On-demand concierge: paying per request instead of holding a retainer

On-demand concierge is the arrangement with no standing commitment: you ask when you need something, you pay for that request, and nothing accrues between requests. Published guidance puts a single task somewhere between fifty and a few hundred dollars depending on complexity.

The absence of a retainer is the whole product. Nobody is holding capacity for you, which is why it is cheap and also why the answer is sometimes no.

Discrete, well-specified tasks with a clear finish. A reservation, a delivery coordinated, a supplier found and briefed, a document collected. Anything that can be described in two sentences and finished in an afternoon is well suited to it.

It is also the honest starting point for anyone unsure of their own volume. A year of per-request use produces the number that decides whether a retainer would have been cheaper, and that number is usually surprising in one direction or the other.

Two failure modes. The first is volume: past a certain number of requests a month, per-task pricing exceeds the retainer it was avoiding, and people frequently pass that point without noticing.

The second is scarcity. On-demand cannot deliver during a market's peak week, because the access that solves those requests comes from a standing relationship and there is no standing relationship here. That is not a service failure; it is the model working as designed.

The transition is worth making deliberately rather than drifting into it. Count the requests over three months, apply the per-task rate, and compare against a retainer quote for the same volume.

Add one further consideration that the arithmetic misses: a retained provider accumulates context, and context makes each subsequent request cheaper in time even when it is not cheaper in money.

Some providers ask for a registration step or a small deposit before taking a first request, which is not the same as a retainer. It exists so that identity and payment are settled before anything urgent arrives, and it is usually refundable against the first task.

The comparison is a division rather than a judgement. Take the monthly cost of the retainer being offered, divide it by the typical per-request fee, and the result is the number of requests a month at which the retainer starts to be cheaper. Below that number, paying per request wins; above it, the retainer does. Published guidance puts the crossover somewhere in the low teens, and the rate estimator works it out against the specific figures on the table.

What distorts the answer is volume estimated from a good month rather than from a year. Households and businesses both tend to remember the busy period and forget the eight quiet months around it, and a retainer bought against that memory is capacity paid for and not used.

The reverse error is rarer but more painful: staying on per-request pricing through a sustained increase in volume, where each individual request looks affordable and the annual total is well past what a retainer would have cost.

Paying per request buys the work and nothing else. It does not buy priority when the provider is busy, it does not buy a named contact who knows your history, and it does not usually buy an answer outside working hours. For irregular and unremarkable requests none of that matters; on the day something is urgent, all three do.

That is the real trade rather than the price. A retainer is largely a purchase of position in a queue you cannot see, and its value is proportional to how often you expect to need to be at the front of it.

A reasonable middle position exists and is worth asking about: a small standing fee that buys a named contact and a response commitment, with the work itself still billed per request. Not every firm offers it and few advertise it.