What to sell the customers you already have, and how to tell if it’s working.
Every brand with a hit product reaches the same meeting. New customers are getting expensive, the base is loyal, and someone puts up a slide: customers who buy from two categories are worth three times as much. So let’s give them a second category.
Those two numbers frame the guide. The first says the next category is harder than it looks from inside the meeting: most moves beyond the core fail, even when chosen by people paid to choose well. The second says the slide is missing a column. Some cross-buyers are your best customers and some are your most expensive: they buy on discount, return what they buy and call support about it. Push them into a third category and you lose more.
None of this means don’t expand. It means expand in the right order: first the headroom left in the core, then the nearest step out, launched to the customers who already trust you, and measured against a holdout instead of a slide.
The best next category is often more of the first one. The second best is the one your customers would guess you’d make.
It builds on The Second Order (the second purchase, cohorts, VIPs) and The First Offer (bundles, basket analysis) without repeating them. This guide is about the decision above both: which category to add, whether to add one at all, and how to tell afterward whether it helped.
Start with The Next-Category Audit. Your lowest checks name the chapters to read first. Or follow a path:
Four tools run in the page. Nothing you type leaves your browser.
Examples that open with Say or Picture use made-up round numbers. Every source is listed in Appendix C.
What this guide argues, and what would prove each claim wrong.
A position says what would prove it wrong. Test each on your own customers.
Four rings around the core. Each ring out shares less with what you already do well, and asks for more proof before you commit.
Picture a coffee brand whose hero is a whole-bean subscription. Here’s what each ring looks like for it, and what should be true before it moves there.
The whole book is above and always will be. These are the same chapters addressed individually, for linking to one idea rather than to ninety.
| Ring | What’s shared with the core | For the coffee brand | Move when | How it goes wrong |
|---|---|---|---|---|
| 0. The core | Everything | More of the hero: sizes, blends, cadence, a better second order | Always first, above all while share of wallet is under about half | Nobody measures share of wallet, so the headroom is invisible |
| 1. One step | Customer, occasion and proof; maybe not supply | Filters, a grinder, cold-brew concentrate | The core’s share of wallet is high, and customers already buy this somewhere | Launched as a discount to the whole list, so it attracts the wrong buyers |
| 2. Two steps | Customer, and one of occasion, supply or proof | Tea, mugs, a branded kettle | Ring 1 is working and a partner can’t do it better | Customers can’t see why you’d be good at it, so they buy once |
| 3. A new business | Little beyond the name | Energy drinks, apparel, a café | Rarely: when the core is capped and the move has its own economics | It borrows the brand, the cash and the team’s attention, and dilutes all three |
Rings are about distance, not product type. Chapter 4 turns them into a score.
Every ring out adds one more thing that has to go right the first time.
Twelve checks on whether you’re ready to add a category, and whether the ones you’ve added are paying. About forty-five minutes with your order data, your last survey and your flows.
The audit isn’t about how many categories you sell. It asks whether you know how much of your customers’ spending you already have, whether you chose your last expansion by distance or excitement, and whether your multi-category numbers are true.
Open your order data, your customer survey results if you have any, your email and SMS flows, and the plan for your next launch. Score each check 0 to 2: 0 if it failed or nobody can answer it, 1 if partly true, 2 if clean. A slide that says “multi-category customers are worth 3x” doesn’t count as an answer to any of these.
A category you can’t measure is a category you can’t defend in the next budget meeting.
Score as you go; your band appears when all twelve are in.
| Score | What it means | Read next |
|---|---|---|
| 20–24 | You expand on evidence. Your job now is picking the next step and launching it to the base with a kill rule. | How Far Is One Step?, then Launch to the Base First |
| 14–19 | Some of what you believe about your multi-category customers is true and some isn’t. Fix the zeros before the next launch. | The chapter linked from your lowest check, then The 3x Customer |
| 8–13 | You’re choosing categories on excitement and grading them on a slide. Measure the core before adding to it. | Part one, starting at Headroom Before Horizon |
| 0–7 | Pause the next launch for a month. Measure share of wallet and contribution per customer first. | Headroom Before Horizon, then The First Thirty Days |
If you sell one category today, checks 6, 7 and 9 score 0 and that’s fine: they’re about categories you haven’t added yet. Look at the other nine instead. A clean 18 on those nine means you’re ready to choose.
Before you sell customers something new, find out how much of their spending on the old thing still goes to someone else.
A customer who buys your coffee every month looks loyal in your data. Your data can’t see the two other brands in her cupboard. Share of wallet, the share of a customer’s category spending that comes to you, is the first number to know before adding a category, because it says how much growth is left in the one you have.
The obvious guess is that your biggest spenders are loyal and your small spenders are the opportunity. Rex Du, Wagner Kamakura and Carl Mela found in 2007 that it doesn’t hold: the volume of customers’ transactions with the firm had little correlation with the volume of their transactions with competitors, and a small share of customers accounted for a large part of all the transactions that went elsewhere Published. Some of your modest customers are big spenders in the category who give most of it to someone else. You can’t find them in your order data. You have to ask.
Timothy Keiningham and his colleagues went looking for the survey question that predicts share of wallet. Satisfaction and Net Promoter scores, they wrote in Harvard Business Review in 2011, “correlate poorly with what matters most: share of wallet” Published. What did predict it was rank: where a customer places your brand among the brands they buy in the category. Their Wallet Allocation Rule turns that into a share with two inputs, your rank and the number of brands the customer uses:
share = (1 − rank ÷ (brands + 1)) × (2 ÷ brands)
For a customer who buys three brands: first gets 50%, second 33%, third 17%. The shares always add to 100%.
