Is a New Customer Really 5 to 25 Times More Expensive Than Keeping One? Every Source (2026)

So the answer to the title is: not as a measured fact. No source we found measured a 5 to 25 times gap, and the 5x half rests on a credit to TARP that names no study. Its "twin", that a 5% lift in retention raises profits by 25% to 95%, is a different claim with a different source, Fred Reichheld of Bain & Company: in 1990 he and W. Earl Sasser published it as "almost 100%" and "25% to 85%", for five-point drops in defection rate, from a few client examples. Measured studies that came later support "retention matters a lot"; they do not support one universal cost multiple.

This page traces each number to the document it came from, keeps the two claims apart, says plainly where we found nothing, and ends with the one comparison that does apply to you: what it costs to win a new customer against what it costs to bring a past one back through database reactivation. Every source was opened on 29 October 2026.

Where does "5 to 25 times more expensive" come from?

The wording comes from "The Value of Keeping the Right Customers", a short refresher on customer churn that Amy Gallo wrote for Harvard Business Review on 29 October 2014, built on an interview with Jill Avery, a senior lecturer at Harvard Business School. The sentence in full: "Depending on which study you believe, and what industry you're in, acquiring a new customer is anywhere from five to 25 times more expensive than retaining an existing one."

Three things about that sentence matter. It is hedged twice ("which study you believe", "what industry you're in"). It names no study. And it carries no link or footnote: in the page's raw HTML, the opening paragraph holds exactly two links, one to a Bain & Company PDF (for a different claim, the 25% to 95% profit figure covered below) and one to a Net Promoter page. The article's only explanation for the range is the next sentence: "It makes sense: you don't have to spend time and resources going out and finding a new client".

So HBR 2014 is the source of the wording, not evidence for the ratio. That is the confidence we give it in the table below.

Did Harvard Business Review say retaining customers is cheaper than acquiring new ones?

Yes, at least twice with a number attached, and neither time with a study behind it. In the July-August 1990 issue, Christopher W. Hart, James L. Heskett and W. Earl Sasser, Jr. wrote in "The Profitable Art of Service Recovery" that it costs much more to replace a customer than to retain one, "five times more, most industry experts agree". No footnote sits on the five-times line. The article's footnote 1 cites a 1986 report for the U.S. Office of Consumer Affairs (Consumer Complaint Handling in America: An Update Study, Part II, by the Technical Assistance Research Programs Institute), and that marker sits after the complaint figures (how many unhappy customers would buy again). We did not open the 1986 report, so we make no claim about what it contains.

Then, in 2014, came Gallo's hedged "five to 25 times". HBR also published the strongest test we found of the usual explanation for it: in 2002, Werner Reinartz and V. Kumar showed that long-standing customers were not consistently cheaper to serve (see below). If someone cites "Harvard Business Review" for 5-25x, the accurate version is: HBR printed the range as a hedged rule of thumb, and HBR also printed research that cuts against its simplest reading.

Who first said a new customer costs five times more?

The earliest trail we can document leads to TARP, a Washington consumer-research firm. A "Consumer Relation Facts for Business" sheet from the Office of Consumer Affairs of the U.S. Department of Commerce, archived by the UNT Libraries CyberCemetery in August 2000, lists: "Attracting a new customer costs 5 times as much as keeping one. (TARP)". The sheet is undated. It gives no study title, year or method; each line just ends with an agency tag. It shows that a federal office repeated the figure and credited TARP. It is not evidence for the number.

The best trace we found comes from loyalty researchers who had repeated the figure themselves. In "Loyalty Myth #8: It Costs Five Times More to Acquire a Customer than to Retain a Customer", an excerpt from the book Loyalty Myths by Keiningham, Vavra, Aksoy and Wallard (Ipsos Loyalty, 2005), the authors write that "the earliest sources that we can find attribute it to research conducted by the Technical Assistance Research Project (TARP) in Washington, D.C. in the late 1980s." They name the 1990 HBR article as the piece that "lent the statement further credibility", note that Tom Peters repeated it in his book Thriving on Chaos, and add: "We, too, have published prior works repeating this fallacy." (Ipsos expands TARP as Technical Assistance Research Project; HBR's 1990 footnote names the Technical Assistance Research Programs Institute.)

