A large retailer wants to white-label your product and sell it to their own customers. Do you do the deal?

Frame the deal before you evaluate it

The mistake is to treat this as an arithmetic question about margin. It is a question about what kind of company you become, and the arithmetic follows from that. So establish the shape first, in three questions the interviewer will expect you to ask unprompted.

Who owns the customer? If the retailer's customers sign with the retailer, are billed by the retailer and raise support with the retailer, then you have no relationship with the people using your product, no ability to research them and no ability to sell them anything else. You have become a supplier of a component. That is a legitimate business, and it is not the business your direct sales team, your pricing and your roadmap were built for.

Who sees the usage data? This decides whether you can improve the product at all. A white-label arrangement where telemetry flows back to you is a distribution channel. One where it does not is a licence agreement with a reporting deadline, and after two years you will know less about how your own product is used than you do today.

And what is the retailer actually buying? Sometimes they want your product. Sometimes they want a checkbox on a bid, in which case usage will be near zero, the revenue is real and the strategic value is nil. Sometimes they want to learn enough to build it themselves, which is visible in the questions they ask during due diligence.

The revenue is not the number that decides it

Work the economics with every input stated. Suppose the retailer offers a flat annual licence of £600,000 with unlimited seats inside their customer base, against a direct business currently at £4m of annual recurring revenue. That is 15 per cent of current revenue in one contract, and it looks decisive.

Now cost it. The partner requires single sign-on against their identity provider, their own branding throughout, a data residency guarantee, and a quarterly reporting pack. Say that is 14 engineer-weeks of build. Your team has roughly 200 engineer-weeks of delivery capacity a year, so the integration is 7 per cent of the annual capacity, and it displaces whatever was in that slot. Then the ongoing cost, which is the part people omit: a bespoke deployment needs to be regression-tested on every release, their support escalations arrive through a named contact with a contractual response time, and each subsequent release has a second configuration to reason about. Call that 1 engineer-day per release at a fortnightly cadence, so 26 days or roughly 5 engineer-weeks a year, plus another 3 weeks absorbed by their escalations. That is 8 engineer-weeks a year, another 4 per cent of capacity, forever.

So the deal is £600,000 in year one against 14 weeks of build and 8 weeks a year of permanent overhead, and in year two it is £600,000 against 8 weeks. On those numbers it is comfortably worth doing on cash. The reason to hesitate is not the cash, it is that 11 per cent of your first-year engineering capacity now serves a single customer whose requirements are contractual rather than negotiable, and that your roadmap for the next two years contains a queue you did not choose.

Every figure there is an assumption, and the point of stating them is that an interviewer can push on any one. If ongoing overhead were 20 per cent of capacity rather than 4, the same deal is a different decision, and knowing which input flips the answer is the actual skill being tested.

Concentration, quantified

One customer at 15 per cent of revenue changes your negotiating position at every subsequent renewal, and it compounds if the deal grows. Do the projection out loud: if the partner grows to 40 per cent of revenue, they set your prices, they influence your roadmap, and a decision by one procurement director can remove nearly half the company. Firms in that position routinely discover they cannot refuse a discount request, because the alternative is worse than the discount.

The structural answers are all things agreed at signing and impossible afterwards. Cap the partner's share of revenue by pricing the deal so growth is per-customer rather than flat. Keep the term short with renewal rather than long with a break clause, because a long term with an exit you will never use is not protection. Refuse exclusivity, or price it explicitly as a revenue floor rather than granting it for goodwill. And insist on the right to contact end users for research even where you cannot sell to them, because that is the one concession that keeps you a product company.

What the deal forecloses

The cost that nobody puts in the model is the direct business you do not build. Two things go quietly. The first is roadmap coherence: a white-label partner's requirements are rarely the same as your direct customers', and every quarter you spend serving them is a quarter your direct product does not differentiate. The second is channel conflict, which arrives the first time your own salesperson meets a prospect who is already being sold your product under someone else's name at a different price. That conversation is worse than it sounds, and it is entirely predictable at signing, so decide in advance which segments are yours and write it down.

There is also a positioning consequence. Once your product is available under a well-known brand, some buyers will conclude that the well-known brand built it, and your case for being the category product weakens even as your revenue grows.

Answer with terms, not with a verdict

The weak answer to this case is a decision. The strong answer is a set of conditions, because that is how the decision is actually made in a room.

Something like: yes, on four terms. Usage telemetry flows to us and we may contact end users for research. Pricing is per end customer rather than flat, so their growth is our growth and their share of our revenue stays bounded. A two-year initial term, no exclusivity, and a defined list of verticals reserved for our direct team. And the integration work is scoped to a fixed 14 weeks with anything further treated as a change request at our rates. If they refuse the telemetry clause, the answer is no at any price, because a product we cannot observe is a product we cannot improve, and in three years we would be selling a component we no longer understand.

That last sentence is what a senior interviewer is waiting for. It shows you have identified which single term the whole deal turns on, and that you are willing to lose the revenue rather than the business.

Decide who owns the customer and the usage data before you look at the revenue. A partnership that removes your access to your own users is not a channel, it is an orderly handover of the product.