โ† All work
Growth experiments Activation A/B testing

Compounding the
top of funnel

Signup is the cheapest place to win: every point gained there multiplies through the whole funnel. We ran disciplined experiments on the first two steps of a regulated onboarding journey and moved both, hard.

My role
Product Lead, experiment strategy & funnel ownership
Where
B2B spend-management scale-up
Team
Product ยท Eng ยท Data ยท Marketing
+34.6%
"Email signup actioned", up from 3.24% to 4.37%
+59%
"Create company" conversion, up from 0.95% to 1.51%
Q1
early-funnel gains landed while the downstream diagnosis ran in parallel

01Situation

A funnel with expensive real estate at the top

Self-serve onboarding for a regulated B2B product. The first two conversions, actioning the email signup and creating a company, were running at 3.24% and 0.95%. Single-digit percentages at the very top mean everything downstream starves.

02Problem

Small numbers, huge leverage, no discipline

Top-of-funnel changes were being debated on opinion, not tested. And because the product is regulated, naive growth tactics (strip the form, hide the requirements) risked pushing unqualified or confused signups straight into the KYC process, moving the problem downstream instead of solving it.

03Goal

Lift the first two conversions, cleanly

Measurably improve signup-actioned and create-company rates through controlled experiments, without degrading the quality of traffic entering the regulated funnel.

04My role

Product Lead. I owned the funnel, set the experiment agenda, defined success metrics, and held the line on "growth that doesn't poison downstream".

05Team

Product & Engineering (variants, instrumentation), Data (experiment design and readouts), Marketing (traffic mix, so lifts weren't just channel shifts).

06Diagnosis

Where intent was leaking

Instrumentation showed intent leaking at the very first actions: prospects who signed up but never actioned the email, and email-actioned users who stalled before creating a company. Both are moments of maximum motivation; losing them there is pure waste.

07Decision making

Experiments over opinions

Every proposed change ran as a controlled experiment against the two named metrics. Clear hypotheses, one change at a time where possible, cohort-aware readouts (so a "win" wasn't just a good-traffic week), and a kill rule for variants that lifted top-of-funnel while worsening downstream quality.

08Solution

Two moved metrics

The winning experiments simplified the first-touch moments: a tighter signup-to-email journey and clearer next-step framing at company creation, while keeping the regulated requirements visible so downstream cohorts stayed clean.

Experiment results: before vs after
Control Winning variant
Signup actioned, control
3.24%
Signup actioned, variant
4.37%
Create company, control
0.95%
Create company, variant
1.51%
+34.6% on email signup actioned and +59% on create-company conversion. Compounding lifts at the two cheapest points in the funnel.

09Trade-offs

Growth that respects the funnel behind it

We left some "easy" conversion tactics on the table (hiding requirements, over-promising speed) because they buy top-of-funnel wins with downstream pain: confused customers, more KYC feedback loops, worse activation. In a regulated funnel, traffic quality is part of the metric.

10Impact

Both target metrics, materially moved

Signup actioned up 34.6% (3.24 โ†’ 4.37%); create-company up 59% (0.95 โ†’ 1.51%). Because the experiments held traffic quality steady, the gains fed the regulated funnel with more of the same-quality prospects, setting up the downstream work captured in Finding the real bottleneck.

11Learning

Top-of-funnel wins are only real if they survive the middle

A lift that degrades cohort quality is a loan, not a gain. Pairing every growth experiment with downstream cohort checks is what made these numbers durable.

"Every point won at signup multiplies through the whole funnel."