Their objections, in short:

1. Costs get misallocated. Much of the advertising counted as acquisition spend also reaches current customers.
2. The ratio moves. It changes with the stage of the product lifecycle.
3. Customers differ. They vary in what they cost to win and to keep: "Often the most expensive customers to retain are those who generate the most profits for the firm".

“supporting any retention strategy based in whole upon this myth is a recipe for financial disappointment.”

A 2025 trace reached the same place. Marketing effectiveness consultant James Hurman presented research commissioned by Klaviyo, a B2C CRM vendor, to an audience of ecommerce and CX specialists; the Australian Marketing Institute reported it in May 2025. Treat it as corroboration of the Ipsos trace, not as a study: the shopper data behind it (1.7 million online shoppers in Australia and New Zealand) is not published, and the work was paid for by a vendor. His words on the 5x line:

“When we trawled back through where this came from, we actually find there was no empirical study, there was no research, there was no data, there's no evidence at all that acquiring a new customer costs five times as much as keeping an existing one.”

Is there any study behind the 25x figure?

Not one we could find. We could not find any study that measured a 25x gap between the cost of winning a customer and the cost of keeping one. The only document behind the 25 is HBR 2014's "depending on which study you believe".

What we read in full, and none of it gives a 25x ratio: HBR 2014; HBR's 1990 service-recovery article (it says five); the federal TARP fact sheet (5); the Ipsos excerpt (five); HBR's 1990 "Zero Defections" (no cost ratio at all; its only acquisition figure is that credit-card companies "spend an average of $51 to recruit a customer and set up the new account"); HBR's 2000 "E-Loyalty"; Bain's 2001 brief; and Gupta, Lehmann and Stuart's 2004 paper. Web searches for "25 times" with acquire, retain and origin, for "up to 25 times", and for "five to 25 times" with myth or source returned only pages published after October 2014 that repeat Gallo's range. None named an earlier study (one 2023 newsletter credits Bain and Gartner without naming one; see below).

This is an absence within our search, not proof that no such study exists anywhere. We also do not guess where the 25 came from.

Which sources say what? Every study in one table

Every source behind the two claims, oldest wording first, with what each one actually measured. Confidence is High where we read the primary document ourselves, and Medium for news coverage, commentary, vendor-paid research or a figure we only saw cited second-hand.

Does a 5% retention increase really raise profits by 25% to 95%?

This is a separate claim with a separate source, and it travels with the 5-25x line: HBR 2014 puts both in the same paragraph. It belongs to Fred Reichheld of Bain & Company, and its numbers changed as it was retold:

• 1990, the original. Reichheld and Sasser, "Zero Defections: Quality Comes to Services" (HBR, September-October): "Companies can boost profits by almost 100% by retaining just 5% more of their customers." The chart title reads "Reducing Defections 5% Boosts Profits 25% to 85%", and the text gives three examples: "Reducing defections by just 5% generated 85% more profits in one bank's branch system, 50% more in an insurance brokerage, and 30% more in an auto-service chain." The basis was Bain's analyses of client companies' customer costs and revenues; the article does not say how many companies, or how the profit effect was worked out.
• 1996, the book. As the Christian Science Monitor reported Reichheld's The Loyalty Effect: "a 5 percentage point increase in customer retention in a typical firm, he reckons, will increase profits by more than 25 percent".
• 2000, the 95% appears. Reichheld and Phil Schefter, "E-Loyalty: Your Secret Weapon on the Web" (HBR), summarising the 1990 work: "The bottom line: increasing customer retention rates by 5% increases profits by 25% to 95%." This is the earliest 25% to 95% wording we found. The same article recalls that "Those numbers startled many executives, and the article set off a rush to craft retention strategies".
• 2001, the Bain brief. Reichheld's "Prescription for cutting costs: Loyal relationships" (Bain PDF, about 1,000 words): "In financial services, for example, a 5% increase in customer retention produces more than a 25% increase in profit." There is no 95% in it, and no acquisition-versus-retention cost multiple.
• 2014, the citation. HBR's Gallo credits "research done by Frederick Reichheld of Bain & Company" for "increasing customer retention rates by 5% increases profits by 25% to 95%", and links that 2001 brief.

