The Concentration Discipline

Most of what circulates online about ideal customer profiles rests on a small number of statistics that get repeated without a source attached. We went looking for the real evidence instead, and the strongest case for a defined ICP turned out not to be a marketing stat at all.

Rule of thumb: the case for focus does not need an inflated statistic to be convincing.KEY TAKEAWAY

Customer value in B2B is never spread evenly across an account base. Research on customer profitability dating back to a landmark 1991 Harvard Business Review study found that a company’s most profitable 20% of customers can generate more than double its total profit, while its least profitable accounts erode a large share of it back. Current B2B case studies show the same pattern in modern go-to-market motion. When targeting and scoring are built around a defined profile, companies report materially higher win rates and engagement, with named examples ranging from 25% to 90% depending on the metric.

THE CHALLENGE

Most companies do not choose a narrow customer on purpose. They arrive at a broad one by accumulation. Every year adds a persona, a vertical the sales team wants to chase, a feature built for a single large logo. Nobody makes an explicit decision to serve everyone. It happens because saying no to any one deal feels like leaving revenue on the table.

The cost of that drift rarely shows up on a quarterly scorecard. It shows up as marketing that has to work harder to explain the same product to five different audiences, a sales team that re-qualifies from scratch on every call because there is no shared definition of a good-fit account, and a product roadmap pulled toward whichever customer is loudest rather than whichever is most valuable.

The evidence for what this actually costs is older and more rigorous than most people assume, and it does not require citing a viral, unsourced statistic to make the point.

WHAT THE EVIDENCE SHOWS

Two bodies of evidence, one nearly 35 years old and one current, describe the same pattern.

In 1991, Robert Kaplan and Robin Cooper published research in Harvard Business Review that has held up for more than three decades. Using activity-based costing to look inside real companies’ customer bases, they found that profit is never distributed evenly across an account list. At Kanthal, a Swedish heating-wire manufacturer that became the canonical example, the most profitable 20% of customers generated 225% of the company’s total profit, and the least profitable 10% gave back 125% of it. The company’s actual reported profit sat in between, propped up by a minority of accounts and dragged down by a minority on the other end. Kaplan and Cooper called the resulting chart a whale curve, and researchers have since found the same basic shape at company after company, in different industries and different decades. The proportions vary. The shape almost never does.

The same concentration shows up in current B2B go-to-market data once a company scores accounts against a defined profile instead of treating every lead the same. Snowflake built an account propensity model using more than 70 data points to identify which accounts were structurally more likely to buy, then focused sales attention there. Once the highest-scoring accounts were prioritized, Snowflake saw a 25% increase in customer engagement, twice the new-customer conversion rate, and a 90% higher opportunity open rate on those accounts compared with unscored outreach. Smartsheet ran a similar exercise, rebuilding its highest-volume demo form and targeting around clearer segmentation and intent data. The result was an 84% increase in marketing qualified leads, a 26% increase in opportunity rate, and a 59% increase in win rate on that channel.

Separately, Factors.ai, a platform built around ICP scoring, reports from its own client base that B2B teams see 20 to 40% higher win rates and 15 to 30% shorter sales cycles once ICP scoring is built into day-to-day pipeline routing rather than kept as a descriptive document. We are citing that as one platform’s reported client data, not as an independently audited study, because that is what it is.

Set the case studies next to the 1991 research, and a mechanism becomes visible, not just a coincidence. Every account in a pipeline is not an equally good bet, and the businesses that measure the difference, whether through activity-based costing three decades ago or account scoring today, consistently find that a defined, well-served core drives a return disproportionate to its size. The businesses that don’t measure it are not exempt from the pattern. They simply cannot see it.

What It Means for Leaders

1. Your account base already has a whale curve, whether or not you have measured it. Every business with more than a handful of B2B customers has some version of Kaplan and Cooper’s pattern sitting inside its own CRM. Not measuring it does not make it disappear. It means decisions about where sales and marketing spend their time are being made without seeing it.

2. Scoring beats describing. A written ICP that lives in a slide deck does not change behavior on its own. The case studies above involved building a scoring or segmentation system that sales and marketing actually used to prioritize accounts day to day, which is a more operational commitment than agreeing on a persona once a year.

3. The bigger gain is usually on the top end, not just at the bottom. It is tempting to treat ICP discipline mainly as a way to stop wasting time on bad-fit leads. The larger, more reliable gain in both the 1991 research and the current case studies came from concentrating more attention on the best-fit accounts, not just filtering out the worst ones.

RECOMMENDED NEXT STEPS

• Action to take now: Run a simple profitability or win-rate cut of your existing account base by segment or tier, even a rough one, to see whether your own whale curve looks the way Kaplan and Cooper’s research predicts.
• Decision or assumption to validate: Whether your current ICP is a scoring system that changes day-to-day prioritization, or a description that mainly exists for onboarding new hires.
• Metric or signal to monitor: Win rate and opportunity engagement split by account score or tier, tracked separately rather than blended into one company-wide number.

CONCLUSION

The clearest decision here does not require trusting an unsourced statistic. Kaplan and Cooper’s research is more than thirty years old and has been replicated across industries since, and current case studies from Snowflake and Smartsheet show the same pattern operating inside modern go-to-market motion. The decision that matters is whether your organization has turned that pattern into an operational scoring system, or whether it is still finding out where its own whale curve sits by accident, one quarter at a time.

Dimension	Why it matters
Customer profitability concentration	Top 20% of customers generated 225% of profit, bottom 10% eroded 125% of it, at Kanthal (Kaplan & Cooper, Harvard Business Review, 1991)
Account scoring vs. blended outreach	Snowflake saw a 90% higher opportunity open rate and 2x conversion rate on accounts prioritized by propensity score (ZoomInfo case study)
Segmentation and targeting	Smartsheet saw a 59% increase in win rate and 26% increase in opportunity rate after rebuilding targeting around defined segments (ZoomInfo case study)
Documented, scored ICP (platform-reported)	B2B teams report 20-40% higher win rates and 15-30% shorter sales cycles when ICP scoring is operationalized (Factors.ai client data)
Written vs. operational ICP	A described persona changes onboarding decks. A scoring system changes which accounts get sales and marketing attention this week.