Jul 28, 2026
The feature request was impossible. The outcome wasn't.
A customer needed a large CSV export immediately. We couldn't build everything in time, so we changed the shape of the solution.
An account executive came to us with a request that sounded simple: a potential client wanted a CSV containing all the insight charts from one of our Media Research dashboards.
They wanted it quickly. Really quickly.
The first version of the request, as described, wasn’t realistic within the timeframe.
So I said that.
Saying no wasn’t enough
The easy response from engineering would have been:
We can’t build that in the time available.
And technically, that was true.
But the AE wasn’t asking for a CSV because they particularly cared about CSV generation.
They needed something they could show the potential client.
That changed the conversation.
Instead of asking how we could build the complete feature immediately, I asked what the smallest useful version could be.
We started with a few metrics
I suggested that we provide a CSV with a smaller set of metrics first, then add more in iterations.
The AE wasn’t convinced at first.
But we pushed ahead with the smaller first version.
When it reached the client, it turned out to be enough to move the conversation forward.
The complete export could come later.
The implementation wasn’t the real requirement
This was a useful reminder that product requests often arrive in the language of implementation.
“Give me all the charts in a CSV.”
But the underlying requirement might be:
“Give the client enough data to evaluate this.”
Those are very different engineering problems.
One might take weeks.
The other might take a day.
What I changed in how I respond to requests
I don’t think the lesson is to always build a smaller version.
Sometimes the complete thing really is necessary.
The lesson is to separate the requested implementation from the outcome someone is trying to achieve.
When a request looks impossible, I now try to ask:
What does the person actually need to be true?
Once you know that, there are often more ways to get there.
The feature request was impossible in the timeframe.
The outcome wasn’t.
← Back to writing
Comments
Loading comments…