Smartlook built a signup flow that many growth teams would celebrate.
The flow reached a 70.6% conversion rate from signup-page visit to completed signup. Among people who started the process, 99% completed it. At one point the company described the experience as nearly frictionless.
Then a more important metric deteriorated.
Activation.
The Smartlook case, written by product manager Nikola Kožuljević and published by CXL in 2020, is one of the clearest examples of why local conversion optimization can damage the broader customer journey.
Smartlook provides behavioral analytics for websites and mobile apps. For the product to deliver value, customers need to connect their own data so they can see real recordings and behavior rather than demos.
Signup was not the value moment.
Activation was.
The original problem
Smartlook had an older four-step onboarding process.
It covered registration, profile, project setup and tracking code.
The flow had real issues.
It did not support mobile projects well. The design was dated. Users were effectively locked into setup before they could explore the application.
The team therefore redesigned onboarding.
Its stated goals included:
- get users to the core feature faster;
- reduce friction.
Those goals sound reasonable.
The problem was how success was measured.
The team optimized the visible funnel
Smartlook simplified the signup journey.
It added social login, fewer explicit interactions, automatic progression after choices, delayed tracking-code setup, delayed consent steps and cleaner UI.
The conversion metrics improved.
According to the CXL article:
- 70.6% of signup-page visitors completed the process;
- 75% of users who saw the signup page converted in later data;
- 99% of people who started signup completed it.
If the team had stopped measurement at account creation, the redesign would have looked excellent.
But the product still required activation.
Activation was the actual value transition
Smartlook defined activation around projects receiving data.
Without live customer data, users could not experience the full value of session recordings.
After the frictionless flow launched, the activation rate declined.
CXL reports that by January 2020 the activation rate was down 18.9%, equivalent to 14 percentage points, compared with the beginning of the previous year.
The flow created more accounts.
But too many accounts were inactive.
This increased infrastructure costs, messaging-tool costs, analytics noise and operational complexity.
The signup funnel improved while the customer system weakened.
Why reducing friction created bad data
One subtle problem involved segmentation fields.
Smartlook asked users to select role and company category.
Because the flow advanced automatically after a selection, many people simply clicked the nearest or first option.
The result was faster completion but poorer information.
That degraded Smartlook’s ability to personalize onboarding.
This illustrates a broader principle:
friction can sometimes create valuable commitment or information.
The correct goal is not zero friction.
It is zero unnecessary friction.
Useful friction can qualify a lead, collect required context, confirm intent, prevent errors or improve personalization.
Smartlook reintroduced friction deliberately
The team changed the signup flow again.
It required users to select role, select company category, move through steps with explicit Previous and Next actions and provide website category as a mandatory field.
This reduced raw conversion.
According to CXL:
- overall signup conversion declined from 70.61% to 66.98%;
- step completion declined by 1.21 percentage points;
- the overall conversion change was a 5.14% decline.
In a conventional CRO dashboard, that might look like failure.
But Smartlook reported important benefits:
- fewer new inactive accounts;
- lower cost from unused accounts;
- more reliable segmentation data;
- more contextual onboarding;
- better product tours.
The company accepted a lower local conversion rate to improve the quality of users progressing through the system.
The key mistake was metric hierarchy
Smartlook did not have a bad signup design problem.
It had a metric hierarchy problem.
The metrics should have been prioritized like this:
Business outcome
- retained revenue;
- paid conversion;
- active customers.
Product outcome
- activated projects;
- meaningful product usage.
Funnel metric
- completed signup.
Interface metric
- completion of individual steps.
When a lower-level metric improves while a higher-level metric deteriorates, the optimization is probably wrong.
This concept applies far beyond SaaS signup.
Ecommerce example
Removing address validation can increase checkout completion.
If it also increases failed deliveries and refunds, the local conversion gain is misleading.
Lead generation example
Shortening a form can increase leads.
If lead qualification collapses, sales productivity may fall.
Newsletter example
Aggressive popups can increase subscribers.
If engagement and deliverability weaken, the audience may become less valuable.
Paid media example
Optimizing to cheap conversions can increase platform-reported volume.
If revenue quality declines, CAC on real customers worsens.
The common failure is optimizing the easiest observable stage.
What operators can copy
1. Define activation before redesigning signup
Ask what action proves that the customer has reached first meaningful value. That event should usually matter more than account creation.
2. Use metric hierarchy
Every local conversion metric needs a downstream check.
Example:
`signup conversion → activation → paid conversion → retention`
3. Track inactive-account cost
Free signups are not free. They can consume infrastructure, support, email, CRM seats and analytics capacity.
4. Treat segmentation data as a product asset
If personalization depends on profile data, form quality matters.
5. Remove unnecessary friction, not all friction
Evaluate each step. Does it reduce risk, improve information, qualify, increase commitment or prevent a later failure? If yes, the friction may be productive.
What not to copy
Do not add fields simply because Smartlook reintroduced fields.
For another product, extra steps may genuinely be waste.
The transferable principle is that optimization should follow the full value chain.
The importance of causal caution
The activation decline occurred after the signup redesign.
That temporal relationship is useful but does not prove the redesign caused the entire decline.
Smartlook itself acknowledged that other factors, including increased lead generation, could affect the activation rate.
The team formed a hypothesis and adjusted the flow.
That is good practice.
Operators should avoid telling a cleaner story than the data supports.
Product activation should shape acquisition reporting
The Smartlook case also changes how acquisition should be reported. If signup is easy but activation is difficult, marketing dashboards that stop at account creation will overstate channel quality. A source generating the cheapest registrations may be sending users least likely to complete setup.
A stronger report follows cohorts beyond signup and compares activation by source, campaign, device and persona. That can reveal whether the problem is onboarding design or audience quality. It also prevents teams from paying more to scale inactive accounts.
For self-serve SaaS, the acquisition metric should therefore sit as close as practical to first value. When optimization platforms need higher event volume, signup can remain a secondary signal, but business decisions should still be checked against activation and retention.
Limitations of the case
The Smartlook story is unusually transparent, but it still has limitations:
- it is a company-authored account published by CXL;
- multiple onboarding changes happened over time;
- no randomized experiment isolates the entire flow change;
- the final article reports operational benefits but not a complete downstream revenue analysis.
Radar Digital therefore treats the case as a strong operational example, not a definitive causal experiment.
The operating lesson
A conversion rate is not a business outcome.
Smartlook’s near-frictionless signup flow made the top of the product funnel look stronger.
Activation showed that the system was weaker.
The correct optimization question was not how to maximize signup completion.
It was:
How do we maximize the number of customers who reach meaningful product value efficiently?
That shift changes how a team designs onboarding, instrumentation and experiments.
Sometimes the best conversion optimization is to accept a lower conversion rate at the wrong stage.