PublishedKeiningham, Aksoy, Williams and Buoye, The Wallet Allocation Rule, 2015. The authors describe the underlying work as a two-year study of more than 17,000 consumers in more than a dozen industries and nine countries Reported.
DerivedFrom the formula above.
Treat the rule as a starting point, not a law. Most of its evidence comes from the authors’ own studies, and it assumes customers can rank brands they buy for different reasons. Check it on your data: survey a few hundred customers on category spend and rank, and compare their stated share with what they spend with you. If the two roughly agree, rank becomes your share-of-wallet tracker. The survey is in Appendix B.
A customer who ranks you second of three gives you about a third of their spend, and feels loyal while doing it.
Say a brand has 20,000 active customers. Its survey says a typical customer spends $300 a year in the category, uses three brands and ranks this one second. The brand gets about $90 a year from each, a 30% share, a little under the rule’s 33% for second place. If it could move three in ten customers from second to first, their predicted share would go from 33% to 50%: about $50 more per customer a year, or $300,000 across 6,000 customers Derived. No new product, no new inventory, no new supplier.
If the next category’s plan says $250,000 in year one, before development and stock, the core is the bigger opportunity, and the cheaper one.
With the defaults, the brand holds a 30% share, the rule predicts 33% for second place, and moving three in ten customers to first is worth $300,000 a year, 1.2 times the new category’s year-one plan.
You move up a rank by finding out why customers put someone else first, and removing the reason. Ask the customers who rank you second what the first brand does better. The answers tend to be practical: a size you don’t sell, a worse price per unit, a flavor you discontinued. Those are core fixes, usually cheaper than a launch. The mechanics of later orders are in The Second Order.
The best-known evidence for “grow from a strong core, one step at a time” comes from Bain & Company. It’s useful, and it was drawn from winners. Here’s how much weight it can carry.
Chris Zook, then a partner at Bain, built a trilogy on one idea: companies that sustain profitable growth usually have a strong core and expand from it into the next ring out, repeatedly, rather than leaping. Profit from the Core (2001, with James Allen) and Beyond the Core (2004) are still the best-known books on the subject.
Notice that last one. It’s another “3x,” and it has the same weakness as the multi-category customer’s 3x in chapter 7: the companies that found a repeatable formula were probably stronger before they found it.
Bain is a consultancy that sells growth strategy. I can’t find where it has published the data, sample rules or method behind these figures, so they can’t be checked. The figures shift between tellings: 11% or 13%, one in four or one in five. Preston Smith, reviewing Beyond the Core for the Product Development and Management Association, pointed out that the companies in Bain’s database were “most likely Bain clients and probably are not selected randomly,” that the interviews and cases were a convenience sample, and that the book “does not reveal the research design, methodology, data, or actual results of the analyses” Reported.
The deeper problem is survivorship. Much of the method starts with the companies that grew and asks what they did. Beyond the Core drew on profiles of about 100 companies with the best growth records and interviews with 25 of their CEOs Reported. Companies that took the next step and failed, or stayed close and stalled, are much harder to see. And distance is judged after the fact: a move that worked looks close in hindsight, and one that failed looks like a stretch. Phil Rosenzweig called this the halo effect: success colors how we describe the strategy that preceded it Published.
A study of winners tells you what winners did. It can’t tell you what the losers did differently.
One example makes the point. Profit from the Core, published in 2001, used Enron as one of its examples; a 2008 Businessweek review of the book noted that Enron was bankrupt by the end of the year the book came out Reported. By 2004, Bain’s own article listed Enron among “the 25 largest business disasters in the past five years,” 18 of which it said were “rooted in major adjacency moves gone awry” Reported. The same company was a core-strategy example in 2001 and an adjacency disaster in 2004. The story changed because the outcome did.
The direction is still probably right, because it agrees with evidence that doesn’t come from Bain. Brand extension research, in chapter 5, finds that fit between the old product and the new one is the strongest single driver of an extension’s success. Scanner-data research in chapter 11 finds that experience with the parent brand drives trial of the extension. Both say the same thing Zook says, from data anyone can inspect: the closer the next step, the more of your existing trust it can use.
So take the adjacency record as a prior, not a law. Stay close unless you have a specific reason not to, and don’t quote “one in four” as a measured success rate for DTC brands; nobody has measured that. And check the thing Zook put first: that the core is strong. A new category launched to cover a core that’s losing share of wallet is the pattern his books warn about most.
Zook counted steps from the core but never published how. Here’s a version you can count yourself: four things a new category can share with the one you have.
“Adjacent” is the most flexible word in a planning meeting: everything is adjacent to something. To make it useful, ask what the new category shares with the core, one dimension at a time. Each no is a step.
A score of 0 or 1 on any dimension counts as a step. No steps means you’re deepening the core. One step is the next ring. Two is a stretch that usually needs a partner or a very good reason. Three or four is a new business, and it should have its own business case, not a slot in the product calendar.
Proof matters most. The brand extension research in chapter 5 found that fit is the first thing customers judge an extension on, and a 2023 meta-analysis of that research found that usage fit, which turns on how the products are used, was the weakest kind Published. So a shared occasion without shared proof is probably weaker than it looks. Filters for a coffee brand pass on both. A coffee brand’s branded mugs pass on occasion and fail on proof: nobody thinks a roaster makes better mugs.
Count the nos. Each one is a thing you’ll have to learn with real money.
