What the customer relationship costs while it stays

Ask a revenue leader what their best salesperson’s customer relationships are worth and you will hear about the day she resigns. The fear has furniture: the book she could take, the accounts that answer her mobile and nobody else’s, the competitor who is not really hiring a person but acquiring a portfolio. Fourteen years ago, in Harvard Business Review, Andris Zoltners and colleagues asked the question by name: who owns your customer relationships, your salespeople or your company? The discipline of answering it has been exit-shaped ever since: bench hiring, multiplexed contacts, CRM capture, pay restructured toward salary and stock.

Here is what that discipline misses, and what this essay adds. The tie your firm pays to protect is not idle while everyone stays. It works, and one of the things it does, in the one setting where selection and severance were measured together, is choose your customers, against you. The evidence for the exit fear, meanwhile, turns out to be thinner than its grip on the discourse suggests: what actually follows a departing salesperson has been measured as behavior exactly once, across twenty-nine defections. This piece takes the three studies that measured what happens when the tie breaks, shows that their apparently contradictory verdicts are measurements of three different channels, and finishes the decision guide one of them started: with the two columns none of them funds.

Which three distinct cost dimensions govern active enterprise customer accounts?

Three figures circulate as the price of a salesperson-owned relationship, and stacked side by side they look like a range. They are not. They are three different kinds of knowledge wearing the same percent sign.

26% is a stated intention. In the study everyone cites for salesperson-owned loyalty: Palmatier, Scheer and Steenkamp, 362 buyer–salesperson dyads in B2B distribution: buyers were asked what share of their purchases they would try to shift to their salesperson’s new firm if she defected. The average answer, over a three-year horizon: 26%. It is a survey answer about a hypothetical, from the buyer’s own mouth. The paper’s one behavioral result compares twenty-nine actual defections: sales growth lower where the salesperson left for a competitor, reported significant at t = 1.72: too few to carry the moderation the authors themselves wanted to test.

(An American Express estimate that 30% of customers would follow their financial advisor circulates through the same paper’s introduction: provenance-grade, not measurement.) The study’s strongest number: salesperson-owned loyalty predicting the selling firm’s latent financial risk at a coefficient of .62, explaining more variance than anything else in the model: runs between two constructs reported by the same buyers in the same survey, whose items nearly restate one another; the authors themselves flag that a common source can inflate such coefficients. The fear is real, and the best evidence for it is intention-shaped.

13.2–17.6% is a measured loss: with the defection channel switched off. Shi, Sridhar, Grewal and Lilien tracked 830 customers of a Fortune 500 electrical-components distributor through their salesperson’s departure, against 1,615 matched controls. Annual customer sales fell 13.2–17.6%. But read the setting before you quote the number: the firm’s noncompete legally barred departing reps from taking customers with them. Nothing “followed the rep out.” This is the cost of the transition itself: the severed working knowledge, the half-finished cycles, the new face, not of defection. And the loss has an address: customers reassigned to new hires lost 21.6%; customers reassigned to existing reps lost 11.0%, a figure statistically indistinguishable from zero. Replacements whose books resembled the departing rep’s brought the loss to statistical zero; the replacement’s past performance, the thing managers select on, did nothing measurable.

95% is a self-report. The brokerage figure: top producers who “can bring 95% of their business with them”: comes from managers describing their own firms in Russell Coff’s case material, as the essay on the asset that can leave already framed it. It is a claim by interested parties about their most extreme cases, and it has done more work in retention negotiations than any measured number.

Intention, transition loss, self-report. The quantity everyone actually fears: the share of revenue that walks out as behavior when a relationship-owning salesperson defects without a noncompete: appears in none of the three. The exposure column of this decision is unpriced. That is not a gap in this essay; it is one of its findings.

What do empirical studies conclude regarding terminating unprofitable customer accounts?

Now the other side of the ledger: what happened when researchers measured firms severing salesperson–customer ties on purpose or by circumstance. Three studies, three verdicts, each welded here to its setting because the settings are the finding.

At the distributor (noncompete; the selection channel unmeasured), severance was a pure cost: the 13–18% above, concentrated in new-hire reassignments.