So the most-cited version of the twin has never been in the document it is linked to. Peer-reviewed work kept the original wording: in 2004, Pfeifer and Farris (Journal of Interactive Marketing) wrote that "A cornerstone of the customer relationship management (CRM) and interactive marketing literature is the Reichheld and Sasser statement that 'reducing defections by 5% boosts profits 25% to 85%.'", and set out to give managers "a more exact quantification of the economic benefits of increased customer retention."

What does the "5%" actually mean?

Five percentage points of defection rate, not a 5% relative change. Going from 10% to 5% defection keeps 5 more customers in every 100, but it halves the defection rate. In the 1990 credit-card example, "as the credit card company cuts its defection rate from 20% to 10%, the average life span of its relationship with a customer doubles from five years to ten and the value of that customer more than doubles". Then: "As the defection rate drops another 5%, the average life span of a customer relationship doubles again and profits rise 75%".

Marketing scientist Byron Sharp of the Ehrenberg-Bass Institute points at that second step: "Their 5% drop in defection is actually a drop of 5 percentage points, ie from 10% to 5%", which is "a halving of customer defection", and the authors "simply assumed that the halving of defection is achieved at zero costs". He calls the work "merely a 'thought experiment'" and writes that the article "does not empirically show that lower defection rates cause massive increases in firm profits." That is his reading. The 1990 article does report client examples (a bank's branches, an insurance brokerage, an auto-service chain); it does not disclose how many companies were analysed or how. Both are true: case examples, method not disclosed.

How did the misquotes spread?

Each common version has a traceable starting point. The pattern repeats: a hedged rule of thumb picks up a famous name and loses its hedge.

The percent version is the one to watch for. The student blog that paraphrased Gallo wrote that acquisition is "anywhere from 5% to 25% more expensive than retaining an existing customer". Five to 25 percent more and five to 25 times more are different claims, and neither comes with a study. The 2023 newsletter states that "Research from various sources, including Bain & Company, Harvard Business Review, and Gartner, suggests that acquiring a new customer can cost anywhere from 5 to 25 times more than retaining clients", and names no study from any of them.

Are loyal customers actually cheaper to serve?

Not reliably, in the best test we found. Werner Reinartz and V. Kumar's "The Mismanagement of Customer Loyalty" (HBR, July 2002) compared "the behavior, revenue, and profitability of more than 16,000 individual and corporate customers over a four-year period" at four companies: a high-tech corporate service provider, a large U.S. mail-order company, a French retail food business and a German direct brokerage house. Their finding: "in none of the four companies we tracked were long-standing customers consistently cheaper to manage than short-term customers."

The link between how long customers stayed and how profitable they were was "weak to moderate in all four companies we studied, with correlation coefficients of 0.45 for the grocery retailer, 0.30 for the corporate service provider, 0.29 for the direct brokerage firm, and just 0.20 for the mail-order company." More broadly, they found "little or no evidence to suggest that customers who purchase steadily from a company over time are necessarily cheaper to serve, less price sensitive, or particularly effective at bringing in new business."

This matters because "existing customers cost less to serve" is the usual explanation for why keeping them is cheap. Four companies are not every company, and the study measured serving costs, not acquisition costs. But it is a measurement, and the 5-25x range does not have one.

Is keeping customers still worth it? What measured studies found

Yes, and the measured research makes the case better than the slogan does. In "Valuing Customers" (Journal of Marketing Research, February 2004), Sunil Gupta, Donald R. Lehmann and Jennifer Ames Stuart valued five firms from public data (Capital One, Amazon, Ameritrade, eBay and E*Trade). Their result: "improving customer retention by 1% is likely to improve customer and firm value by 3-7%. In comparison, margin elasticity is about 1 and acquisition elasticity is only 0.02-0.3." In plain terms, in these five firms a 1% gain in retention moved value far more than a 1% cut in the cost of winning a customer.

