Agency SOPs That Actually Stick

I'm Kurt Schmidt, founder of Schmidt Consulting Group. I work with B2B services firms and agencies on go-to-market strategy, pricing, and growth operations. And one of the patterns I see most consistently in agencies that plateau. Regardless of their service line, size, or market. Is this: their best knowledge is completely undocumented.

The term "agency SOPs" gets thrown around like it's an administrative chore. A binder full of how-to guides that nobody reads. But that framing misses what SOPs actually are when they're built right. A real SOP is a rule that runs. There's a meaningful difference, and that difference is what separates agencies that compound their experience from agencies that keep paying for the same lessons over and over.

---

Why Do Most Agency SOPs Fail Before Anyone Reads Them?

Most agency SOPs fail because they're written for the wrong moment. Someone documents a process after things are calm, in a format that requires the reader to seek it out, at a time when the reader is usually rushing. That's a hope.

The senior person on your team knows to check the thing that went wrong last quarter. They know which client detail always causes a scope problem. They know the step everyone skips when they're under pressure. That knowledge is real, hard-earned, and completely trapped. It works until that person is on vacation, pulled onto another account, or gone. And then the mistake the agency already paid to learn happens again. At full price. Because the lesson never left one skull.

I've worked with agencies that had strong, capable operators who carried enormous institutional knowledge in their heads. Those agencies weren't weak operationally. They were fragile. There's a difference. Fragile means everything works until one thread pulls.

A checklist nobody opens is still better than no checklist. But the goal is producing a rule that shows up exactly when the situation calls for it, whether or not anyone remembers it exists. That's what a durable agency SOP actually does.

---

What's the Right Way to Encode a Lesson So It Outlives the Person Who Learned It?

Before you can fix how you're building agency SOPs, you need a framework for evaluating encoding strength. Not all ways of capturing a lesson are equally durable. I think about this as a hierarchy, from weakest to strongest.

Training sits at the bottom. Telling people what happened and what to do differently. It decays fastest because people forget, and new hires never heard it in the first place.

Documentation is a step up. Writing the lesson down in a wiki, a Notion page, a Google Doc. Better, but it only works if someone reads it at the exact moment they need it. In practice, they rarely do.

Checklists are where things get meaningful. A checklist item that must be checked at the point of work brings the rule forward in time to exactly when it's relevant. This is a qualitatively different thing from documentation that lives somewhere else.

Templates are stronger still. When you bake a rule into the artifact. A proposal template with the right clause already drafted, a project brief with the required field that can't be left blank, a contract with the language already inserted. The right thing becomes the default. Doing it wrong now takes effort. That's a significant shift.

Automation is the top of the hierarchy. A system that enforces the rule or catches the violation without any human in the loop. A Zapier workflow that flags a missing field. A project management trigger that won't advance a stage without a sign-off. A CRM rule in HubSpot or Salesforce that prevents a deal from closing without required fields completed. The lesson runs even when everyone has forgotten it exists.

The test of real learning isn't "we talked about it in the debrief." It's "the next person literally can't make that mistake, or gets caught if they do."

Push every important lesson as far down this hierarchy as it can go. A minor lesson might be fine as a checklist line. A costly, recurring problem deserves to live in a template or get automated outright.

---

How Do Agency SOPs Create a Compounding Return on Experience?

This is where the framing shifts from operations to strategy. Encoding lessons is a growth lever.

An agency that encodes its lessons gets compounding returns on experience. Every mistake. Every client escalation, every missed deliverable, every scope bleed. Makes the whole system permanently smarter, and every new hire inherits the accumulated judgment on day one. They don't re-learn the hard way. They start from where the last person finished.

An agency that doesn't encode is stuck re-paying for the same lessons on a rolling basis. The senior people carry the weight. New hires take 6 to 12 months to get up to speed because the knowledge isn't accessible. And that dependency. On a few key heads. Is also what caps the price the agency can sell for. Buyers doing due diligence on an acquisition or a large retainer ask one version of the same question: what happens if your best person leaves? If the answer is "things get rocky," that's a valuation problem, a retention problem, and a business development problem all at once.

