Four pricing changes in eighteen months, and what each one cost me

I have raised prices four times since the start of 2025. Every one of them increased revenue and none of them caused the collapse I quietly expected. That is not the interesting part. The interesting part is that the damage, when it came, had nothing to do with the number.

The four changes

Briefly, so the rest makes sense.

The first was a 20% rise on new signups only, existing customers untouched. The second added a higher tier above the existing one and left both lower prices alone. The third moved everyone — including existing customers — after sixty days’ notice. The fourth was not a rise at all; it was removing the cheapest plan.

Revenue went up each time. Churn moved by less than a percentage point in three of the four.

What the third one taught me

The third change was the one that hurt, and it was not the largest increase. It was the one where I wrote the announcement email badly.

I led with the reasoning — hosting costs, support load, the honest stuff — and buried the actual new price in the fourth paragraph. I thought I was being transparent. What I had actually done was make forty-one people read four paragraphs looking for a number, which reads as evasion whether or not you meant it that way.

If you are announcing a price, the price goes in the first line. Everything else is context for a decision the reader has already started making.

One customer — the one with the largest account and the longest tenure — replied with a single line asking whether this was going to happen every year. That was the real question, and I had not answered it anywhere in the email. I have since put a sentence in the pricing page saying when I next expect to review prices, and nobody has asked again.

Three replies out of forty-one

The number I keep coming back to is not churn. It is that across all four changes, out of everyone affected, three people replied.

I had budgeted a week of my time for handling the fallout of the third change. I used about forty minutes. That gap between the anxiety and the actual workload is the single most useful thing I have learned about pricing, and it is not something you can be told — I had read the same reassurance on a dozen blogs and still spent a fortnight not sending the email.

What I would do differently

Grandfather by default, and say so. The first change cost me nothing in goodwill because existing customers were explicitly exempt, and I said that in the subject line. It cost almost no revenue either, because the existing base was small. That trade gets worse as the base grows, so this is advice with a shelf life.

Announce the number first. Covered above, learned the hard way.

Say when the next review is. Uncertainty about the future price is worth more anxiety to a customer than the current price. A sentence fixes it.

Stop modelling it. I built a spreadsheet before the second change that projected three churn scenarios. All three were wrong, and none of the differences between them would have changed what I did. That was an afternoon spent on the feeling of being careful.

The one I got right by accident

Removing the cheapest plan was the change I worried about least and it was the most valuable. Not because of the revenue — most of that cohort moved up rather than leaving — but because it removed the plan that generated the most support tickets per pound. My support inbox halved within two months.

I did not plan that. I removed the plan because the margin was bad. Finding out that price and support load were connected was luck, and it is the thing I would look at first if I were doing this again.