On Tuesday, UPS told the market it had finished removing roughly two million packages a day from its own network — the tail end of a deliberate, years-long retreat from its largest customer. Revenue still grew 7.6%. Domestic margin expanded. The company paid $891 million in severance to get there.
That is what firing your biggest customer looks like at $91 billion of scale. The mechanics are identical at $91,000. One account pays most of the bills, sets most of the terms, and quietly owns your calendar, your hiring plan and your sleep. Everyone knows it’s a risk. Almost nobody has a written plan for it, because the honest version of the plan starts with “earn less money for a while.”
This is that plan. Not “diversify” as a slogan — a sequence, with the math, the scripts, the calendar and the five ways it goes wrong.
Part 1 — Measure it before you feel it
Concentration isn’t a vibe. Run these four numbers on your last twelve months of revenue. Ten minutes, one spreadsheet.
- Top-1 share. Largest client revenue ÷ total revenue. Under 15% is healthy. 15–30% is a watch item. Over 30% is a structural risk. Over 50% means you don’t own a business, you own a job with extra paperwork.
- Top-3 share. Over 60% and a single bad quarter at any one of them takes the whole year down.
- Margin-weighted share. Redo Top-1 using gross profit, not revenue. Big accounts are usually discounted. A client at 40% of revenue and 22% of profit is a different animal than one at 40% of both — and far easier to release.
- Attention share. Estimate the percentage of your working hours the account consumes. When attention share exceeds revenue share, you are subsidizing them with the thing you can never buy back.
Write the four numbers on one line and re-run it quarterly. That line is the whole dashboard.
Part 2 — The K.E.E.P. test
Not every big account is a problem. Concentration is only dangerous when the account has leverage and you have no alternative. Score the relationship 1–5 on four axes:
- K — Kill risk. If they left in 90 days, do you make payroll? (5 = you’re fine, 1 = you’re done.)
- E — Economics. Gross margin on this account versus your book average. (5 = above average, 1 = you lose money and call it “strategic.”)
- E — Escalation. Payment terms, scope creep, weekend requests, contract one-sidedness. (5 = clean and mutual, 1 = they set every term.)
- P — Portability. Could you sell the same capability to five similar buyers tomorrow? (5 = yes, it’s productized; 1 = everything you built is bespoke to them.)
16–20: keep and grow — this is a great account that happens to be large. 11–15: reprice and de-risk. Below 11: plan the exit. And note the trap in the P score: the most dangerous accounts are the ones that pay well and quietly made you unsellable to anyone else.
Part 3 — Do the replacement math first
Never cut before you know the arithmetic of the hole. Worked example, deliberately uncomfortable.
A consultancy bills $600,000 a year. The anchor client is $270,000 of it — 45% of revenue, at a 28% gross margin against a 47% book average. So the account contributes about $75,600 of gross profit. Fixed costs are $290,000.
The instinct is “I need to replace $270,000.” You don’t. You need to replace $75,600 of gross profit. At the 47% book margin, that is roughly $161,000 of new revenue — 40% less top line for the same money, plus the roughly 600 hours a year the anchor was eating.
Three numbers to write down before you do anything else:
- Replacement revenue: lost gross profit ÷ target margin.
- Runway: months of fixed costs you can cover if the account vanishes tomorrow with zero replacement. Under six months, you glide down — you do not cut.
- Pipeline gap: replacement revenue ÷ your average deal size = number of new deals. Multiply by your close rate and sales cycle. That product is your real timeline, and it is almost always longer than the one in your head.
Our take: The replacement-margin calculation is the single highest-leverage thing in this playbook. Most owners stay trapped for years because they benchmark against a revenue number that overstates the account’s actual contribution by 40–60%. Run it once and the exit usually turns out to be a third smaller than it felt.
Part 4 — The three exits, with scripts
You have three moves, in escalating order. Try them in order. Most situations resolve at step one.
Exit 1: Reprice. The cheapest de-risking is charging what the concentration is worth. You are carrying their volatility; price it.
