Try Before You Buy: Do the Economics Work?

Try before you buy inverts the normal order of commerce: the goods go out first and the money follows only for what is kept. Conversion rates on the offer are extraordinary, because the customer risks nothing. Return rates are equally extraordinary, for exactly the same reason.

Whether the model works is not a matter of opinion. It comes down to three numbers, and most merchants evaluating it never separate them cleanly.

The Three Numbers

First, the keep rate: what share of dispatched value is actually paid for. Second, the incremental conversion: how many of these orders would not have happened at all without the offer. Third, the handling cost per returned unit, including the fact that returns arrive in bulk and in mixed condition.

The programme works when incremental conversion multiplied by kept margin exceeds the total cost of handling everything that comes back. It fails when the offer mostly cannibalises orders that would have happened anyway, because you have then added a full reverse-logistics cost to a sale you already had.

Incrementality Is the Whole Question

The single most common failure is offering try before you buy to everyone, including customers who were going to purchase regardless. Those orders convert at the same rate as before, return at a much higher rate, and consume handling cost that the original sale never incurred.

Testing incrementality properly requires withholding the offer from a control group and comparing net contribution, not conversion. If net contribution is equal, the programme is a cost with extra steps. This is the same rigour required of any conversion-lifting intervention, and it is the discipline missing from most published case studies of the model.

Target It Rather Than Offering It Universally

The programme works best where uncertainty is the binding constraint on purchase: high-fit-risk categories, first-time customers, high price points, or shoppers who have abandoned a cart at the size selection step. In those segments the offer is genuinely unlocking a sale that would not otherwise happen.

Offered to a repeat customer buying a basic they have bought before, it adds nothing but cost. Segmenting eligibility is what separates programmes that pay from programmes that quietly consume margin, and the segmentation inputs are largely the same as those used for predictive returns propensity.

• Measure keep rate on value, not on unit count.
• Run a holdout; incrementality cannot be assumed.
• Restrict eligibility to segments where uncertainty blocks the sale.
• Cap the number of items per trial and the trial window.
• Model handling cost separately from your standard returns cost.

If the customer would have bought anyway, try before you buy is not a growth programme. It is a subsidy for returns.

Operational Load Is Different in Kind

Trial returns do not behave like normal returns. They arrive in batches, contain multiple items from one order, and are more likely to include items that were tried on. That changes receiving, grading and resale economics in ways that an average return cost hides.

Plan for the batch shape rather than the volume. A programme that generates the same number of returned units as your baseline will still congest receiving if they arrive as multi-item parcels at the end of trial windows, which is a scheduling problem more than a capacity one. The relevant patterns are the ones described in returns receiving workflow.

Resale Rate Is the Silent Variable

Everything depends on what a returned trial item is worth. If items come back in resaleable condition at a high rate, the model can absorb a low keep rate. If they come back worn, unpackaged or damaged, the arithmetic collapses quickly and no amount of conversion lift compensates.

This is why the model works better in some categories than others regardless of demand, and why measuring resale rate specifically for trial returns rather than blending them into your overall figure is essential. Retail analysis on post-purchase models, including work published by the National Retail Federation, consistently points at recovery value rather than conversion as the deciding factor.

Payment Mechanics Shape Behaviour

How the money is held changes the keep rate materially. An authorisation on a card that is captured for kept items produces a different psychology from a fully free trial with no payment instrument attached, because the customer has already made a financial commitment in the first case.

Instalment and pay-later mechanics complicate the refund sequencing further and need designing rather than inheriting, particularly around when a plan is cancelled versus partially adjusted. The specifics are covered in BNPL returns and refund timing, and getting them wrong produces support volume that swamps the programme.

Finally, set a hard trial window and enforce it automatically. Programmes without a firm end date accumulate a long tail of items that are neither paid for nor returned, and chasing them manually costs more than the goods are worth.