One of the most valuable lessons I have taken from building risk frameworks across multiple markets is this: the best intervention is not always the broadest one.
In cash on delivery markets, platforms invest significant effort in designing rules that reduce failed delivery costs. Those rules work. They prevent real financial leakage and protect platform margins in markets where cash changes hands at the doorstep rather than at checkout. Getting this right is genuinely hard and the teams that do it well deserve credit for it.
But as these frameworks mature, the most interesting question is no longer whether to intervene. It is how precisely to intervene.
Precision as a sign of maturity
Early stage risk frameworks tend to prioritise breadth. Cast a wide net, catch the problem population, and optimise from there. That is a reasonable starting point when the primary goal is stopping the bleeding.
As frameworks evolve the question shifts. Instead of asking how do we stop more failed deliveries, the question becomes which intervention produces the best outcome across both risk and commercial dimensions simultaneously.
That shift in thinking is what separates a reactive risk function from a strategic one.
What simulation teaches you
One of the most useful exercises in rule design is simulating the same intervention at different thresholds before deploying it. Not just measuring how many problem cases each threshold catches, but understanding who is in each cohort and what their broader value profile looks like.
When you do this exercise consistently, a pattern emerges. A broader threshold catches more of the problem population but also captures a segment of buyers whose long term contribution to the platform is positive. A more selective threshold catches a population that is near-zero in lifetime value contribution, where the intervention has almost no commercial downside.
The difference in protection between the two thresholds is meaningful but not dramatic. The difference in commercial impact is substantially larger. That asymmetry is the insight.
The metric that changes the conversation
Lifetime profitability contribution of the intervention cohort is not a metric most risk dashboards carry. But it is the metric that most clearly separates a good threshold from a great one.
When you can show that a more selective threshold removes near-zero lifetime value buyers, prevents a substantial volume of failed deliveries, and leaves commercially valuable buyers untouched, the conversation in the room changes. It stops being a debate about how aggressive the rule should be and starts being a conversation about how precise it can be.
That is a fundamentally healthier place to be making decisions from.
The collaboration that makes this possible
This kind of analysis does not sit entirely within risk. It requires connecting risk data with commercial and financial data in a way that most organisations have not yet systematised.
The platforms that are doing this well have built a working relationship between risk, finance, and growth functions where intervention decisions are evaluated against a shared view of buyer value, not just a single-function view of cost reduction. That alignment is harder to build than the analytical framework itself. But the payoff in decision quality is significant.
The takeaway
Mature risk thinking is not about being lenient. It is about being precise. The goal is not to block fewer buyers. It is to make sure that every buyer you block is one whose blocking genuinely improves the platform's position across all the dimensions that matter, not just the ones that appear on a single dashboard.
That precision takes more work upfront. It pays back in better decisions, fewer unintended commercial consequences, and a risk function that leadership trusts precisely because it thinks beyond its own domain.