Script: “We’ve restructured how we price ongoing work. Starting [date, 60–90 days out], this engagement moves to [new rate]. The scope stays the same and I’d rather keep building with you than reset with someone new — but I’m being straight that at the current rate we can’t staff it the way you deserve. Can we walk through the new structure Thursday?”
Two rules: never apologize for the number, and never negotiate against yourself in the same email. If they accept, your concentration risk is unchanged but far better compensated. If they push back hard, you have learned something free.
Exit 2: Shrink. Keep the account, cut the surface area. Drop the lowest-margin workstream, move to a fixed retainer with a defined ceiling, or hand back the piece that eats weekends.
Script: “Going forward I want to focus our work on [high-margin core], which is where we move the needle for you. I’m going to transition [low-margin workstream] back over the next 90 days and I’ll document it and help you hire for it. Same partnership, sharper scope.”
Exit 3: Release. The full glide-down. Never abrupt, never emotional, always with a runway and a handoff.
Script: “I want to give you a lot of notice rather than a little. Our business is moving toward [new focus], and I don’t think we’ll be the right long-term fit for [their work]. I’d like to propose a six-month transition: we keep delivering at full quality through [date], I document everything, and I’ll introduce you to two firms I’d trust with it. I’d rather hand this off well than have it degrade.”
Say it on a call, then send it in writing the same day. The written version is not a re-argument; it is a summary with dates.
Part 5 — The twelve-month glide-down calendar
- Months 1–2 — Build before you cut. Nothing changes with the account. Productize one repeatable offer, write the one-page version, and open the pipeline. Target: five qualified conversations.
- Month 3 — Reprice. Deliver Exit 1 to the anchor and, importantly, to your next two largest accounts at the same time. Repricing the whole book makes it policy, not a targeted move.
- Months 4–6 — Land two. Two new accounts at book margin. The goal is not replacement yet; it is proving the offer sells to strangers.
- Month 7 — Shrink or release. With two new accounts closed, deliver Exit 2 or 3 depending on the K.E.E.P. score. You now have leverage you didn’t have in month one.
- Months 8–11 — Transition cleanly. Deliver at full quality through the last day. Document, introduce successors, collect a written reference while goodwill is at its peak.
- Month 12 — Re-measure. Run the four Part-1 numbers again. Set a written ceiling: no single client above 25% of revenue, enforced by turning work down.
Part 6 — The five failure modes
- Cutting before building. The rage-quit after a bad Friday. You lose the revenue and keep the fixed costs. Sequence is the entire playbook.
- Replacing one anchor with another. The relief of a big new logo is how a 45% concentration becomes a 45% concentration with a different name. Cap new accounts at 25% of revenue on the day you sign them.
- Announcing an exit you can’t fund. If runway is under six months, don’t give notice — reprice and shrink instead. Notice you have to retract costs more than the account did.
- Degrading delivery on the way out. Your last ninety days write your reference and your referrals. Deliver the exit better than you delivered the middle.
- Counting revenue instead of profit. Covered in Part 3, and still the most common reason people stay stuck for a decade.
Part 7 — Your first 90 minutes
- Minutes 0–20: Pull twelve months of revenue by client. Compute Top-1, Top-3, margin-weighted share and attention share.
- Minutes 20–40: Score your largest account on K.E.E.P. Be honest on P — portability is where people flatter themselves.
- Minutes 40–60: Run the replacement math. Write down replacement revenue, runway in months, and the number of new deals required.
- Minutes 60–75: Draft the Exit 1 repricing email. Don’t send it today. Sending it in month three is the plan; writing it today is what makes month three real.
- Minutes 75–90: List ten companies that look like your best non-anchor client. That list is the pipeline.
The uncomfortable truth in the UPS numbers is that the transition costs real money and the market punishes you during it. The comfortable truth is that they did it anyway, on purpose, and came out with better margins on a smaller base. Concentration isn’t a moral failing — it’s a stage. It only becomes a trap when you keep measuring it in revenue and hoping it resolves itself.