How Substack uses Deceptive Design to Fuel Growth

If you look at Substack's marketing, you'll clearly see their hook: we will grow your newsletter.

Their showcase testimonial claims that 78% of their new subscribers come from other Substack recommendations.

How are they getting this kind of meteoric growth? Is it solely network effects?

It might be intentional, or it may be the outcome of relentless A/B testing.

I've seen first-hand that these techniques work—I would just never recommend that you use them.

Let me show you.

There is a possibility that none of these techniques were intentionally deceptive, but instead simply the result of poorly designed A/B tests.

For example, imagine that you were experimenting on the style of the back button:

And then separately ran an A/B test looking at the most effective title combinations:

This doesn't mean that the product teams at Substack are intentionally trying to deceive users, but it's plausible that they're so driven by metric optimisation that they didn't notice.

I see this happen all the time.

But what usually stops teams shipping these grey area designs, is either a moral compass, or a long-term mindset.

A/B tests measure short term bumps, not long-term changes in perception.

But what happens when AI has the freedom to "self-optimise" designs, and "A/B test at scale"?

It's somewhat inevitable that if AI is allowed to creatively self-optimise, we will see a lot more of this.

I've shown a series of "growth at all costs" moments in this study.

But let's finish on another really simple example: double opt-ins.

Anyone with a newsletter knows that over the long term, engagement rates are key. Zombie email lists arouse the suspicion of spam filters, and it can be hard to claw your way back out.

One popular option is to force "double opt-ins"—getting users to confirm their email addresses before they're added to your mailing list.

But as a writer on Substack, if you go to turn this on, you're met with a warning:

I know that there are mixed opinions about double opt-ins, but this is clearly one-sided.

I would argue that Substack should be educating their users about these features, not warning them about the downsides.

Perhaps they're worried that you'll get fewer new subscribers, and as a result, their growth won't appear so impressive.

Growth, growth, growth.