Customer Concentration Risk

While large contracts were counted only as revenue, they arrived as good news and nothing else. An annual deal closes, the chart goes up, and a chart going up is good. What share of the total that one deal represented went unrecorded.

The year after Pagedeck signed a $150,000 annual contract with one large enterprise shows what that share means. The deal itself was the best news the company had. The problem came from that single account becoming 38 percent of recurring revenue. At renewal, the customer asked for a 30 percent reduction and two features nobody else would use. Grace could not say no. Saying no removed 38 percent of revenue that day; saying yes filled the next two quarters of the roadmap with one customer's requests.

Customer concentration risk exists to put that position into a number before the contract gets signed. Read revenue totals alone and a good deal and a dangerous deal look identical. Record the share alongside, and which contracts can shake the company becomes visible before anyone signs.

The measurement divides what one customer pays by the total. The denominator is annual recurring revenue (ARR), revenue restated on a yearly basis. One-time setup or training fees get left out, because mixing them in makes any customer with a big first year look dangerous.

Three values go down side by side.

1. The largest single customer's share. Above 10 percent, it goes on the risk table.
2. The combined share of the top five. Past half, that is treated as a signal.
3. The share of the top ten. When the first two look fine and this reads 80 percent, the customer count itself is small.

The risk arrives in two forms, and the first to arrive is the loss of negotiating position. The customer does not have to leave: the fact that they must not leave is enough to make refusing anything impossible. Requests for discounts, or for features no other account would use, come in through that door. The second form is the actual departure. A customer paying 38 percent walks and revenue drops 38 percent that month, while the headcount hired against that revenue stays.

Reaching this state is not a mistake in itself. In a company with twenty customers, one at 20 percent is arithmetic. The problem is leaving the state in place. Of the two ways to lower the share, cutting the big customer's spend is one nobody takes, which leaves growing the denominator. In practice that means protecting the large contract while adding small ones quickly.

Contract terms can also buy time. A 90-day notice requirement pushes any sudden departure out by a quarter, and a two- or three-year term creates room to grow the denominator. Splitting a single enterprise into separate contracts per department helps too: one department leaving still leaves the others.

Whether renewal dates cluster gets checked alongside. If the top five contracts all end in March, half the company rides on that one month regardless of the shares. Staggering renewal dates as new contracts get signed costs nothing and removes that cluster before it forms.

The first place is investor due diligence. Any company preparing a next round gets asked for a top-customer share table, and registration documents for a public offering require disclosing customers above 10 percent of revenue. Two companies both at $7.5 million in revenue get valued very differently when one has a 40 percent customer and the other's largest is 6 percent.

In quarterly board reporting, the direction matters more than the level. A largest-customer share moving from 38 percent to 29 percent says the company is working its way off that one account.

On the risk table it occupies one row, next to items like key person risk. In a table that ranks rows by likelihood times magnitude, this item's magnitude is always large. Rows near the top get a contingency plan attached — a document written in advance saying what gets cut and what gets protected on the day a top customer signals they will not renew. Business continuity, which keeps the business running through an incident, covers system failures instead, so it is not the pairing for this row.

It also enters the design of sales targets. Given a revenue target alone, a sales team correctly judges that chasing one large deal is faster. Adding a new-customer count and a ceiling on the largest customer's share changes that judgment.

"Win this and that account is 30 percent of recurring revenue. This goes to the board first."

The share needs review separately from the contract value.

"Top five are at 54 percent. Let's set this quarter's target in customer count rather than dollars."

Shifting the work toward growing the denominator.

"Every renewal lands in March, so half the company rides on one month."

Clustering of dates, separate from the shares themselves.

Grace takes this table to the board every quarter. Last quarter the largest customer was 38 percent and the top five were 61. Sofia used that meeting to ask for next quarter's target to be new customer count instead of new contract value. Twenty small deals can be closed in the four months one large deal takes, and a dollar target gives nobody a reason to choose the twenty.

Grace attached one condition to the change: the relationship with the current largest customer stays exactly as it is. Refusing that customer's requests in order to lower the share shrinks the numerator and shakes the company now. Growing the denominator takes four quarters, and the account has to be protected through all four.

The most common reading is that a large customer is simply a good thing. Revenue rises, the name becomes usable in sales conversations, and other prospects get easier to convince — none of that is wrong. Those three belong in the revenue column, though, and concentration belongs in the risk column, so they get counted separately. Written into one cell, the fact that the best news of the year arrived with the largest risk of the year leaves no trace, and the first place it becomes visible is usually the renewal conversation twelve months later.

The other one is that adding customers solves it, since "concentration" sounds like a statement about count. What sets this risk is the distribution of dollars, not the count. A company with 100 customers where one pays 40 percent and the other 99 split the remaining 60 is exactly as concentrated as before. Make count the target and the work goes into small deals while the largest share falls by a couple of points. Both values have to sit in the target for the direction to be right, which usually means one goal for how many customers and one ceiling for how much any single one may represent.