One of the most important growth features at early Facebook was the contact importer. You’d upload your email contacts, we’d match them against our user base, and you’d find your friends. Simple.
The teams working on it were measured on contacts imported. Millions of them. Every week the numbers went up. Big smiles all round.
Problem was, nobody was tracking what happened next. How many of those imported contacts actually got invited? How many of those invitations led to a registration? How many of those registrations turned into someone who actually used Facebook?
When we finally instrumented the full flow — contact imported → invitation sent → registration → active user — we found massive drop-offs at stages nobody was watching. We’d been celebrating the sound of water entering a bucket full of holes.
I see this everywhere. Companies obsessing over the top of the funnel while the bottom leaks. It’s the most common mistake in digital marketing and probably the most expensive.
At eBay we had a similar moment. We measured CRU — Confirmed Registered User. Anyone who finished the registration form. But “finished the registration form” and “actually bought something on eBay” turned out to be very different things. When we shifted to ACRU — Activated CRU, meaning someone who’d actually done something on the platform — it completely changed which marketing was working and which wasn’t. Sounds like a small definitional tweak. It was ground-breaking.
The lesson is always the same: draw out every step between the first touch and the action you care about. Measure every step. Find the cliff.
We did this with a 10-step email conversion flow at Facebook and found that one of the middle steps — an asynchronous processing tier — was silently crashing and losing messages. Users would start the flow, hit this invisible wall, and vanish. Nobody noticed because the steps on either side looked fine. Fixing that one thing gave us a “tens of percents” improvement. Not single digits. Tens of percents. From one broken step that nobody knew existed.
There’s a balance though. I always say the journey from click to conversion should be as short as possible but no shorter. At Meta this thinking led directly to lead gen ads (where users submit info without leaving the platform) and click-to-messaging ads (where the conversion is just starting a conversation). We put the conversion inside the creative itself. But you can go too far — if you strip out so much friction that people convert without understanding what they’re getting, you end up with chargebacks and churn. In 2024, selling Quest headsets, my team made exactly this mistake. Twenty years in and I’m still learning.
The other thing that changed how I think about conversion was growth accounting. Danny Ferrante introduced it at Facebook in 2008:
Acquisitions − Churn + Resurrections = Net Growth
At Facebook, churn and resurrections were each double the size of new acquisitions. So a 1% improvement in either had twice the impact of a 1% improvement in acquisition. Most companies throw all their conversion effort at acquisition. The maths says that’s probably wrong.
I’ve put all of this into a one-page Conversion Rate Optimisation cheat sheet as part of the companion stuff for my book Click Here. It covers the 5-step conversion audit, growth accounting, friction reduction, the Big Green Button test, and why you should always log with two independent paths (we found a >10% discrepancy at Facebook between server-side and client-side counts because of a pre-fetching bug that was logging “active” users who never opened the app).
It’s free, it’s one page. Print it out and stick it next to your monitor. The bottom of the funnel is where the money is.
