Payment before execution is an asymmetry, and there is no version of the sentence that makes it symmetrical. The buyer moves money before anything has run, which means the buyer carries the risk during the exact period when the supplier is least proven. A supplier who arranges it that way owes both an explanation for the arrangement and something equally concrete on the other side of the scale. pipera arranges it that way.
The explanation is that the engagement does not begin with operating. It begins with building. Before a single day of running there is a site or a store to build or rebuild, a catalogue and its data to reconcile, creative to produce, advertising accounts and measurement to set up properly, email infrastructure to stand up, and access to be wired through every system the company already pays for. That work has real cost on the day it is done and it returns nothing while it is being done. There is no arrangement in which it is free. An operation funding it from its own balance sheet and recovering it over the following months would be lending, and lending is priced. The buyer would pay for the risk anyway, with less visibility into what the payment was for.
That is the argument for the asymmetry. On its own it is not enough, because it is the same argument every supplier makes for a deposit, and deposits routinely vanish into work nobody can point at afterwards. The counterweight has to be equally concrete, and the counterweight is a refund.
Deliverables are agreed in specifics on the call. They are written into the contract item by item, each line stating what the thing is and when it lands. If a deliverable in that contract does not happen, the fee for that deliverable is returned. It is not credited toward the following period. It is not offset against something else that did happen. It is returned.
Those three options are not variations on a theme. Credit forward keeps the money inside the relationship and converts a failure into a reason to stay in it, which is the opposite of what a failure ought to do. Offset lets a supplier net a miss against a hit and present the buyer with a total, which means the supplier sets the exchange rate between the two. Returning the fee is the only version where the failure costs the supplier what it costs the buyer, and the only version the buyer can verify without argument, because money either arrives in an account or it does not.
This is why the scope conversation takes as long as it does, and why it can feel disproportionate to a buyer who has already decided. A vague deliverable cannot be owed, and what cannot be owed cannot be refunded. Improve the site is not a deliverable. Grow the list is not a deliverable. Neither has a state that a reasonable person could look at and call a failure, so neither can trigger a return. The specificity is not administration wrapped around the deal. It is the mechanism the refund runs on. Every hour spent making a line precise is an hour spent making it enforceable, and a supplier pressing for loose wording is not offering flexibility, it is removing the claim before there is anything to claim.
The same requirement runs down into the operation. Whether a deliverable landed is settled on the record rather than in a meeting, because the record states what was done and when, and both parties read the same rows. An operation that reported by summary would be deciding its own refund cases.
What a refund does not do is give back time. If a deliverable does not land, the fee comes back and the period it was supposed to affect does not. The season is gone, the launch window is gone, and the position the business would have held if the thing had worked is not purchasable afterwards at any price. A refund is a floor under the loss. It is not compensation, and a buyer who treats it as compensation has mispriced the downside of the whole arrangement.
That has a direct consequence for what should be checked before signing. Each line should name a thing rather than an activity, because activities always happened and things either exist or do not. The dates should be staged rather than clustered at the end of the term, since a schedule that reveals failure early leaves the business time to respond, while a schedule that reveals it at the end makes the refund the only remedy left. The deliverables the business is genuinely depending on commercially should sit among the earliest, because the returnable part of the risk is money and the unreturnable part is the calendar, and the ordering decides how much of each is exposed. And the buyer should be able to tell from the record that a line was met without asking anyone, since a refund clause resting on the supplier's account of events is a clause the supplier controls.
None of this makes payment before execution comfortable. It makes it accountable, which is a different and smaller claim, and the smaller claim is the one that holds.
Read the deliverable lines before you sign anything, and ask which of them lands first, because the fee is returnable and the quarter is not.