Distance tells you the risk; the launch still has to pay. The quickest check is how many existing customers must buy in year one to earn back the launch cost: development, samples, inventory you may not sell, photography and the team’s time.
Say a coffee brand with 20,000 active customers is weighing cold-brew concentrate. It expects buyers to spend $60 a year on it at a 35% contribution margin, so each buyer is worth $21 a year. Launch costs come to $60,000. It needs about 2,860 buyers, or 14% of its active customers, to break even in year one on the base alone Derived. If its own plan assumes 10%, the launch loses money in year one unless it brings in new customers, and new customers cost money to acquire.
With the defaults, the cold brew scores 8 of 12 and sits one step out, on supply chain. It needs 14% of customers to buy in year one to break even, and the plan’s 10% leaves an $18,000 loss on the base. That’s not a reason to kill it. It’s a reason to run the waitlist in chapter 11 before ordering stock, and let 2,000 sign-ups or 200 tell you which plan is real.
Score candidates as a group, with someone who talks to customers every week. The scores are judgments; writing them down before launch stops them changing after it.
Thirty-five years of brand extension research agree on one thing: customers judge a new product by whether it fits what they already believe about you. The details are messier than the textbooks say.
Liking a brand isn’t the same as believing it can make something else. Customers who love your sunscreen may not trust your supplements. The research on why started with one paper.
In 1990 David Aaker and Kevin Lane Keller asked consumers to rate proposed extensions of well-known brands. Their model said an extension would be judged on the perceived quality of the parent brand, on three kinds of fit (is the new product a complement to the old one, a substitute for it, and would the brand’s skills transfer to making it), and on how hard the new product seemed to make. What they found was that fit and difficulty mattered, and quality mattered mainly when there was fit. The parent brand’s quality on its own didn’t lift the extension Published. A great brand stretching into something unrelated didn’t carry its reputation with it. And an extension that looked too easy to make was rated lower, as if the brand were cashing in its name.
Then came the replications. Sunde and Brodie repeated the study in 1993 and got different results, and several more replications around the world disagreed with each other Published. In 2001 Paul Bottomley and Stephen Holden pooled the original data with seven replications and found support for the full model, but concluded that “the level of contribution of each of these components varies by brand and culture” Published. In 2006 Raj Echambadi and colleagues reanalyzed the same pooled data and argued that Bottomley and Holden’s simple effects were estimated incorrectly: what predicted extension ratings was quality and fit together, not either alone Published.
The field argued about the details for fifteen years. Nobody overturned the finding that fit matters.
Franziska Völckner and Henrik Sattler tested ten success factors from the earlier research at once, in 2006, and ranked them. Fit between the parent brand and the extension came first, then marketing support, then parent-brand conviction (how strongly customers believe in the parent), retailer acceptance and parent-brand experience Published. For a DTC brand, “retailer acceptance” is your own site and email: whether the extension gets real placement or a tile at the bottom of the collection page.
In 2023 Chenming Peng, Tammo Bijmolt, Völckner and Hong Zhao published the meta-analysis: 2,134 effect sizes from research between 1990 and 2020. Both parent brand equity and fit raised extension success. Among the kinds of fit, usage fit had the weakest effect Published. That’s one reason chapter 4 weighs proof, whether the brand’s skill obviously carries over, above a shared occasion.
Customers don’t ask whether they like you. They ask whether you’d be good at this.
A poor extension doesn’t only fail on its own. Barbara Loken and Deborah Roedder John found in experiments that extensions whose attributes contradicted what people believed about the family brand could dilute those beliefs, though less when people saw the extension as untypical of the brand Published. Vanitha Swaminathan and colleagues, using household scanner data on six real extensions, found signs that unsuccessful extensions could hurt choice of the parent brand, while successful ones helped it, most of all among people who hadn’t bought the parent before Published.
For a DTC brand the risk is specific: a brand known for testing launches something untested, or a durability brand launches something that breaks, and its reviews sit on the same site as the hero’s.
The research method is also the cheapest fit test there is. Describe the candidate in one line and ask customers, on a 1 to 7 scale: how well does this fit what you know of us, how good would we be at making it, how likely are you to buy it. Get at least 200 answers; the survey is in Appendix B. Stated intent overstates buying, so compare candidates with each other and let the waitlist in chapter 11 do the real test.
A cooler company that became mostly a drinkware company, and a shoe company that stopped making its first apparel line. What their filings say, and what they don’t.
Both of these companies sell mostly direct to consumers, both are public, and both expanded beyond their first product. One expansion became the larger part of the business. The other’s first line was discontinued. Everything here comes from their SEC filings.
YETI was founded in 2006 by Roy and Ryan Seiders, who wanted a “nearly indestructible hard cooler with superior ice retention” for hunting and fishing Filed. In 2014 it launched the Hopper soft cooler and, in its words, “entered a new category with the Rambler drinkware line” Filed. Net sales went from $89.9 million in 2013 to $468.9 million in 2015 Filed. By 2019 drinkware was 58% of net sales, and in 2024 it was 60%, with coolers and equipment at 38% Filed. The direct channel was 59% of net sales in 2024 Filed.
FiledYETI Holdings, Form S-1, July 2016; Form 10-K for 2019, February 2020; Form 10-K for 2024, February 2025.
