Saying no is a scheduling tool
Every yes is a claim on a future week. Treating refusals as calendar management rather than a personality trait made them much easier to give. There really is power in the word "NO"
This post's shape: Lead 60 words, 01 153 words, 02 147 words, 03 149 words, 04 129 words.
The one thing I've learnt over the past year is the power of the word "NO".
For a long time I treated it as a question of personality - either you're the kind of person who says no or you aren't, and I wasn't. What changed was realising a no isn't really about the person asking. It's about the calendar.
Every yes is a claim on a future week
Whenever you say yes, whether to a new client or a new feature request, this claims at least another week of your time. Not only does this time need to be budgeted, but you need a safety net for any surprises that might come up (a service you're reliant on goes down, a feature breaks, the client decides to modify something without telling you).
That second part is the one that gets skipped. The estimate covers the work you can see. The safety net covers everything else, and everything else always shows up. A week of feature is rarely a week on the calendar once the surprises have had their turn.
So a yes isn't free just because today is quiet. It's a claim on a week you haven't seen yet, made on behalf of a version of you who might already be full.
When "just do it with AI" sets the deadline
I've lost a lot of sleep trying to quickly deliver on features simply because I said "YES". I'm not saying deny everything, but if the timeline for whatever is pitched is denied because "Oh, just do it with AI", you need to say no - or at least have a clear paper trail outlining that in trying to meet this deadline, certain features will have to be scoped out. If the trade-offs aren't clear, you're setting yourself up for a reputation hit.
AI does make parts of the work faster. It doesn't make the testing, the edge cases or the back-and-forth with the client any shorter, and those are usually what the deadline forgot. A no here doesn't have to be a refusal. It can be a list: here's what fits in the time, here's what doesn't, pick.
Six projects and four frowns
I experienced this the most at work. When I first started, I was so excited to have people specifically requesting me to implement their ideas that I forgot to say NO to anybody. Next thing I knew, I had 6 projects sitting in my lap with everyone wanting an update in a week.
Valuable lesson to all devs, this is why product and project managers exist. I assumed everybody spoke to each other but instead I had 4 people sitting with frowns because it looked like I ghosted them on the project - only to arrive a week later with the prototype and an apology.
None of them were wrong to ask. Each of those projects was reasonable on its own. The problem was that nobody, me included, could see all six at once - so nobody could tell me which one should wait.
What I do now
I still say yes more than I say no. The difference is that the yes now comes with its cost attached.
- Route new requests through whoever owns the priorities. If that's a product or project manager, deciding what waits is their call, not a failure of mine.
- Put the trade-off in writing. If the deadline stays, the message says which features move out to meet it.
- Budget the safety net. An estimate without room for surprises is a best case, and a best case is not a plan.
A no that comes with a date isn't a rejection, it's a schedule. Most people are happy to wait when they know what they're waiting for - what they can't stand is finding out a week late.