Case Study

Case Study: 30 to 37 Accounts With the Same Seven People

How a white-label SEO team went from 30 to 37 accounts with no senior hire, cutting delivery turnaround from 11 days to 3-4 by writing down what done means.

Asif Dilshad
Nov 2, 2026 · 5 min read

W3whiz went from 30 to 37 client accounts with the same seven people and no senior hire. Delivery turnaround fell from 11 days to 3–4 and has held there. The fix was not more reviewers or a new tool. We wrote down what "done" means for each deliverable and checked it before work could reach a senior.

We are a white-label SEO team: two seniors, five juniors, delivering for several partner agencies. Two years ago we turned down five accounts because we couldn't say whether we had room. Today we have room for five more, and we can say so as a number.

The problem: 3 days of work, 11 days to deliver

Production took three days. Delivery took eleven. Eight of those eleven days were deliverables waiting in a queue.

The queue was built into the structure. Five juniors produced work in parallel across 30 accounts, about six each. Every deliverable then went to one of two seniors, who also handled escalations and pitch support. Five people producing, two people clearing in series.

So when a partner offered five more accounts, we couldn't say what they would do to turnaround. We said no.

The first fix made it worse

Adding a third reviewer cut turnaround by about a day, then revisions started climbing. We reversed it after six weeks.

Three reviewers meant three slightly different standards. Work that passed one came back flagged by another, and juniors started writing for whoever they expected to review them. That proved the limit was never reviewer hours. "Good" only existed in two people's heads.

What we measured: three weeks of review notes

Roughly 70% of what our two seniors asked for in review was checking that rules had been followed. About 30% was real judgment.

For three weeks, both seniors logged every change they requested, as they made it. Then we sorted the log:

TypeWhat it meansExamples from our log
Rule checkHas one right answer you could write on a checklistWrong template version, data from the wrong date range, missing client disclaimer, broken internal link, metric labelled differently than last month, brand colour off-spec for the partner
JudgmentDepends on context a checklist can't holdIs this the right call for this account? Will it land with this partner's client? Is the risk worth taking?

Two caveats. 70/30 is our ratio, from our deliverable mix and team; yours will differ. And we counted changes requested, not hours spent.

What we built: three simple parts

The key decision was checking quality when work is created, not at the end of the chain. It took about a quarter: three weeks of logging, six weeks writing the spec, then configuring and holding the line.

  1. A written definition of done for each deliverable type. A specification, not a task list: required sections, data sources and date-range rules, each partner's brand and disclaimer rules, naming conventions. My co-founder and I split ownership: I took reporting and technical deliverables, he took content and strategy.
  2. Rules enforced inside the tool juniors already use. Each deliverable type became a template with its own checklist. Mechanical checks must pass before a task can move to review, so failing work never reaches a senior's queue. This was configuration, not software development.
  3. An exception log. Every override, and every issue a senior catches that slipped through, gets logged. Recurring exceptions become new rules, about five or six a month.

This changed accountability. Before, seniors caught problems, so loose work could ship downstream. Now quality belongs to the person who produced it, with seniors as a safety net.

Where AI fits: AI helps produce first drafts. It doesn't decide what "done" means. AI output goes through the same template checks as anything a junior writes, and seniors still make the judgment calls. If anything, AI made the written spec more important: faster drafts just meant more work arriving at review.

Results, two years on

Seven more accounts on the same 1,000 delivery hours a month, with senior review time down by a third.

MetricBeforeAfter
Client accounts3037
Team2 seniors, 5 juniorsSame, no senior hire
Delivery turnaround (days)113–4
Total delivery hours per month~1,000~1,000
Delivery hours per account per month~33~27 (19% less)
Senior review hours per week~30~20 (33% less)
Senior review hours per account per week~1.0~0.54 (46% less)
Review volume handled by seniors100%About one third
Time for a new junior to become productive6+ weeksUnder 3 weeks
Spare capacityUnknown (turned down 5 accounts)Room for 5 more

Why did review volume fall by two thirds but review hours by only one third? Three reasons. Seniors now see only judgment calls, which take longer per item. Thirty-seven accounts produce more deliverables than thirty. And part of their time now goes to the exception log, turning recurring issues into new rules.

It also changed my job. About 60% of my week now goes to the operation. The other 40% goes only to accounts that aren't working.

We don't publish revenue or margin, for the same reason we don't publish pricing. The hours tell the story.

What it cost, and what it doesn't prove

For the first few months, juniors treated the checklist as the ceiling, not the floor. Work passed validation and stopped there: technically complete, strategically flat. We had to say explicitly that passing the checks is the minimum.

  • One agency. Seven people and 37 accounts. A 50-person shop will hit different bottlenecks.
  • Our ratio. 70/30 came from our deliverable mix. The transferable part is the three weeks of logging, not the number.
  • Our pricing helped. Partners pay a flat monthly fee per account, so faster delivery became margin, not a smaller invoice. If you bill by the hour, efficiency can cut your revenue.

Find your own ratio

You don't need a project to start, just three weeks of logging.

  1. Have every reviewer write down each change they request, as it happens. No formal reviewers? Log what you change before work goes out.
  2. Sort the log into two piles: broke a rule, or needed judgment.
  3. Small rule pile: your bottleneck really is senior capacity, so hire. Large rule pile: you have a definition problem, and hiring will make it worse.

Then it's build or buy. Build the system yourself in about a quarter, and I can be a second set of eyes. Or plug into an operation that already runs this way through white-label fulfillment.

Not sure which side you're on? Send me your two piles and I'll tell you straight. Book 20 minutes.

The full story: We Said No to Good Revenue (LinkedIn) · Method: AI SEO automation: start with the rule checks

The founder side of the same change, what I stopped doing and what my week looks like now: How I Got Out of My Agency's Delivery Queue.

How the exception log grew to 300+ rules: Two Years of Our Exception Log.

Keep reading

Send me your two piles

Log your reviewers' changes for three weeks, sort them into rule checks and judgment calls, and I will tell you straight whether you need a hire or a definition of done.

Book 20 Minutes