Show Customer Concentration Before One Account Becomes the Whole Growth Story

Five logos sit in a row across the slide, evenly spaced and evenly sized. Under the heading, revenue is up sharply from the prior period. Nothing on the slide is false. But the row of logos implies something that was never checked: that five customers are pulling roughly comparable weight. A long logo list cannot establish that revenue or activity is evenly distributed. Only the distribution underneath the total can.

So show the distribution. Name the activity being measured and the period it covers, say how related accounts were grouped, calculate each relevant share against the same defined total, and then describe what is actually known about the relationships behind the largest shares. That is the whole method, and it fits on one slide or one appendix page.

What it produces is a measure of dependence. It does not predict that a customer will leave, it does not establish a percentage above which a business becomes unsafe, and it does not prove that a company is diversified. It shows where the activity came from. The reader can decide what to do with that.

Define the activity and customer unit

Pick one base and name it: recognized revenue, transaction value, usage, seats, gross merchandise value, whatever the slide is actually about. These are not interchangeable, and the difference changes the answer. A company can take a large share of transactions from one participant while earning a small share of revenue from it, particularly in marketplace settings where a take rate sits between the two. Andreessen Horowitz's February 2020 note on marketplace metrics works through concentration using shares of marketplace GMV, and it keeps GMV separate from revenue rather than treating one as a synonym for the other (a16z.com, published February 21, 2020; the sections headed "Concentration or fragmentation of supply and demand" and "Take rate"). That separation is the useful part to borrow. The specific numbers in that piece belong to its marketplace examples, and it does not propose a threshold at which concentration becomes a problem. Don't import one from it.

Then fix the period. A share measured over twelve months and a share measured over the most recent quarter are different claims about the same company, and mixing them in one chart produces a number that means nothing. Use a consistent window and say which one.

Finally, decide what counts as one customer. Account records are not the same as independent customer relationships. A parent company and three subsidiaries may be four records and one relationship. A reseller and the brands it sells to may be one route to market. Agencies, franchisees, and regional operating units raise the same question. Whatever rule you use — shared ownership, shared contracting entity, shared decision-maker — state it, because changing the rule changes the concentration. If two of the logos on the slide turned out to be the same corporate family, the top share would climb without a dollar of new activity appearing.

Reconcile the largest relationships to the whole

Once the base and the grouping are settled, calculate each relevant share against that same total and keep the remainder visible. In a set of one hundred units, if the largest relationship accounts for sixty, the useful information is not just the sixty. It is that everything else accounts for forty.

Two moves to avoid. The first is switching denominators partway through. If the top share is computed against total revenue, the second share cannot be computed against revenue from a subset of segments because that subset tells a friendlier story. Pick the denominator once and hold it. The second is combining periods: a top share from the last twelve months next to a remainder from the trailing quarter. Reconciling means every figure in the view descends from the same records, the same window, and the same grouping rule — so that someone with the source data can reproduce it. A visual estimate of logo sizes is not a measurement.

Confidentiality shapes how much detail belongs on the slide. The reconciled view does not require naming anyone. "Relationship A," a share, and a period give an investor what they need to understand the dependency. Identities can stay in the appendix or out of the deck entirely, subject to whatever agreements govern disclosure.

Explain the relationship behind the share

A share tells the reader how much. It does not tell them what kind of thing the amount is, and that distinction often matters more than the percentage.

Separate ongoing activity from a one-time project. A large share that came from a single implementation, a pilot that ran once, or a seasonal peak is a different fact from a large share that comes from recurring use. Neither is automatically better. A one-time project can be a foothold that leads to a long relationship; recurring activity can sit on a contract that expires in six months. What matters is that the reader knows which one they are looking at.

Then separate contractual facts from expectations. "The agreement runs through next March and renews automatically unless either party gives notice" is a fact the records support. "We expect them to grow with us" is an expectation, and it belongs in language that marks it as one. The same goes for dependencies. If a customer's use of your product depends on a platform that could change its rules, or on a partner relationship that could end, say so where a record supports it and stay quiet where none does. Guessing at a dependency is not the same as documenting one.

What the share alone cannot tell anyone is whether the account will renew, expand, or leave. It is worth saying that plainly rather than letting the chart imply otherwise. A concentration figure is a description of one period's composition, not a forecast.

Keep diversification plans separate from achieved change

If the company intends to broaden its base, that belongs on the slide — in the future tense, with its assumptions attached. "We plan to add accounts in two adjacent segments, which would reduce this share if current relationships hold level" is a plan. "Concentration is decreasing" is a claim about results, and it needs two comparable periods to support it.

Comparable means the definitions and the coverage stayed the same. If the grouping rule changed, if a new segment entered the reported base, or if the period length shifted, the two shares are not measuring the same thing and subtracting them produces noise. When definitions do change — and they sometimes should — say so and show the restated figure rather than a smooth line that hides the seam.

