Agency Sales Process: Build the System Before You Train the Seller

An agency sales process is the documented path a deal travels from first contact to signed work, with a clear definition of what has to be true before it moves to the next step. That last part is where most of them fall apart. Plenty of agencies have stage names in a CRM. Far fewer can tell you what has to be proven before a deal earns the next stage.

I bring this up first because of what usually prompts the search. Revenue softens, a couple of good opportunities slip away, and somebody decides the team needs to get better at closing. So the hunt starts for better discovery questions, a script for the business development hire, a trainer with a repeatable method.

And nobody has figured out where the deals are actually failing.

That's what makes training a guess. You're buying a fix before you know what broke.

Two agencies can both tell me sales is weak and be describing opposite problems. One barely gets in front of a buyer who could ever afford them. The other gets plenty of strong opportunities and loses them somewhere between the first call and the signature. Those need different work, and the same two-day workshop won't touch both.

So here's the sequence I'd use: map the process you already run, define what each stage requires, find where deals actually die, and then decide whether what's broken is the process itself, something upstream of it like positioning or pipeline, or a person who needs coaching.

What the stages actually are

Most agency sales processes have six stages. The names matter less than the exit conditions, which I'll get to, but you need the shape before you can find the leak.

Each one below gets an exit condition, which is the thing that has to be observably true before the deal moves on. That's the part to copy. The stage names you can change to whatever your team already says.

Fit check

Somebody raised a hand or you reached out. Keep this to a cheap screen: does this buyer look like somebody you can help, in a category you serve, at a size that can fund the work? You're reading signals here and confirming them later.

Exit: the buyer matches your written boundaries on category, size, and problem type, and you've identified a specific reason to think there's something to talk about. Anything that fails here gets declined rather than parked.

Discovery

Now you confirm what the screen only suggested, and go considerably deeper. You're working out what they're trying to change, why now, who else votes, and what result would justify the spend.

Exit: you can state the business problem in the buyer's own terms, you know why it matters now rather than next year, you know who influences the decision and who signs, you've confirmed the fit you assumed at the screen still holds once you understand the situation, and you have a specific agreed next step on the calendar.

Shaping

You and the buyer converge on the problem, a rough scope direction, and a price range that won't shock anyone. This is the stage most agencies don't have, and its absence is one of the more common reasons proposals come back for rewrites.

Exit: the buyer has explicitly agreed to the problem statement and the approach direction, has heard a price range and not flinched, and knows what the proposal will contain before it arrives.

Proposal

The document confirms what you already agreed rather than introducing it. If anything in the proposal is news to the buyer, shaping wasn't finished.

Exit: the proposal is delivered live rather than emailed cold, every decision-maker has either attended or been briefed, and you have a date for their answer.

Decision

Procurement, legal, budget holders, and whoever else was never on your calls. Deals that felt closed often sit here.

Exit: the paperwork path is known and moving, with a named owner on their side for each step.

Handoff

What sales promised becomes what delivery owes.

Exit: the delivery team has read the discovery notes and the scope, has flagged anything they can't honor, and the client hears the same story from delivery that they heard from you.

Those exit conditions are the whole point of writing a process down, so let me be specific about why.

Give every stage an exit condition somebody can observe

A stage earns its keep when everyone can say why a deal is sitting in it and what has to happen before it moves.

"Discovery complete" fails that test if all it means is that a call happened. The calendar invite proves people talked. It says nothing about whether the deal is real.

Look at the discovery exit above and notice what it doesn't say. It doesn't say a call happened, or that the seller built rapport, or that the prospect seemed interested. Every item is something a third party could check by reading the notes.

Your criteria will be different from somebody else's, and they should be. What matters is that they describe evidence rather than activity.

"Verbal yes" is the one I'd define carefully, because enthusiasm and approval are two different things, and plenty of enthusiastic deals still have procurement and a budget holder in front of them.

Better forecasting is the obvious benefit. The one I care about more is diagnostic. When a deal stalls, you can point at which condition was missing instead of wondering whether the seller lacked confidence.

One caution, since this is where the sequence gets misread. Almost everything on that list is something discovery itself is supposed to establish. What has to be true beforehand is different: that you're reaching buyers you can help at all, and that you've given them some reason to take the call. Those are pipeline and positioning questions, and a seller can't create either one on a call.