On the four shared things, drinkware scores high on customer (the same outdoor buyer), occasion (the same trips and job sites) and above all proof: a brand trusted for ice retention is believed when it says it keeps a drink cold. It’s weaker on supply chain, a different product to make at a much lower price. That’s one step out. The company describes its method the same way: “anchor products, followed by product expansions,” such as sizes and colorways, then accessories Filed. Core first, then the next ring, then the ring after.
Two cautions. Scoring a success after the fact is the hindsight chapter 3 warned about. And the filings don’t split drinkware’s growth between cooler owners and new customers; the 2016 filing credits the 2014 launches with expanding YETI’s “reach beyond the premium hard cooler category” Filed. A cheaper product that carries the brand’s proof is also an entry point.
Allbirds built its brand on wool sneakers. Its 2023 annual report says footwear “represents the vast majority of our revenue and is the foundation of our brand” Filed. The same report explains what happened to its first apparel line: “our customers have not purchased certain of these products in sufficient quantities,” demand “failed to meet our expectations,” and in the second quarter of 2022 the company “determined that we needed to adjust our overall apparel strategy and discontinue the product line” Filed. It still sells tees, sweats, socks and underwear, described as secondary offerings.
Net revenue fell 14.7%, from $297.8 million in 2022 to $254.1 million in 2023, and the net loss widened from $101.4 million to $152.5 million Filed. The 2024 annual report sets the goal of growing “within our existing customer base” and increasing “closet share by focusing on our core franchise products” Filed. Closet share is share of wallet, in apparel. After the expansion, the plan was headroom in the core.
FiledAllbirds, Form 10-K for 2023, March 2024; Form 8-K with fourth-quarter 2023 results, March 12, 2024; Form 10-K for 2024, March 2025.
Be careful with the lesson. Apparel wasn’t Allbirds’ only problem: the March 2023 transformation plan also slowed store openings and reconsidered its international approach Filed. And the filing says customers didn’t buy enough, not why. The honest reading is narrower: a brand whose proof was comfort and materials in shoes found that the proof didn’t carry customers into first-generation apparel in the numbers it planned for, and the plan it wrote afterward pointed back at the core.
One expansion borrowed the core’s proof. The other borrowed its name.
Both companies ended up talking about the core: YETI’s method starts with anchor products, and Allbirds’ recovery plan starts with core franchises and closet share. Filings show outcomes, not mechanisms, so use cases like these to ask better questions about your own candidates, not to predict them.
Customers who buy from two categories spend more than customers who buy from one. Most of that gap was there before they bought the second category.
The slide is built the same way everywhere. Split customers into those who have bought from more than one category and those who haven’t, compare a year’s spending, and divide. Then comes the leap: get single-category customers to buy a second category and they’ll be worth that multiple.
To buy from two categories, a customer has to buy at least twice, or buy a bigger basket. Customers who buy often, and have been around longer, have more chances to try something else. So “multi-category” is partly another name for “already buys a lot,” and the slide reports that the group who buy more buy more.
The research says the same. V. Kumar, Morris George and Joseph Pancras studied what predicts cross-buying at a catalog retailer and found that, alongside the firm’s marketing, it tracked how customers already behaved: how often they bought, how much they returned, how focused their buying was, and which category they started in Published. Werner Reinartz, Jacquelyn Thomas and Ganaël Bascoul tested the direction directly, on two data sets, and concluded that “cross-buying is a consequence and not an antecedent of behavioral loyalty”: loyal behavior drives the number of categories people buy from, not the other way around Published.
Annual reports make the claim the other way. Warby Parker’s 2024 10-K says customers who shop “across product lines and channels tend to convert to highly loyal returning customers” Filed. That may be true, but as written it’s a correlation, and it would read the same if loyal customers simply went on to buy more lines.
Multi-category customers aren’t valuable because they buy more categories. They buy more categories because they were already valuable.
Recommendations have the same problem. Amit Sharma, Jake Hofman and Duncan Watts used natural experiments on Amazon and estimated that at least 75% of the clicks through recommendations would likely have happened without them, through search or another route Published. Attribution credited the recommendation for traffic that was coming anyway.
You can’t remove selection from observational data, but you can size it. Compare what the two groups spent before any of them bought a second category, then compare growth, not level.
Say a brand’s multi-category customers spent $450 in their second year and its single-category customers spent $150: the slide says 3x. Now look at their first six months, before anyone crossed categories. The future multi-category buyers spent $200; the others spent $100. The gap was already 2x. From that start, the multi-category group grew 2.25 times and the single-category group 1.5 times, so the most cross-buying could explain is a 1.5x difference. A converted single-category customer would go from about $150 to at most $225 a year: worth up to $75 more, not $300 Derived.
DerivedFrom the example’s made-up round numbers. Your own split comes from the query in Appendix A.
With the defaults, the 3.0x becomes at most 1.5x, and a converted customer is worth up to $75 a year more, a quarter of the $300 the slide implies. That’s still worth having. It’s a different budget.
Across five companies, between a tenth and a third of cross-buyers lost money, and they accounted for up to 88% of the losses. Find yours before your cross-sell flow finds them.
Chapter 7 said the value of cross-buying is smaller than the slide claims. This chapter says some of it is negative. Denish Shah, V. Kumar, Yingge Qu and Sylia Chen went looking for the customers conventional wisdom ignores: the ones who buy across categories and lose money doing it.
They analyzed the customer databases of five firms, in consumer and business markets. Between 10% and 35% of each firm’s cross-buying customers were unprofitable, and those customers accounted for 39% to 88% of the firm’s total loss from customers. The unprofitable cross-buyers shared “persistent adverse behavioral traits”: limited spending, “excessive revenue reversals” (returns and cancellations), excessive service requests and buying on promotion. For these customers, more cross-buying meant bigger losses, a “downward spiral” Published.
