How to Run Lean Ecommerce Operations Without Becoming the Bottleneck

For a lean ecommerce team, the thing that keeps the founder out of every decision isn’t more SOPs — it’s a few operating systems: one source of truth for information, policies that make each decision once, and hiring based on real skills instead of a gut-feel thumbs-up. Here’s the operating system I build across the brands in our portfolio.

Why does adding more SOPs usually make operations worse?

Walk into most growing brands and you’ll find everyone working hard, getting things done — and somehow still disconnected. One team is drowning while another has capacity. A few people seem to know everything; everyone else is stuck in an information desert. And the company “documentation” lives in nine different places.

The instinct is to fix it with more: more SOPs, more process, more documentation. That’s the mistake. The real problem usually isn’t documentation — it’s alignment and leadership.

The fix that actually works is the opposite of more process. Build one knowledge base that’s the single source of truth — not another folder, the place to look. Then cross-train people across functions, so teams share context and the gaps in your knowledge base surface naturally. That’s operational leadership: aligning the team to work as one unit, not burying it in paperwork.

What systems does every ecommerce brand actually need?

Operate enough brands and the same core systems show up every time, regardless of category:

• A way to measure performance and protect cash flow
• A way to consistently deliver customer service
• A way to create and test marketing — messaging, page design, ad creation

The win isn’t doing these once. It’s building them as frameworks so you’re not reinventing the wheel on every product, every campaign, every hire. Lean teams move fast because the system carries the load, not because everyone is heroic.

How do you stop being the bottleneck for every decision?

Treat policies as decision machines. Make a decision once, write the rule, and you never have to make that decision again.

In practice that looks like clear policies for things like:

• When to issue a refund
• How to handle customer-service edge cases
• Which marketing tests to run, and how to run them
• When to reorder inventory

Each policy removes a decision from someone’s plate and creates consistency across the business. The result: things move faster, the team is more confident, and you stop being the person every question routes through. If you’re answering the same question twice, that’s your next policy.

Where should your operational information live?

In one — and only one — place.

Here’s how I learned to take this seriously. We were filling out a vendor application for Chewy: a 150-column spreadsheet, genuinely one of the worst forms I’ve seen. It took weeks — but the form wasn’t the real problem. We were. Product information was scattered across old emails, different teammates’ heads, and texts to the CEO asking things like “what’s the case pack quantity on the 60-day?”

The deeper issue: a single product had six variants (different servings, bulk sizes), multiplied across a dozen products, each with its own dimensions, weights, case packs, and SKUs across Shopify, Amazon, and Chewy. Spreadsheets force an impossible choice — repeat the same data for every variant, or fragment it across tabs nobody can navigate.

What you actually need is a relational structure: a product has many variants; a variant belongs to one product. Shared information (formulation, usage, nutrition) lives at the product level; variant-specific details attach to each variant. I built ours in Notion — no developer, it adapts as we grow, and anyone on the team can maintain it. The payoff: the next vendor application is an afternoon, not weeks of detective work.

How do you scale a team without quality falling apart?

Years ago, running my ad agency, I learned this the painful way. I managed 20 media buyers running Facebook ads for clients, and we were growing fast. Our system: pair each new buyer with a senior one, let the senior train them, and when the senior gave a thumbs-up, the new buyer took their own accounts.

Then a newly “ready” buyer was handed an account and lit tens of thousands of dollars of a client’s budget on fire. We got fired. He’d been blessed as ready on one account — and was nowhere near ready for that one.

The root cause: we’d reduced a complex, fiduciary job to a single subjective metric — one senior person’s thumbs-up. The fix is to enumerate the skills the role actually requires and grade each one individually. Only then do you really know what someone can handle. “Great at X, still developing Y” is a readiness map; “ready / not ready” is a coin flip — especially when money is on the line.

Where does AI fit into all of this?

This is the part most people skip. Clean data, clear policies, and real measurement are exactly what make AI useful in operations. Point AI at a messy operation and it produces confident nonsense; point it at a well-structured one and it becomes leverage — the thing that lets a lean team compete with a company three times its size. (More on that in Why AI Tools Can’t Tell You What’s Actually Driving Your Ecommerce Growth.)

The systems come first. AI amplifies what’s already in order.

The bottom line

Lean operations aren’t about doing less. They’re about building systems that do the deciding, remembering, and measuring so the team — and the founder — don’t have to. A single source of truth, policies that make each decision once, hiring on real skills, and measurement you can trust. That’s the operating system that lets a small team run a real business.