Customer concentration risk is one of those numbers everyone knows they should track and almost nobody actually does. It’s the single biggest predictor of small-business failure that doesn’t show up on a P&L. Your top customer pays your rent — what happens when they leave?
This guide walks through how to compute it, what the numbers mean, and three ways to get the answer ranked by what they cost.
What “concentration risk” actually is
Concentration risk is the share of your trailing revenue that comes from your top few customers. The standard cuts are:
- Top-1 share — what percentage of revenue does your single biggest customer represent?
- Top-3 share — same question, top three combined.
- Customer-count tail — how many customers contribute to the bottom 5% of revenue? (A long tail is a healthy buffer; a short tail means you’re a few-customer business pretending otherwise.)
A reasonable rule of thumb most lenders use:
| Top-3 share | Risk band | What it means |
|---|---|---|
| Below 25% | Low | Diversified book. Loss of any one customer is a bad quarter, not an existential event. |
| 25–50% | Moderate | Worth tracking quarter-over-quarter. Diversification work compounds. |
| 50–75% | High | Most lenders flag a covenant carve-out at this level. |
| Above 75% | Critical | You’re effectively a 3-customer business. One leaving is a layoff event. |
These bands are conservative. The exact threshold your specific lender uses will be in your facility agreement — read clause 5 or 6.
Option A: Excel + a Saturday
Cost: $0 + 4 hours.
What you do:
- Export every paid or authorised invoice from the trailing 24 months (Xero: Reports → Sales by Customer; QBO: Sales → Customers, then export).
- Build a pivot table grouping by Customer, summing Total.
- Sort descending. Compute each customer’s share of the total.
- Sum the top 3 shares.
- Write it up.
When this is the right answer: if you have under 20 customers and a clean ledger. The whole job is a 30-minute pivot table.
When this falls apart:
- Credit notes. If you don’t subtract
ACCRECCREDITinvoice types, you’ll over-count revenue by 5–15% for any customer with returns. - Customer-name mismatches. “Acme Roofing”, “Acme Roofing LLC”, “Acme Roofing Limited” are three rows in your pivot. You’ll either combine them by hand or report a falsely diversified book.
- Period boundaries. “Trailing 24 months” needs a moving window, not a calendar one. Excel makes this fiddly.
Option B: Big4 advisory or a compliance SaaS
Cost: $2,400+ per quarter for advisory; $948+/yr for the SaaS subscription.
What you get: a competently produced report, signed-off methodology, a meeting if you bought the advisory.
When this is the right answer: when your bank has specifically asked for a third-party-prepared report, or when you have an audit-defence requirement that needs a CPA’s signature on the methodology.
When this is overkill: every other case. The methodology is the same five-step pivot table from Option A; the value is the signature, not the analysis.
Option C: RevenueGuard
Cost: $19, once. Runtime: about 4 minutes from clicking “Connect Xero” to the PDF arriving in your inbox.
What you get: a 3-page PDF with:
- A 0–100 risk score banded against the table above.
- Top-3 share of trailing revenue.
- Revenue at risk (top 3 in dollars).
- Top-10 customer ranking with revenue + share bars.
- Plain-English findings ranked by severity.
- Methodology footnotes.
The bot ranks by revenue, not margin contribution — your accounting data doesn’t carry COGS. If you need margin contribution you’ll need a P&L cross-reference, which is a different job.
When this is the right answer: most operators with under a few hundred customers running this quarterly. It’s faster than Excel and cheaper than a renewal.
When this is not the right answer: if you need someone to sign the report. The bot doesn’t sign anything.
What no one tells you
Concentration is volatile quarter-to-quarter at small scale. A 10-customer book with one customer at 35% can swing to a different customer at 35% in 90 days as projects roll over. Trend it across four quarters before you panic; one print is a snapshot, not a signal.
Also: the score weights both top-3 share and customer count. A 60% top-3 share is meaningfully more dangerous in a 5-customer book than a 200-customer book. Most off-the-shelf calculators don’t capture this and over-warn small firms with naturally low customer counts.
When to do it yourself anyway
If your top-3 customers are 90%+ of revenue and you’ve known about it for two years, you don’t need a $19 PDF to tell you that. You need a diversification plan and a couple of difficult conversations. The bot doesn’t write either of those for you.