At a microfinance bank in Mexico: 461 loan officers, 129,839 loans: Kim, Sudhir, Uetake and Canales measured the opposite. The bank randomly transfers officers between branches, deliberately destroying their customer relationships. Transfers hurt collections, as the exit literature would predict: repayment on the transferred portfolio worsened: a weaker leg, significant only at the ten-percent level. And the bank still came out ahead: transferred officers originated visibly better loans, and total value created per officer-month rose. The reason is the piece of this literature the exit folder has not absorbed: Kim and colleagues themselves name the tension with the departure studies and leave it open: loan officers both own relationships and choose customers. Their relationship-derived private knowledge: who is good for the money, who merely looks it: is exactly what a credit score cannot see. Under acquisition pressure, officers used that knowledge against the bank, signing customers who looked fine on paper and were privately known to be marginal: one point of acquisition performance cost about half a point of return on the new loans it produced. A compensation stake in later repayment moderated the selection; without it, the authors conclude: from an interaction the paper reports at the ten-percent level: the acquisition bonuses were buying the bank almost nothing.

At a European logistics firm: 2,040 B2B customers through salesperson changes, in the study whose decision guide this essay completes: Schmitz, Friess, Alavi and Habel split revenue in two and found the verdict depends on which half you watch. Resale revenue, the repurchasing of what the customer already bought, fell 28.8%. New-sale revenue: products this customer had never bought: rose 52.2%, because the incoming salesperson re-explored needs the departed one had stopped asking about.

Before this reads as an argument for shuffling your sales force annually: the average total effect was negative, −6.8%; the total turned positive only in measured combinations: favorable purchase dynamics alone yielded about +10.7% at average relationship strength, strength alone yielded no average gain, and the largest totals needed the right replacement behavior on top; the new-sale gain concentrated where the incoming rep actually was a broader cross-seller: the disruption effect on new sales loses statistical significance once the replacement’s behavior enters the model; and the authors’ own warning deserves quoting whole: “We strongly urge managers to avoid the conclusion that proactively disrupting an interpersonal relationship is a certain route to increased revenue.”

How do three structural operational channels explain customer cost divergence?

Read as one literature, these verdicts contradict. Read as measurements of different channels, they route.

Continuity is what Shi measured: the working knowledge in the tie, lost in any transition, mitigated by similar-book replacements and by not handing the account to a novice. Every funnel has this channel.

Selection is what Kim measured: the tie’s private knowledge deciding which customers enter. This channel exists only where a salesperson has discretion over entry that nothing re-screens. Kim’s loan officers held wide discretion over who entered: they recruit, take the application and recommend the loan, and the bank’s approval runs on observables the officer can see past. That discretion is the precondition, and it is the scope condition of this essay.

Revitalization is what Schmitz measured: the value of re-asking questions a settled relationship has stopped asking, realized only where there was unexplored potential and a replacement capable of exploring it.

These channels do not sum, and this essay will not add them into a number: the designs overlap (a maintenance loss in Kim’s bank and a resale loss in Schmitz’s logistics firm are plausibly the same loss twice), and a severance that ends selection may also end the trust revitalization needs. What the channels do is route. The question is never “is breaking ties good?” It is: which channels does my funnel have?

What constitutes the unmeasured standing maintenance cost of enterprise accounts?

The practitioner literature has circled this. The 2011 ownership piece named three standing costs of rep-owned relationships from anecdote: complacency in owned accounts, pay levels that outrun contribution, and salespeople blocking reorganizations their book gives them the power to block. This January, also in HBR, Janssen, Denenberg and Shapiro gave the acquisition version a name: “sales debt,” the accumulated cost of signing poor-fit customers under revenue pressure, and prescribed harder qualification and comp that only rewards the right customers.

Set the measured record beside the folk one and the map is instructive. The sales-debt complaint has a measured mechanism, and it is Kim’s: where reps hold private knowledge of customer quality and discretion over entry, wrong-fit acquisition is not a discipline failure, it is adverse selection: the tie itself doing the choosing. The comp fix the practitioners reach for: counting only right-fit customers toward quota: is the observable-quality weighting Kim and colleagues say fails exactly where quality is privately known. What the bank’s data validates is different: a stake in what the customer does after signing. Complacency, meanwhile, maps to Schmitz’s finding that strong, settled relationships are where disruption pays on the new-sale side: though routine itself cuts both ways in his tables. And hold-up: the rep whose ownership blocks change: has no measured study at all. Some of the folk wisdom was ahead of the evidence; some of it still is.

What actionable diagnostic guide completes the account maintenance audit?

Schmitz and colleagues end their paper with a decision tree for the keep-or-disrupt call: evaluate the relationship’s strength and untapped potential, then manage retention or expansion accordingly, with a carefully hedged branch for proactive disruption. It is the right artifact with two columns missing. Here is the completed version, every row a question your systems can answer, no coefficient imported from anyone else’s setting.