PublishedShah, Kumar, Qu and Chen, Journal of Marketing, 2012. Shah and Kumar summarized it for managers as “The Dark Side of Cross-Selling,” Harvard Business Review, December 2012.
Every DTC brand has these customers: the buyer who orders three sizes and keeps one, the one who only buys in the sitewide sale, the one who opens a ticket on every order. They show up as multi-category, which is what the cross-sell flow was built to create.
A cross-sell flow that doesn’t know contribution works hardest on the customers who cost the most.
Revenue can’t see these customers. Contribution can: what’s left after discounts, product cost, returns, fulfillment and service. The calculator below takes one customer, or the average of a segment, and adds it up.
With the defaults, the customer keeps $168 a year in revenue and loses $15.60. Each extra order adds only $1.10, and at an average discount of 18% or less, instead of 25%, they’d break even Derived. On a revenue report this customer looks fine: four orders a year, several categories. On contribution they’re a cost, and a discounted cross-sell to a new category makes them a bigger one.
Two of the biggest cross-selling bets in modern business, both at banks. One was quietly unwound. The other became a fraud case.
If cross-buying followed from loyalty rather than causing it, you’d expect companies that made cross-selling the goal to be disappointed. The two most famous cases were worse than disappointing.
In 1998 Sandy Weill’s Travelers Group merged with Citicorp on a simple thesis: one company selling banking, insurance and investments to the same customers would sell more of each. Amey Stone and Mike Brewster’s King of Capital tells that story. The thesis didn’t hold. In January 2005 Citigroup agreed to sell Travelers Life & Annuity and substantially all of its international insurance businesses to MetLife for $11.5 billion. Its chief executive, Charles Prince, said the sale “sharpens our focus on Citigroup’s long-term growth franchises” Filed. The company built to cross-sell insurance from inside one firm sold its life insurance business to a specialist and agreed to distribute the specialist’s products instead.
Wells Fargo went further and made cross-selling the number. From at least 2000 until the third quarter of 2016 it published a cross-sell metric, “the ratio of the number of accounts and products per retail bank household,” and presented it to investors as central to its community bank Filed. In September 2016 the Consumer Financial Protection Bureau fined the bank $100 million, alongside $35 million from its bank regulator and $50 million to the City and County of Los Angeles, after employees opened roughly 1.5 million deposit accounts and applied for roughly 565,000 credit card accounts that may not have been authorized, driven by what the Bureau called sales targets and compensation incentives Filed. In February 2020 the bank agreed to pay $3 billion to settle criminal and civil investigations by the Justice Department and the SEC. The SEC said the metric had been “inflated by accounts and services that were unused, unneeded, or unauthorized” Filed.
FiledCitigroup Form 8-K, January 31, 2005; CFPB, September 8, 2016; SEC press release 2020-38 and order, February 21, 2020.
When the number of products per customer is the target, people will find a way to add products.
No DTC brand is going to open accounts in a customer’s name. But the mechanism, a count that stands in for value and then becomes a target, shows up in small ways. An agency paid on “multi-category customers” gets there with free samples from a second category, or a deeper discount on it. Each raises the count and lowers contribution per customer, which was the point of the count.
Goodhart’s law says a measure that becomes a target stops being a good measure. Report categories per customer; it tells you whether the base is broadening. Pay people on contribution per customer, measured against a holdout.
Every cross-category email, text and recommendation gets credit for purchases. A holdout tells you how many it caused.
Your cross-sell flow reports every purchase by someone who got the message, within an attribution window. Some bought because of it. If the Amazon estimate for recommendation clicks in chapter 7 carries over, many would have bought anyway. The only way to know the split is to hold some customers back.
Take everyone who qualifies for the flow, say everyone who has placed a second core order, and randomly assign a fixed share, often 10%, to get nothing from it. They still get newsletters and core flows. After 60 to 90 days, compare the groups on three numbers, counted from the day each customer qualified, opened or not:
Say 40,000 customers qualify and 4,000 are held out. In 90 days, 6.0% of the customers who got the flow buy the new category, against 4.5% of the holdout. Attribution would credit the flow with about 2,160 buyers, 6.0% of 36,000. The holdout says it caused about 540, 1.5 points on 36,000: a quarter of what the dashboard shows Derived. At these sizes the difference is about four times its standard error, well clear of noise Derived. With a smaller list, work out first whether the test can see the lift you care about; the sample-size math is in The Honest Test.
Attributed revenue is what the flow was near. Incremental revenue is what it did.
Bundles and cart add-ons are the same question at checkout: they need a holdout too, and they’re covered in The First Offer.
A small permanent holdout, 5% of new customers who never get cross-category messages, lets you check every quarter whether the whole cross-sell program still adds contribution. It costs a little revenue, and it’s the one cross-sell number nobody can argue with.
Your existing customers will try the new category more readily than anyone. That makes them the cheapest test, as long as you judge the result on the second order.
The usual launch builds the product, shoots the ads, spends on acquisition and emails the list on launch day with everyone else. It pays to acquire strangers into a product your own customers could have judged for free.
Vanitha Swaminathan, Richard Fox and Srinivas Reddy followed six real brand extensions through a national household scanner panel. Experience with the parent brand had a significant effect on whether households tried the extension, and not on whether they bought it again Published. Your customers will try the new thing because they like you. They’ll buy it a second time only if the product earns it.
