How do I tell whether my growth is real or just a spike from a launch or a discount?
The short answer
Separate one-time bumps from underlying growth by watching your retained-cohort base and your organic (non-paid, non-promo) new users over several weeks. A launch or a coupon inflates signups but the honest signal is whether those users are still active a month later and whether growth continues once the push ends. If removing the spike leaves a flat or falling baseline, you bought attention, not growth.
Go deeper, your way
3 hand-picked resources, 3 link-checked. Pick how you want to dig in.
▶️ Video
✓ Link checkedFreeIntermediate
Why we picked it
Balfour (founder of Reforge, former VP Growth at HubSpot) reframes growth away from one-off pushes and toward a model where one cohort of users feeds the next, which is exactly the difference between a bump and durable growth. Watching it gives you the mental picture of what compounding growth looks like, so you can ask whether your recent numbers are self-sustaining or just a spend-fuelled spike. Treat it as a starting point for building your own growth model, not a checklist.
Durable growth comes from a compounding loop where output (users, revenue) is reinvested to produce the next cohort, unlike a launch or discount that spends down and stops.
Sustainability mostly comes down to retention: if a jump in users does not hold over time, it will not compound.
Building a qualitative and quantitative growth model helps you spot your real constraints instead of chasing a one-time bump.
Why we picked it
This is the piece that named the problem you are asking about: the difference between numbers that make you feel good and numbers you can actually act on. Eric Ries (the Lean Startup guy) wrote it as a guest post, and it is still the clearest short read on why a traffic spike or an upvote count tells you almost nothing. Treat it as a starting point for deciding which one or two numbers your launch should live or die by.
A metric is only useful if a change in it tells you what to do next. Total hits and signup counts almost never pass that test.
Cohort analysis (following a group of users through registration, trial, and purchase over time) shows whether your launch actually changed behaviour, or just briefly inflated the top of the funnel.
Look at per-customer and per-segment numbers, not one big aggregate, because a healthy total can hide the churn and drop-off that decide whether a launch worked.
Why we picked it
Once you suspect a spike, this guide shows you the mechanic that actually proves it: split customers into cohorts by the month they arrived and watch each group's retention curve separately. It makes the point that a stable-looking aggregate can hide a launch or promo cohort that is churning out fast while your older, organic cohorts quietly prop up the average. A practical starting point for telling promo-driven arrivals apart from durable ones.
