What it means
The raw material is event data: a page viewed, a button pressed, a basket abandoned, a support ticket raised, each stamped with a time and a user identifier. Analysing the sequence of those events, rather than a total count of them, is what separates this from ordinary reporting.
It matters commercially because it connects product decisions to revenue. Once you know that customers who complete a particular step renew at twice the rate of those who do not, the value of improving that step becomes a calculation rather than an argument.
The everyday tools are funnels, which show where people fall out of a sequence, and cohort curves, which follow a group of customers who joined in the same month to see how many are still active later. Controlled experiments then test whether a change actually causes the improvement rather than coinciding with it.
The main pitfall is reading correlation as cause. Customers who use an advanced feature may renew more because they were committed to begin with, not because the feature made them loyal, and only an experiment can tell the two apart.
Data protection is the other constraint. Most of the value comes from aggregated patterns rather than individual profiles, so a well-designed setup collects the minimum needed, honours consent choices and keeps identifiers separate from personal details.
In practice
Real-world examples.
Example
An online homeware retailer replays anonymised mobile sessions and finds shoppers stalling at the delivery options screen. Simplifying it raises mobile checkout completion by 8 percentage points, which on 25,000 monthly mobile checkouts is 2,000 extra orders at a $54 average basket, or $108,000 a month.
Example
A retail bank compares customers who set up a standing order in their first 30 days with those who do not, and finds annual account closure rates of 4% against 11%. The onboarding journey is rebuilt to put that single action front and centre in the first week.
Example
A software company sees that accounts with fewer than three active users churn at roughly three times the rate of those with more. Customer success stops working alphabetically through the account list and starts contacting single-user accounts within their first fortnight.
Formula
Calculation
Step conversion rate = users completing the step / users entering the step
Overall conversion rate = step 1 rate x step 2 rate x step 3 rate
A subscription business measures one month of activity on its website.
40,000 visitors, of whom 8,000 create an account: 8,000 / 40,000 = 20%
Of those, 2,400 finish onboarding: 2,400 / 8,000 = 30%
Of those, 600 start paying: 600 / 2,400 = 25%
Overall conversion = 20% x 30% x 25% = 1.5%, which matches 600 / 40,000
Session data shows most of the loss at onboarding happens on a single form. Simplifying it lifts the onboarding rate from 30% to 36%, giving 8,000 x 36% = 2,880 activated users and 2,880 x 25% = 720 paying customers, an increase of 120. At an average annual revenue of $480 per customer, that one change is worth 120 x $480 = $57,600 a year.Case study
Seen in the real world.
Larkfield Fitness is an illustrative chain of 18 gyms with about 9,000 members paying $45 a month. Its app records every check-in, and an analysis of two years of data showed that members visiting fewer than four times in their first month cancelled within six months 62% of the time, against 21% for everyone else.
The chain introduced a free coaching call in week two, aimed purely at getting new members through the door a fourth time. Six-month retention across the membership base rose from 71% to 78%, which on 9,000 members is 630 memberships saved.
In this fictional example, holding those 630 members for six extra months each was worth 630 x $45 x 6 = $170,100, against a coaching programme costing a small fraction of that. The finance director's note to the board made the point that the insight came from behaviour already being recorded, not from new market research.
Watch out
Common mistakes.
- Reading a correlation in event data as proof of cause, when the customers who take an action are often different from the start.
- Tracking every possible event first and deciding what matters later, which produces a large bill and a dataset nobody trusts.
- Comparing cohorts of different ages, so a group tracked for two months looks better than one tracked for a year purely because it has had less time to churn.
Questions
People also ask.
Is behavioural analytics the same as web analytics?
Not quite, since web analytics counts pages and sessions while this follows the sequence of actions a known user takes over time.
Do you need personal data to do it?
No, most of the value comes from aggregated patterns and pseudonymous identifiers, which keeps the work within normal consent and data protection rules.
How quickly does it pay for itself?
Usually within the first two or three funnel fixes, because removing friction at a single high-traffic step tends to be cheap relative to the revenue it recovers.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%