That’s the case for launching to the base first, and the warning attached to it. The base gives you trial cheaply and quickly. It also inflates first-order numbers with goodwill. So read a base launch on repeat: the share of first buyers who buy the new category again within about one and a half times its natural reorder interval. Timing the reorder itself is covered in The Second Order.
Trial measures how much they like you. Repeat measures how much they like the product.
Warby Parker sells glasses first. It launched Scout daily contact lenses in 2019. Five years later, in 2024, contacts were 10.2% of its $771.3 million in net revenue, and eye exams and vision care another 5.3% Filed. That’s one step out on the four shared things (the same customers, the same need to see well, and the same optical proof) and it still took years to reach a tenth of revenue. Plan the next category’s first year as a test, not a second engine.
Say the coffee brand from chapter 4 emails its 20,000 active customers about cold-brew concentrate, and 1,600 join the waitlist, 8% of the base. If 40% of the waitlist buys in the first month, that’s 640 buyers, 3.2% of the base, against the 14% it needs in year one to earn back the launch cost Derived. More customers will buy over the year, so the gap isn’t fatal, but it says: small first run, then wait for the repeat rate.
Some products make yours more valuable. You don’t have to make them to benefit from them.
Adam Brandenburger and Barry Nalebuff’s Co-opetition (1996) added a player that most strategy maps leave out. Next to customers, suppliers and competitors sit complementors: companies whose products make customers value yours more when they have both Published. The textbook example is computer hardware and software. For a DTC brand, it’s the grinder for the coffee, the vet service for the pet food, the carrier for the baby clothes.
A candidate that scores two steps out in chapter 4 usually fails on supply chain, proof or both. A complementor already has both. Partnering gives your customers the product without your needing its factory, its inventory or its credibility. It also gives you evidence. If your customers buy a partner’s product in numbers, you’ve tested the category without owning it, and you can decide later whether to build.
A partner sells you the option to launch later, for the price of a conversation now.
Score the partner’s category on the same four shared things, from your customer’s side. Then ask two more questions: do their customers look like yours, so the benefit runs both ways, and would their worst failure embarrass you? Their product sits next to your name, so the dilution research in chapter 5 applies to them too.
Test a partnership like a nudge: hold out a slice of your list and compare contribution per customer after 60 days. A partnership that only moves the partner’s revenue is a favor, not a strategy.
Twelve numbers, reviewed each quarter: three for the core, three for the candidate, six for categories you’ve already launched.
Revenue can’t tell you whether a new category took share from the core, whether its buyers were going to buy more anyway, or whether they return half of it. The scorecard puts those next to revenue.
| Number | How to get it | Healthy | Warning |
|---|---|---|---|
| The core | |||
| 1. Hero share of wallet | Survey plus orders, chapter 2 | Rising, or above the rule’s share for your rank | Falling while a new category launches |
| 2. Share of customers ranking you first | Same survey | Rising | Falling |
| 3. Core revenue per active customer | Orders | Flat or rising | Falling two quarters in a row |
| The candidate | |||
| 4. Steps from the core | Scorer, chapter 4 | 0 or 1 | 2 or more, with no partner considered |
| 5. Fit survey: “how good would we be at making it” | Chapter 5 | Top of your candidates | Bottom, whatever the purchase intent |
| 6. Waitlist sign-ups against break-even share | Chapters 4 and 11 | Close to or above break-even | Well below it, with a full stock order placed anyway |
| Launched categories | |||
| 7. Repeat rate of first buyers | At 1.5 times the reorder interval | At or above the bar in the brief | Below it with no kill decision |
| 8. Adjusted multi-category multiple | Calculator, chapter 7 | Reported next to the raw one | Raw multiple used in plans |
| 9. Cross-buyers with negative contribution | Calculator, chapter 8 | Known, and suppressed from cross-sell | Unknown |
| 10. Incremental contribution per customer | Holdout, chapter 10 | Positive and reported quarterly | No holdout |
| 11. Core revenue among new-category buyers, against the holdout | Same holdout | No lower than the holdout’s | Lower: the new category is replacing the core, not adding to it |
| 12. Return rate, new category against core | Refunds by category | Similar | Much higher, especially with a new sizing system |
Numbers 1, 2, 5 and 6 need a survey or a waitlist; the rest come from order data.
Write it before launch, in one sentence, with a date and an owner: “If fewer than [x]% of first buyers reorder by [date], or contribution per buyer after returns is below $[y], we stop reordering stock and sell through.” Categories rarely fail loudly. They linger, tying up cash and site space, because nobody decided to end them. The kill rule is that decision, made while everyone is still objective.
Every category you launch needs a date on which someone is allowed to end it.
Four weeks from “we should launch something” to a decision you can defend, with evidence for it.
None of this needs new software: a survey, three queries, a holdout flag in your email platform and one scoring meeting. Here’s the order.
At the end of the month you’ll have a share-of-wallet number, an honest multi-category multiple, a list of cross-buyers who cost money, a running holdout and a scored shortlist.
Six things the person deciding the next category needs on the first day.
Whoever owns the decision, a new head of growth, an agency or you, needs six things on day one. Without them, the first month goes on hunting for them.
The books and papers this guide leans on, and what to take from each.
And the research: Peng and colleagues’ 2023 meta-analysis on brand extensions, Reinartz, Thomas and Bascoul (2008) and Shah, Kumar, Qu and Chen (2012) on cross-buying, and Sharma, Hofman and Watts (2015) on what recommendations cause. Full references are in Appendix C.