Two caveats come from the paper itself. It works out acquisition cost by "dividing the total marketing cost by the number of newly acquired customers for each time period", and says: "Although some of the marketing cost is incurred for retention purposes as well, we do not have information to separate the two costs." And its line that "customer acquisition costs are significantly higher than customer retention costs" cites Reichheld's 1996 book, which is not a published study of costs, plus one estimate: "Thomas (2001) estimates acquisition cost per customer as $26.94 versus retention cost per customer of $2.15." That is about 12.5 to 1. We could not open Thomas's paper, so the figure is known to us only as cited by Gupta and colleagues, and we give no industry or sample for it (confidence: Medium).

The 2005 follow-up by Werner Reinartz, Jacquelyn S. Thomas and V. Kumar (Journal of Marketing 69(1); INSEAD working paper) studied "23,476 prospects and 5,725 customers of a single firm", a high-tech company. It found that "a suboptimal allocation of retention expenditures will have a larger impact on long-term customer profitability than suboptimal acquisition expenditures", that "it is more detrimental to underspend than overspend from the optimal budget", and that "the profit maximizing strategy does not maximize the acquisition rate nor the relationship duration."

What is keeping an existing customer worth compared with winning a new one?

Work it out for your own business. It takes two numbers and a calendar:

1. Cost to win one new customer. Take everything spent on marketing and sales in a period and divide it by the new customers won in that period. This is the definition Gupta, Lehmann and Stuart use, so keep their caveat in mind: some of that spend also reaches existing customers, which is the Ipsos authors' first objection too.
2. Cost to bring one past customer back. Take the full cost of one reactivation run, staff hours on the phone plus any offer, and divide it by the past customers who bought again within a set window, for example 90 days. Count the staff time: it is usually the real limit.
3. Compare the two. That ratio applies to you. HBR's range does not.

A home-services example: for a heating and plumbing firm, the first number is a quarter's ads, quoting time and sales cost divided by the new customers won that quarter. The second is the hours spent calling past customers whose boiler service is due, plus any offer, divided by the ones who booked within 90 days. We give no figures here on purpose; yours are the only ones that count. To size the list before you start, use the dormant-lead sizing method or the dormant database revenue calculator, and for the trade-off against paying for fresh leads, see lead reactivation vs buying new leads.

Who to call first, and what to offer

Win-back research gives two practical rules. Call the strongest past relationships first: Kumar, Bhagwat and Zhang (Journal of Marketing, 2015) found that "the stronger the first-lifetime relationship with the firm, the more likely a customer is to accept the win-back offer." And treat the price to come back as its own decision: in Thomas, Blattberg and Fox (Journal of Marketing Research, 2004), "the optimal pricing strategy for their application involves a low reacquisition price and higher prices when customers have been reacquired." The same authors argue that "customer winback should also be an important part of a customer relationship management strategy."

Kumar and colleagues also note that "some pertinent questions remain as to whether lost customers are worth the investment in reacquisition and whether they will remain profitable if reacquired." That is why the sum above comes first: it answers the question for your list, not for a study's.

Why do so few businesses work their past-customer list?

In most businesses the past-customer list already exists, in the CRM or the booking system. What runs out is team hours to call it: the people who would make those calls are busy serving today's customers. That is a capacity problem, and it is where AInora fits. AInora's AI voice agents make the reactivation calls the team can't get to, log each outcome in the CRM, and hand the conversations that need a person to a person. The same team reaches more past customers.

Where this is set up: database reactivation for a whole dormant list, service-due customer calls, overdue veterinary clients and old HVAC and plumbing customers. For the step-by-step, read how to reactivate lost customers with AI voice outreach; for how the two channels compare in practice, the reactivation cost comparison.

If you want to see how a reactivation run would work on your own list, book a free consultation.

How did we check these sources?

Every source was opened on 29 October 2026. We read each HBR article in full; quotes are kept short and the links go to the HBR pages, which may ask for a subscription. The Bain, Ipsos, Columbia and INSEAD PDFs were downloaded and read as text. Journal abstracts come from each DOI's Crossref record. Where a number had no source, we left it out or said so. We ran the same kind of audit on lead response time statistics and webinar attendance statistics.