Agency Delivery Handoffs: From Sales to Delivery Without Losing Context

Ask an agency owner where a project went wrong and most point to delivery: a slipped deadline, a scope argument, a client who seems to remember a different conversation than the one you had. Trace it back far enough and a large share of those problems start earlier, in a handoff that never really happened. Sales closed the deal. Delivery picked it up. Nobody actually handed anything over.

I have spent twelve years running delivery operations for marketing agencies across 92 completed engagements, and the handoff is where I have seen more damage originate than any other single point in the process. I have written before about where scope creep starts, and the sales-to-delivery handoff is the first of the three origins named there. This piece does not repeat that argument. It is the repair manual for the handoff itself: what has to cross the line from sales to delivery, why writing it down is not enough on its own, and the one question that tells you whether the handoff actually worked.

What actually counts as the handoff?

Most agencies think the handoff is a moment: the signed contract, the kickoff email, or the day the project appears in the delivery tool. It is not a moment. It is a transfer, and a transfer either happened or it did not, regardless of what got sent. If the person now running the project could not answer a client question about what was promised without pinging sales, the transfer did not happen, whatever paperwork exists.

That distinction matters because most agencies have a document and mistake it for a handoff. A signed statement of work is proof a deal closed. It is not proof that the person delivering it knows what the client believes they bought.

What has to actually cross the line?

Five things, and a scope document on its own usually carries one of them.

• Scope, with its exclusions attached. Not just what is included. What was explicitly ruled out, and why, so delivery can point to a reason instead of a rule when the client asks later.
• The promises made verbally. Every sales conversation produces sentences that never make it into the proposal: we can probably look at that, that is usually pretty quick, most clients get that included. None of them are lies. All of them are debts, and the client is holding the ledger even if your agency is not.
• The client’s actual decision-maker. Not the signatory. The person whose feedback stops the clock and whose objection actually matters. Sales usually knows this from the sales process. Delivery usually finds out the hard way, three rounds in, when the person approving the work is not the person who has been giving notes.
• The commercial shape of the deal. Fixed fee or hourly, what triggers a change order, what margin the estimate assumes, and what happens if the timeline slips on the client’s side rather than yours. Delivery makes dozens of small calls a week that depend on this and frequently makes them blind.
• Known risks. Anything sales noticed and downplayed to close the deal: a difficult stakeholder, an unrealistic deadline the client insisted on, a dependency on an asset that does not exist yet. These are exactly the details a closer has the least incentive to surface, and exactly the details delivery needs first.

A scope document, even a good one, reliably carries the first item and rarely carries the other four. That is not a knock on whoever writes it. Those four live in conversations, not contracts, and conversations do not transfer themselves.

Why does a document alone fail?

Because a document answers questions delivery has not thought to ask yet. A new project lead reads a scope of work the way anyone reads an unfamiliar document: for structure, not for gaps. The gaps only become visible once the work is underway and a client says something that does not match what is on the page. By then the correction happens in front of the client, which is the most expensive place for it to happen.

There is a second reason, and it is more structural. Whoever writes the handoff document is usually the same person who ran the sales process, which means they are writing from inside their own memory of the deal. What feels obvious to the person who lived every call is not obvious to the person reading a summary of it. The gap between those two states of knowledge is exactly what a document format cannot close, no matter how thorough the template.

None of this means documentation is pointless. It means documentation is necessary and not sufficient, and treating it as sufficient is the actual failure.

What does a live handover conversation need to cover, and who owns it?

The fix is not a better template. It is a scheduled conversation, before the project starts, between whoever sold the work and whoever is about to run it, with both people present at the same time. Not a forwarded thread. Not a document plus a Slack message saying let me know if you have questions, because a new project lead who does not yet know the client will not know which questions to ask.

The conversation has one job: surface the four things a document does not reliably carry. What was said that is not written down. Who actually approves. What the deal assumes commercially. What sales is quietly worried about. A short conversation covers this on most projects, not a lengthy meeting, and the cost is trivial next to a scope argument in week six.

Ownership has to sit with delivery, not sales, and this is the part agencies get backwards most often. If the outgoing owner is responsible for scheduling the handoff, it competes with whatever deal they are trying to close next, and it loses. The incoming owner has the incentive that matters: they are about to be accountable for a project they did not scope, and delivering blind is the outcome they should refuse before they refuse anything else.

This sits directly inside the delivery process the handoff feeds. Intake is the first of the seven stages a project moves through, and a handoff that transfers only a document is intake starting from a partial file. Everything downstream inherits the gap.

What is the test that the handoff actually worked?

Ask the delivery lead, cold, without letting them check the file first: what was sold. If they can answer including the scope, the exclusions, the decision-maker, the commercial shape, and the risk sales already knew about, the handoff worked. If they have to go find out, or ask sales, it did not, and the gap will surface in front of the client instead of in front of you.

That test is worth running on every new project for a month before you conclude your handoff process is fine. Agencies that skip it usually discover the gap the way I described at the top: as a scope argument that looks like a client problem and is actually a transfer that never happened.

So what does the repair manual actually say?

Fix the handoff at the source rather than managing its consequences downstream. A document that lists scope is necessary and never enough on its own. What has to move from sales to delivery is five things, not one: scope with its exclusions, the verbal promises, the real decision-maker, the commercial shape of the deal, and the risks sales already noticed. The only reliable way to move all five is a scheduled conversation owned by the person about to be accountable for the project, not the person trying to close the next one. And the only real test of whether it worked is whether your delivery lead can state what was sold without picking up the phone to sales.

Where the handoff sits inside a wider process that has no defined stages at all, fixing one stage in isolation will not hold. That is business systems work, and it comes before a handoff process can be trusted to stick.

Where to start

You do not have to guess whether the handoff is where your agency is actually leaking. The first step I recommend is the Agency Ops Audit, a $1,500 fixed-fee review that maps where delivery actually breaks, whether that is the handoff, scoping, scheduling, or somewhere else entirely, and hands you a written 90-day roadmap in two weeks with a 60-minute readout call. It is recommended, not required: a retainer can also begin after a scoping call. If the handoff turns out to be solid and the leak is somewhere else, I will say so in the readout. That is the only reason a review like this is worth paying for.