Andrew Lauchner runs Growth Legend, embedding inside consumer brands to own lifecycle, email and SMS, and revenue operations. He is the author of The Second Order, on turning first-time buyers into second-time buyers, and The Whole Machine, on the fundamentals of DTC growth, along with a series of field guides for DTC operators at andrewlauchner.com.
As Senior Director of Growth and Retention Marketing at Gallery Furniture, he rebuilt the customer journey and the sales playbooks together. He has worked on growth and retention at Binance and 3Commas, and has been Head of Growth and Retention at Greatness Wins and at Nexus Agriscience.
“Andrew led retention, lifecycle, and email/SMS, but what separates him from most in this space is how deeply he understands the role retention plays in the overall growth engine.”
Akram Khan, Head of Marketing at Gallery Furniture, senior to Andrew but didn’t manage Andrew directly
Andrew answers every note from operators working on this, including those looking for someone to own it. Write to andrew@growthlegend.com or message him on LinkedIn.
The formulas behind the four calculators, and four queries every brand with more than one category should be able to run.
| For | Formula | Notes |
|---|---|---|
| Wallet Allocation Rule | share = (1 − r/(n + 1)) × 2/n | r: your rank; n: brands the customer uses. Keiningham and colleagues. |
| Break-even buyers | launch cost / (spend per buyer × margin) | Divide by active customers for the share. Year one, base only. |
| Adjusted multiple | (A/B) / (C/D) | A, B: outcome-year spend, multi and single. C, D: early-window spend. An upper bound on the causal effect. |
| Contribution per order | v(1−d)(1−r) − vc(1−r) − r×rc − f | v: full-price order value; d: discount; r: share returned; c: product cost share; rc: cost per return; f: fulfillment per order. Per year: orders × this − service. |
| Break-even discount | 1 − (s/o + vc(1−r) + r×rc + f) / (v(1−r)) | s: service cost per year; o: orders per year. |
| Holdout lift | (pT − pH) ± 1.96√(pT(1−pT)/nT + pH(1−pH)/nH) | p: share who bought the new category; T: treated, H: holdout. Incremental buyers: lift × nT. |
-- survey_responses: customer_id, category_spend_12m, brands_used, our_rank
WITH ours AS (
SELECT o.customer_id, SUM(ol.net_amount) AS our_spend_12m
FROM orders o
JOIN order_lines ol ON ol.order_id = o.id
JOIN products p ON p.id = ol.product_id
WHERE p.category = 'coffee' -- the hero's category
AND o.created_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY o.customer_id
)
SELECT s.our_rank, s.brands_used, COUNT(*) AS customers,
AVG(LEAST(1, COALESCE(ours.our_spend_12m, 0)
/ NULLIF(s.category_spend_12m, 0))) AS actual_share,
AVG((1 - s.our_rank::numeric / (s.brands_used + 1))
* (2.0 / s.brands_used)) AS rule_share,
SUM(s.category_spend_12m - COALESCE(ours.our_spend_12m, 0)) AS spend_elsewhere
FROM survey_responses s
LEFT JOIN ours USING (customer_id)
WHERE s.brands_used BETWEEN 1 AND 20
AND s.our_rank BETWEEN 1 AND s.brands_used
GROUP BY s.our_rank, s.brands_used
ORDER BY s.brands_used, s.our_rank;
If actual and rule shares roughly agree across rows, rank is a usable tracker. Stated spend is noisy: use medians in the calculator.
-- groups customers by when they first bought outside their first category
WITH lines AS (
SELECT o.customer_id, o.created_at, p.category,
ol.net_amount - COALESCE(rf.amount, 0) AS net
FROM orders o
JOIN order_lines ol ON ol.order_id = o.id
JOIN products p ON p.id = ol.product_id
LEFT JOIN (SELECT order_line_id, SUM(amount) AS amount
FROM refunds GROUP BY order_line_id) rf
ON rf.order_line_id = ol.id
),
firsts AS (
SELECT customer_id, MIN(created_at) AS first_at FROM orders GROUP BY customer_id
),
first_cat AS (
SELECT DISTINCT ON (l.customer_id) l.customer_id, l.category
FROM lines l JOIN firsts f USING (customer_id)
WHERE l.created_at = f.first_at
ORDER BY l.customer_id, l.net DESC
),
crossed AS (
SELECT l.customer_id, MIN(l.created_at) AS crossed_at
FROM lines l JOIN first_cat c USING (customer_id)
WHERE l.category <> c.category
GROUP BY l.customer_id
),
per AS (
SELECT f.customer_id, f.first_at, x.crossed_at,
SUM(l.net) FILTER (WHERE l.created_at < f.first_at + INTERVAL '180 days') AS early,
SUM(l.net) FILTER (WHERE l.created_at >= f.first_at + INTERVAL '365 days'
AND l.created_at < f.first_at + INTERVAL '730 days') AS outcome
FROM firsts f
JOIN lines l USING (customer_id)
LEFT JOIN crossed x USING (customer_id)
WHERE f.first_at < CURRENT_DATE - INTERVAL '730 days'
GROUP BY f.customer_id, f.first_at, x.crossed_at
),
grouped AS (
SELECT *, CASE
WHEN crossed_at IS NULL OR crossed_at >= first_at + INTERVAL '730 days' THEN 'single'
WHEN crossed_at >= first_at + INTERVAL '180 days'
AND crossed_at < first_at + INTERVAL '365 days' THEN 'multi'
END AS grp
FROM per
)
SELECT grp, COUNT(*) AS customers,
AVG(COALESCE(early, 0)) AS early_spend, -- days 0 to 179
AVG(COALESCE(outcome, 0)) AS outcome_spend -- days 365 to 729
FROM grouped
WHERE grp IS NOT NULL
GROUP BY grp;
Customers who crossed in their first 180 days, or in the outcome year, are left out, so both groups’ early windows are clean. The four averages go into the chapter 7 calculator. Syntax is Postgres.
