I have been running Treendly since 2019. In that time it has collected 44,296 signups, and until last week I could not have told you how many more it has left in it. I had monthly charts, a growth rate, the usual dashboard. None of that answers the actual question, which is: is this curve flattening because I stopped pushing, or because the market is running out?

Those two look identical on a line chart and they call for opposite decisions.

So I fitted the Bass diffusion model to it. Here is what it said, and where it turned out to be worth trusting.

The model in one paragraph

Frank Bass published this in 1969 to forecast sales of new consumer durables, and it has been quietly correct about product adoption ever since. It assumes only two kinds of adopter. Innovators find you on their own: they see an ad, read a post, search for a tool. Imitators adopt because other people already did: word of mouth, a colleague's recommendation, seeing it used somewhere.

Two coefficients describe them. p is the innovation rate, the pull of the outside world. q is the imitation rate, the pull of existing users. A third parameter, m, is the total market: everyone who will ever sign up.

That is the whole model. Adoption at any moment is driven by p acting on everyone who has not adopted, plus q acting on the fraction who have. Early on the q term is near zero because there is nobody to imitate. Later it dominates. That is what produces the S-curve, and unlike an exponential or a straight line, it has a built-in ceiling.

What it said about Treendly

I fitted it on 91 complete months of signups, using monthly new signups rather than the running total. This matters more than it sounds and I will come back to it.

p (innovation)  0.00201
q (imitation)   0.0440
q/p             21.9
m (ceiling)     71,290
reached         44,296  =  62%
peak            month 67 of 91

Three things fall out of that.

Growth was almost entirely word of mouth. The ratio q/p is 21.9, meaning the imitation force is roughly twenty two times the innovation force. An innovation coefficient of 0.002 is very low. In plain terms: over eight years, external acquisition has barely moved this product. People arrived because other people were already there.

That is not a compliment or a criticism, it is a constraint. It says the historical evidence for spending on ads here is close to nil, because it has never been the mechanism. If I wanted to change that I would be doing something the product has never done, not doing more of what worked.

The peak is behind me. Peak monthly adoption landed at month 67, about two years ago, at roughly 857 new signups a month. The model says the fastest growth already happened and I was not especially aware of it at the time.

But it is not over. The ceiling comes out at 71,290, which puts the product at 62% penetration with about 27,000 signups still available. That is a materially different picture from a product that has run out of market. The curve is flattening because the imitation engine is running out of people to imitate, not because demand vanished.

What I actually do with this

The useful output was not the forecast. It was the diagnosis.

Treendly's growth engine is word of mouth, it is past its peak, and it has roughly 27,000 signups of headroom left at the current mechanism. That points at a specific decision: either accept a long slow tail, or change the acquisition mechanism to something the product has never had, knowing the historical data offers no evidence it would work.

That is a real strategic fork, and I could not see it in any dashboard I already had. A model from 1969, fitted in an afternoon, put it in front of me in numbers.

The other products are next.