Technique sits further downstream than people assume

I'm not going to argue technique doesn't matter, because it does. It just sits underneath a stack of things that have to be true first.

A sharper script won't turn a wrong-fit buyer into a qualified one. A better deck won't create urgency where none exists. And when a buyer has four agencies in front of them who all describe themselves the same way, objection handling is being asked to supply a distinction the agency never made.

When those upstream pieces are missing, you've handed the seller a job nobody could do. They're supposed to generate demand, explain what you stand for, build credibility, diagnose a business problem the buyer hasn't articulated, design the engagement, defend the price, and close it. Then when it doesn't work, everyone calls it a sales execution problem.

This is why I start by finding the constraint instead of the symptom. The four ways agency growth leaks are positioning, pricing, pipeline, and AI capacity, and they want different responses. Training improves execution inside a system that already works. It can't stand in for the system.

Founder referrals hide a weak agency sales process

The founder-led agencies I work with have almost all grown on relationships for years before anyone wrote a sales process down.

An old client introduces you to someone. A peer forwards an opportunity. Somebody who's known you for a decade calls with a problem.

Those deals arrive with some of the trust already built. The buyer knows how you think and how you work. They're not asking you to prove you're different, because the introduction did a lot of that for you. Gaps in your positioning and your proof get papered over because a person they trust vouched for you.

That produces genuine revenue, and it hides a fragile system.

You find out how fragile the first time you sell to somebody who's never heard of you, or the first time a business development hire has to carry a conversation while you sit and listen.

That new seller can know the services cold and still lose, because they don't have the years of context and can't reach for a relationship that was never theirs. They don't know which question cracks a vague brief open, which work sample matters to this particular buyer, or when a strange request means the deal is drifting.

If the agency only wins when you take over the call, some of that is skill you've earned and some of it is context nobody else can see. The second part is the part you can fix, and hiring another seller won't do it on its own.

Agency selling is diagnostic work

Agency deals rarely show up with a clean spec.

You get an incomplete brief, a budget somebody guessed at, internal disagreement nobody mentions on the first call, and a request framed around the wrong solution. Several people influence the decision and not all of them are on the Zoom.

Say a prospect asks for a new website. Underneath that could be weak positioning, a conversion problem, technology nobody wants to maintain anymore, a launch with a date attached, or an executive who's embarrassed to send people to the current site. Those lead to completely different engagements at completely different prices.

So the seller's job on that call is diagnosis. Work out what they're trying to change, why it matters now, who else gets a vote, and what result would justify the spend. Presenting comes later, if at all.

That's what your process has to support. A marketing agency sales process should help a seller shape a sound engagement, which is a bigger job than moving a contact toward a proposal.

Let the pattern tell you where to look

Sales problems leave fingerprints, and those are more useful than a general sense that the team should be better.

Treat these as places to start looking rather than verdicts. You still have to go read the deals, because more than one explanation fits most of these rows. Price pressure can mean the buyer sees interchangeable vendors, and it can also mean procurement is doing its job or your seller has no authority to hold a number. The table tells you where to point the flashlight.

Two rows deserve a note. If buyers can't tell your agency apart, negotiation training asks a seller to defend a distinction you never made. And when discounting and scope creep keep showing up together, look at whether the offer is built right and whether the price accounts for the uncertainty you're absorbing, which is work on your pricing model rather than work on the seller.

Rebuild the process from your own deals

Generic sales funnels are seductive because they arrive finished. Stage names, a tidy diagram, activity targets, all before you've looked at a single deal of your own.

Use the six stages above as a starting shape, then correct them against your wins and losses.

Pull a batch of recent opportunities and make sure it isn't just the happy ones. Include the losses, the deals that stalled, the work you discounted to get, and the ones that only closed because you jumped in. Then go through the records, the email threads, the call notes, the proposals, the follow-up.

For each one, reconstruct where it came from, what made the buyer move, why they called you specifically, who was involved in the decision, what you learned in discovery, which objections came up, when the proposal appeared, how scope and price shifted, where you personally got involved, and what finally moved it or killed it.

You're hunting for things that keep happening. Referrals closing easily while cold prospects can't figure out why you're different tells you the referral was carrying your positioning. Proposals going out before anyone knows how the decision gets made tells you the shaping stage doesn't exist yet.