There is no universal threshold to reach for here, and no investor preference that converts into a rule. Some businesses run for years with one relationship at half of activity and manage it deliberately. Others treat that as a problem to solve. Both can be reasonable, and neither is a fact about your company. The benchmark worth applying is the one you can defend from your own records.

The figures themselves need a second set of eyes. Finance owns the revenue definitions; the data owner owns the grouping and the coverage. Someone who did not build the slide should be able to trace every number back to the same source and confirm that the period and the customer units match. That check is not paperwork. It is the difference between a reconciled view and a slide that happens to add up.

A worked example

The following situation, the figures in it, and the relationships described are invented for illustration. No customer records, contracts, or disclosure permissions have been inspected.

A fictional period contains one hundred units of recognized revenue across five independent customer relationships: 60, 15, 10, 10, and 5. The deck shows five equally sized logos.

The largest relationship contributes 60% of the period's recognized revenue. The top two together contribute 75%. The other four relationships contribute 40 units between them.

Two facts about Relationship A's 60 units come from the records. Forty of them are ongoing activity in this period. Twenty come from a one-time project. Neither amount guarantees a future receipt, and the note should say so rather than leaving the reader to assume the 60 is stable.

The relationship note, kept short enough to sit under the table or in an appendix, might read:

Relationship A contributed 60 of the 100 units of recognized revenue in this period: 40 from ongoing activity and 20 from a one-time project. The 40 reflects activity recorded in this period and is not a commitment for the next one. The 20 came from a project classified as one-time; the records presented here neither commit to a repeat nor rule one out. Renewal terms for the ongoing work are stated in the agreement. Relationships B through E contributed 40 units together.

One conditional observation is worth adding, clearly labeled as arithmetic rather than outlook. Set the one-time 20 aside and 80 units of activity remain in the period: 40 from Relationship A's ongoing work and 40 from the other four relationships. Within that remainder, the top account is half rather than 60%. This is not a forecast that the project will not repeat. It is a statement about what the remaining figure is made of, and it is often the most useful sentence on the page.

The diversification effort goes in a separate section, in future tense. Something like: the company plans to add accounts in adjacent segments, aiming to bring Relationship A below half of the defined base over the next four quarters. The assumptions: existing relationships hold at roughly current levels, the new accounts contribute to the same defined base, and the grouping rule stays unchanged. Each of those could fail. If Relationship A also grows, its share may not fall even though the plan succeeded on its own terms. A plan with visible assumptions is useful. A plan presented as a result is a problem.

What the slide should leave the reader with

Sixty percent of one period's recognized revenue came from a single relationship, forty points of it from ongoing work and twenty from a project. That is the honest headline, and it does not cancel the good news. A customer large enough to reach 60% is a real win. It can also be a real dependency. Both are true at once, and a deck that says so is more persuasive than one that shows five logos and lets the reader assume otherwise.

What remains unknown is the part worth naming out loud: whether the ongoing 40 renews on the current terms, whether the project repeats, and what would fill the gap if either stopped. Those questions do not need answers before the deck ships. They need to be asked in the place where an investor will ask them anyway, with whatever the records actually support.

Frequently asked questions

What does a row of evenly sized customer logos fail to show?

It does not establish that revenue or activity is evenly distributed. Only the distribution underneath the total can. The slide should name the activity and period, group related accounts, calculate each share against the same defined total, and describe what is known about the relationships behind the largest shares.

How should the activity base and customer unit be defined?

Pick one base and name it: recognized revenue, transaction value, usage, seats, or gross merchandise value. These are not interchangeable; a marketplace can take a large share of transactions from one participant while earning a small share of revenue from it. Fix the period, and state the rule for what counts as one customer, such as shared ownership or contracting entity.

How should a large share be described beyond the percentage?

Separate ongoing activity from a one-time project, and separate contractual facts from expectations. A share tells how much, not what kind of amount it is. It cannot show whether the account will renew, expand, or leave. Dependencies should be stated only where a record supports them.

Can a diversification plan be presented as achieved change?

No. Plans belong in the future tense with their assumptions attached. A claim that concentration is decreasing needs two comparable periods with the same definitions and coverage. If definitions changed, say so and show the restated figure. There is no universal threshold that makes a concentration level safe.

What does the worked example show?

In the invented period, 100 units are split 60, 15, 10, 10, and 5. The largest relationship is 60%, the top two are 75%, and the other four are 40 units together. Of Relationship A's 60, 40 are ongoing and 20 are one-time. Setting the one-time 20 aside leaves 80 units, where the top account is half rather than 60%.

Further reading

• Sales, Product, and Finance Use the Same Metric Name for Different Things — Use this when finance, sales and product report different totals under the same customer label; it traces each figure to its unit, event, period and owner.
• Several Revenue Streams Make the Total Look Stronger Than the Business Underneath — A complementary view when a total hides its composition: split revenue by stream rather than by customer, and reconcile each component to the reported total.