Column	The question	Where the answer lives
Continuity (Shi)	If this tie broke tomorrow, does a replacement with a similar book exist in-house, or does the account go to a new hire?	Your CRM: industry mix per rep’s book
Revitalization (Schmitz)	Has this customer’s complex-purchase share plateaued while the breadth of purchased lines still grows? That pairing is Schmitz’s measured favorable case: his tenure check found no old-versus-young difference, and he excluded ordering variability as countervailing	Order history: complex-product share and count of distinct product lines, per quarter
Selection (Kim)	Does any salesperson choose which customers enter, and does any screen see what the closer sees? In Kim’s bank, approval on observables did not close the channel. Does the closer’s pay carry a stake in what the customer does after signing?	Your approval workflow and comp plan: both documents you hold
Exposure (unpriced)	What share of the customer interface does this rep personally hold? And what would actually follow them out?	Interface share: calendar and CRM contacts, one afternoon. The second question: measured once, across twenty-nine defections. Palmatier’s two-item buyer survey is the only purpose-built instrument in print: you could field it

The routing follows from the columns. A funnel where no salesperson picks the customer: inbound-qualified, product-led: has no selection channel, and there the exit literature governs: protect continuity, groom similar-book successors, treat “disruption pays” stories from lending as travel writing. A motion where reps qualify their own deals and nothing re-screens them is running Kim’s experiment without Kim’s controls, and the first question is not whether to rotate anybody; it is what stake the closer has in month twelve: the same discretion that lets a discount outlive the deal it closed. And a book whose complex-coordination share has plateaued while its line breadth still grows is Schmitz’s measured favorable case, where the brave conversation is not about the rep who might leave but about whether the relationship’s next salesperson question gets asked while she stays.

Everything above stops at its evidence. The three studies are one distributor, one bank, one logistics firm; none is your firm, which is why the guide imports questions and not numbers. Repeated deliberate disruption is unstudied: nobody knows the scar tissue it builds. And the exposure cell stays honestly empty: the most feared number in sales management is still, in 2026, a survey answer and twenty-nine cases. The relationship your firm is paying to keep is choosing customers, holding pay, and blocking reorganizations while it stays, and the literature has priced those channels better than it has ever priced the leaving. That inversion is the finding. What the exit actually costs is a question someone should finally measure; what the staying costs, you can now look up: column by column, in your own systems.

Where are the empirical limits of customer maintenance cost analysis?

Boundary. The coefficients describe a bounded relationship model, not a portable account-risk score. Measure actual exit exposure and customer continuity before converting the finding into integration terms. For portfolio margin dynamics, remember also that long customer relationships can be low-profit, as service costs often escalate over account tenure.

Evidence base. The analytical frame also draws on these additional sources: Janssen et al. 2026; Zoltners et al. 2011; Zoltners et al. 2017. The links identify the exact works; they support the mechanisms and boundary conditions discussed here, not every claim in isolation.

References

1. Kim, M., Sudhir, K., Uetake, K., & Canales, R. (2019). When salespeople manage customer relationships: Multidimensional incentives and private information. Journal of Marketing Research, 56(5), 749–766. https://doi.org/10.1177/0022243719847661
2. Palmatier, R. W., Scheer, L. K., & Steenkamp, J.-B. E. M. (2007). Customer loyalty to whom? Managing the benefits and risks of salesperson-owned loyalty. Journal of Marketing Research, 44(2), 185–199. https://doi.org/10.1509/jmkr.44.2.185
3. Schmitz, C., Friess, M., Alavi, S., & Habel, J. (2020). Understanding the impact of relationship disruptions. Journal of Marketing, 84(1), 66–87. https://doi.org/10.1177/0022242919882630
4. Shi, H., Sridhar, S., Grewal, R., & Lilien, G. (2017). Sales representative departures and customer reassignment strategies in business-to-business markets. Journal of Marketing, 81(2), 25–44. https://doi.org/10.1509/jm.15.0358
5. Janssen, E., Denenberg, B., & Shapiro, B. P. (2026, January 7). The risks of prioritizing short-term revenue over customer fit. Harvard Business Review. https://hbr.org/2026/01/the-risks-of-prioritizing-short-term-revenue-over-customer-fit
6. Zoltners, A. A., Sinha, P., & Lorimer, S. E. (2011, December 21). Who owns your customer relationships: Your salespeople or your company? Harvard Business Review. https://hbr.org/2011/12/who-owns-your-customer-relatio
7. Zoltners, A. A., Sinha, P., & Lorimer, S. E. (2017, November 17). How to reduce the costs of salesperson turnover. Harvard Business Review. https://hbr.org/2017/11/how-to-reduce-the-costs-of-salesperson-turnover