ended up doing the piece of work the tool was supposed to take over. If the customer reads the draft but still has to go back through the source material, sort out what goes in the report, and write it from scratch, the tool now reads the spreadsheet correctly, but the customer is still drafting the old way. If they can work from the draft, check the date as usual, make a few edits, and send it, then the way they draft has changed. Keeping a human review step doesn't mean the customer hasn't adopted the tool. And using it for this report doesn't mean they'll use it for every report from now on.

Feedback doesn't have to come in round after round. Sometimes there really is just one thing in the way. Fix it, and the rest of the steps can fall into place. If the tool has reliably handled the drafting on several similar tasks and the customer later goes back to working without it, it's time to ask different questions. Does it save enough effort to be worth learning a new way of working? Have access permissions and responsibilities kept up? Have any similar tasks come up since? That early payment matters. But it doesn't mean the customer will keep making time to work through revisions with you.

Top comments (0)