-- last 12 months. order_contribution: one row per order, with customer_id,
-- created_at, categories (array), and contribution = revenue after discounts
-- and refunds, minus product, fulfillment, shipping, payment and return costs.
SELECT c.customer_id,
(SELECT COUNT(DISTINCT cat)
FROM order_contribution c2, unnest(c2.categories) AS cat
WHERE c2.customer_id = c.customer_id
AND c2.created_at >= CURRENT_DATE - INTERVAL '12 months') AS categories,
SUM(c.contribution)
- COALESCE(MAX(t.tickets), 0) * :cost_per_ticket AS contribution
FROM order_contribution c
LEFT JOIN (SELECT customer_id, COUNT(*) AS tickets FROM support_tickets
WHERE created_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY customer_id) t USING (customer_id)
WHERE c.created_at >= CURRENT_DATE - INTERVAL '12 months'
GROUP BY c.customer_id;
Wrap it to count customers with negative contribution among those with two or more categories, and their share of all negative contribution. That’s your version of Shah and colleagues’ 10% to 35% and 39% to 88%.
-- flow_assignments: customer_id, arm ('treat' or 'holdout'), qualified_at
WITH w AS (
SELECT a.customer_id, a.arm,
BOOL_OR('cold_brew' = ANY(c.categories)) AS bought_new,
COALESCE(SUM(c.contribution), 0) AS contribution
FROM flow_assignments a
LEFT JOIN order_contribution c
ON c.customer_id = a.customer_id
AND c.created_at >= a.qualified_at
AND c.created_at < a.qualified_at + INTERVAL '90 days'
WHERE a.qualified_at < CURRENT_DATE - INTERVAL '90 days'
GROUP BY a.customer_id, a.arm
)
SELECT arm, COUNT(*) AS customers,
AVG(COALESCE(bought_new, false)::int) AS bought_new_category,
AVG(contribution) AS contribution_per_customer
FROM w GROUP BY arm;
Count every assigned customer, including those who never opened a message. Comparing openers with the holdout brings the selection problem back.
Six forms and messages. Copy them into whatever your team already uses, and have counsel review anything that goes to customers.
SEND TO Active customers (bought in the last 12 months). Aim for 300+ answers.
Q1 In the last 12 months, about how much did you spend on [category],
from any brand or store? ($ amount, or ranges)
Q2 Which brands of [category] did you buy in that time? (list, with "other")
Q3 Rank the brands you picked, from the one you'd choose first.
OPTIONAL
Q4 For the brand you ranked first: what does it do better than us?
JOIN Answers to customer_id, then to 12 months of orders.
SHOW One line per candidate: "[Brand] is thinking about making [product]." ASK, 1 TO 7: Q1 How well does this fit what you know about [Brand]? Q2 How good do you think [Brand] would be at making it? Q3 How likely would you be to buy it from [Brand]? Q4 Do you buy [product] today? From whom? (free text) READ Rank candidates against each other. Q4 feeds "shared customer."
SUBJECT Something New From Us, and You're First
PREVIEW before anyone else sees it
BODY
Hi [first name],
You've trusted us with your [hero product]. We've been working on
[new product], because [the one-sentence proof it borrows from the hero].
We're making a small first batch, and customers on the waitlist get
it first, before we announce it anywhere else.
[Join the waitlist]
No payment now. We'll email you when it's ready.
[Signature]
FOOTER [Company legal name, postal address]
You're receiving this because you bought from [Brand].
[Unsubscribe] | [Email preferences]
SEND ONLY TO subscribers who agreed to receive marketing texts. [Brand]: we're making [new product] in a small first batch. Customers get first access before anyone else. Join the waitlist: [link] Reply STOP to opt out.
CANDIDATE product, price, contribution margin:
STEPS FROM CORE customer / occasion / supply chain / proof, 0 to 3 each:
PROOF IT BORROWS one sentence:
HEADROOM COMPARED core gain from one rank up ($/yr) vs this plan ($/yr):
BREAK-EVEN share of active customers who must buy in year one:
WAITLIST sign-up rate by segment:
FIRST RUN units, sized to the waitlist, not the plan:
LAUNCH PRICE full price. Early access instead of a discount.
REPEAT BAR [x]% of first buyers reorder by [date]
KILL RULE below the repeat bar, or contribution per buyer
after returns under $[y], by [date]: stop reordering
and sell through.
HOLDOUT [x]% of eligible customers get no launch messages
OWNER name, and the date the kill rule is read:
PARTNER brand, product, why our customers use it with ours:
SCORE customer / occasion / proof, from our customer's side:
THEIR CUSTOMERS how much they overlap with ours (their estimate):
FORM recommend / co-market / co-branded edition / resell
EACH SIDE SENDS to its own list only. No list swaps or data sharing
without counsel's review.
DISCLOSURE any commission or payment disclosed clearly wherever
we recommend them.
TEST [x]% holdout from our side; read contribution per
customer at 60 days.
EXIT notice period, and how we'd respond to their
worst-case failure:
Every external source, by chapter. Web sources were read in September 2026.