1.First of all we will look over the data and see if its true
2. then will identify in which areas the volume has gone down, what types of user do we have there , if there are multiple areas are there any similar pattern visible of user and all
3. are the external factors needed to support a succesfull transaction is fine
4. has any specific event happened in the area of low trade volume which broke the trust
5. will look over the users feedback
6.then will see if theres any scheme or anything has been launched or done by competitor, basically we will do competitor analysis
Dhruv’s approach
Two people matter. The shopper in Europe at the payment screen, who wants it quick and wants to know the money moved. And the merchant hosting the button, who watches conversion. PayPal keeps a cut of every payment, the take rate, so volume is revenue.
Safety and completion pull against each other. Every fraud check, authentication step and compliance rule makes the payment safer and makes some people give up. In Europe that friction is partly law, not our choice.
I cannot ship for a month, so I will not guess at a fix. I will spend it working out how much of the 10% each cause is worth, then build in month two.
- VALIDATE THE DATA
At 10%, the number is as likely to be broken as the behaviour. Check it first.
- Does analytics match the settlement ledger? If both agree, the drop is real.
- Did the definition change? Refunds subtracted, transfers dropped, a dedupe fix.
- Did tracking break? An SDK release, a dead event in one build, a failing pipeline.
- Did the reporting boundary move? Countries moved out of Europe, merchants re-mapped.
- Count or value? If value, a weaker euro against dollar reporting shows 10% with no behaviour change.
- QoQ or YoY? European e-commerce falls every Q4 to Q1, so part may be calendar.
End with a number, not "looks fine": of the 10%, maybe 4 points season, 1 definition, 5 real.
-
FACTORS
Volume = active users x transactions per user x completion rate. Completion has its own funnel: checkout opened, PayPal picked, authentication, authorisation, done. Find which part moved and most causes stop mattering.
Where causes sit: measurement; the consumer (fewer users, more churn, less often); the merchant (a big one removed the button, or pushed it below the fold); product and technical (checkout regression, latency, a risk model over-blocking); regulation (SCA and 3DS step-up adding drop-off at authentication); competitors (iDEAL, Klarna, Bizum, Apple Pay); macro; pricing.
-
SEGMENTATION
I want to end up saying one sentence: it dropped in X, starting Y.
Cut by week first, because the shape tells you the kind of cause. A sharp step means something switched on a date: a release, a config, a fraud rule, a merchant leaving. A slope means erosion, so competitors or macro. A sawtooth means intermittent, like latency. Flat but low against last quarter is season.
Then country. Two or three markets points at merchants, local rivals or rules; everywhere points at us. Transaction type next, since recurring takes the worst of SCA. Then merchant, since the top 20 carry the volume. Then cohort, splitting acquisition from retention. Then platform and build, funnel step, funding instrument.
Now the other two sources. Data says where, feedback says why, so read verbatims from that market in those weeks. Then check if rivals there gained what we lost.
-
HYPOTHESES, ranked by likelihood x impact / cost to check
Each needs a confirm and a kill. A theory I cannot disprove is an opinion.
H1 Season. Confirm: YoY drop much smaller than QoQ. Kill: they match.
H2 SCA or 3DS friction. Confirm: drop-off between authentication and authorisation, EEA only, decline codes shift, complaints about codes and redirects. Kill: authentication pass rate flat. Ranked top: Europe is the reason here, not a coincidence.
H3 Merchant lost or button demoted. Confirm: step drop in a few merchants. Kill: decline spread evenly.
H4 Local competitors taking share. Confirm: a slope, country-matched, rivals growing. Kill: a sharp step, or rivals falling too.
H5 Risk model over-blocking. Confirm: declines up, matching a deploy date. Kill: declines flat.
H6 Macro. Last on purpose: what is left after ruling out what we control.
DELIVERABLE
Not one cause, but a split of the 10%, so leadership knows which fix is worth the engineering time. Primary metric: volume in affected markets, constant currency. Counter-metrics: fraud and chargeback rate, and compliance exceptions. The obvious fix is removing friction, and that buys volume with risk.