What you end up with will look messier than the downloaded funnel, and that's the point. It describes how your buyers decide and how your agency earns confidence, which the template never could.

Turn your instinct into something you can hand to somebody

Founders tend to describe their sales ability as instinct. A lot of what's in there is pattern recognition that never got written down.

You know which prospects turn into difficult clients. You know which question exposes a brief nobody thought through. You can feel when a buyer needs educating, when they need proof, and when you should walk.

Getting that out of your head takes five pieces, and none of them require a consultant to produce.

Start with who you're for and who you're not: the clients you serve well, the ones you should turn away, the readiness signals you look for, and the disqualifiers you currently apply silently. A seller needs enough detail there to recognize fit without routing every lead back through you.

Then write down your questions along with what each one is for. A question list on its own doesn't help much, because the seller won't know what to do with the answers. Say what you're trying to learn, why it matters, and which answers should change the engagement you propose.

Sort your proof by what the buyer is afraid of rather than by industry or recency. Your portfolio isn't a gallery. Somebody nervous about stakeholder alignment needs a different case study than somebody nervous about technical complexity, and the seller should know which one to reach for.

Write the proposal rules down too. When you'll create one, what has to be agreed beforehand, how you use options, who reviews it, and which unanswered questions should stop it going out at all.

Last are the follow-up and escalation rules, covering how the team handles silence, a stakeholder appearing late, a discount request, or an ask outside what you do. Include when you should join the call, because founder involvement can be deliberate and useful in a way that constant rescue never is.

That work turns your contribution into something the team can use, which is most of what building an agency pipeline that runs beyond the founder actually involves.

Then train, once the process is explicit

Technique still matters. I said that at the top and I meant it.

I should be fair about what good training is, too, because the version I'm arguing against is the two-day workshop with a binder. Serious sales training includes deal reviews, call coaching, process design, and a manager who reinforces it afterward. Some of it will do the documentation work I just described, as part of the engagement. If that's what you're buying, you're further along than this article assumes.

What I'd still push back on is buying it before you know what you're fixing. A trainer who does process design will design a process. If your actual constraint is that buyers can't tell you apart, you'll get a well-built process that produces the same losses.

Training is the right call when the process is defined and the seller keeps missing the same standard anyway. Your discovery stage requires establishing urgency and decision authority. You listen to the recordings and they skip both subjects again and again, on deals where the questions clearly belonged. They know the requirement and they aren't doing it.

Even then, check the incentives before you book anything. A seller who avoids disqualifying might be protecting a pipeline number somebody yells about. That's a management problem wearing a skill problem's clothes, and no amount of coaching fixes it.

Measure whether it's transferring

Activity metrics don't tell you much on their own. Call counts and proposal counts can climb while the process stays broken, and chasing bad-fit prospects will inflate both.

What you want to know is whether the agency is getting better at winning the right work without you rescuing it. The numbers I'd watch are the share of discovery calls that meet your fit and urgency bar, close rate by source, close rate on deals you didn't lead, how often proposals get revised, how many active deals needed you to step in, loss reasons broken out by stage, and how many proposals went out before anyone understood the decision process.

Read them together and hold them loosely. A rising close rate on deals you didn't touch is a good signal, though it can also mean the lead mix shifted or the deals got smaller. Compare like with like before you conclude the system is working.

None of this is about removing yourself from important sales. Some deals should have the founder on them. What you're ending is founder rescue as the mechanism that makes an undefined process work.

Start with the constraint that's actually binding

The question in front of you is bigger than whether to buy sales training.

If you can't generate enough right-fit opportunities, the work is buyer selection, positioning, and pipeline. If prospects see interchangeable vendors, clarify what makes you different and back it with proof. If good deals keep collapsing into discount and scope fights, look at pricing and how the proposal is built. If the process is clear and sellers still fail at one specific stage, coach that skill.

Then watch what moves, and give it long enough that you're reading a pattern instead of a quarter.

If your agency still needs you personally to turn an uncertain conversation into signed work, that's where I'd start. Write down how you make those calls. Building a repeatable agency pipeline will show you what the next seller needs, what the process is missing, and where training can finally do the job you hired it to do.