I've seen agencies where a strong project manager carried so much institutional knowledge that their departure literally reset the team's capability. The agency had to re-make mistakes it had already paid to avoid. That's a systems problem. The fix is building a system that doesn't depend on any one person's memory.

This connects directly to agency productivity and profitability and how services firms can grow without proportionally growing their senior headcount.

---

What Should You Do Immediately After Something Goes Wrong?

The debrief is where most agencies stop. They talk about what happened, they identify what should have been done differently, and they move on. The lesson exists as a shared memory for about two weeks. Then it fades.

The practice I recommend is a single question at the end of any debrief: what rule would have caught this, and where does that rule now live?

If the answer is "we'll remember," you haven't learned the lesson. You've felt it. The debrief has to produce an artifact. A new checklist line. A change to a template. An automated check. Something with a name on it and a ship date. Assign it before the meeting ends. Ship it before the memory fades. And I mean within 48 hours, not "next sprint."

This is the operational habit that separates agencies that get smarter over time from agencies that stay at a fixed capability level regardless of experience. It's also the habit that makes agency SOPs genuinely useful: a living system that gets updated every time something breaks.

Project management tools like Asana and ClickUp both support template-level rule enforcement well. For CRM-based rules, HubSpot's required fields and deal stage gates are a practical starting point. The specific tool matters less than the habit of pushing each new lesson to the strongest encoding level it warrants.

This connects to institutional knowledge transfer. Because the downstream benefit of a well-encoded system shows up most clearly when a new person joins and hits the ground running instead of spending months figuring out what the senior person already knows.

---

How Do You Avoid Building a Rule System That Nobody Follows?

This is the nuance that trips agencies up once they start taking encoding seriously. You can over-encode. A rule for every one-off creates process bloat, and process bloat teaches people to ignore the rules that actually matter.

I've worked with agencies that had built out extensive SOP libraries. 80, 90, 100 documented processes. And the actual compliance rate was close to zero. The team had learned to route around the system. That's worse than no system, because a system people ignore actively erodes the culture of following rules. When the checklist for something genuinely important shows up, nobody takes it seriously.

The filter matters. Encode lessons that are costly, recurring, or easy to get wrong under pressure. A one-off client situation that's unlikely to repeat doesn't need a process. A billing step that multiple team members have missed and that has cost real money. That deserves to be in a template at minimum, and automated if possible.

And revisit the rules. A quarterly pass through your checklist and template library to archive what's stale and update what's outdated is maintenance that pays. The goal is a system where every rule is one someone would actually follow if they read it. Because it's current, it's relevant, and it's clearly connected to something that matters.

Per Harvard Business Review's research on organizational learning, companies that build systematic post-failure review processes outperform peers on long-term operational consistency. The mechanism is straightforward: the lesson gets encoded, the encoding runs, and the organization doesn't re-pay for the same mistake. That's the whole thing.

This also connects to knowledge transfer strategies. The broader question of how services firms retain institutional knowledge as they grow.

---

Key Takeaways

• Agency SOPs only work when lessons are encoded into systems that show up at the point of work. Training and documentation alone decay too fast to be reliable.
• The encoding hierarchy runs from weakest to strongest: training, documentation, checklists, templates, automation. Push each important lesson as far down that hierarchy as it can go.
• The debrief question that matters is: what rule would have caught this, and where does that rule now live? If the answer is "we'll remember," the lesson hasn't been learned yet.
• Every encoded lesson compounds. New hires inherit accumulated judgment on day one instead of re-learning it the hard way over 6 to 12 months.
• Over-encoding is a real failure mode. A system full of stale rules nobody follows is as dangerous as no system, because it teaches people to ignore the rules that do matter.
• The agencies most dependent on a few key heads are the most fragile. And the most constrained in what they can sell for or hand off to a buyer.

The agencies I've watched scale without breaking weren't the ones with the most talented operators. They were the ones where every hard lesson. Every client escalation, every missed step, every scope bleed. Left the system a little smarter. The talent was real, but scaling your agency made it transferable.

The question to sit with after your next project goes sideways: is this team going to feel the lesson, or are they going to encode it?