Across 939 B2B software companies, the median SMB-focused product retains 97% of its revenue year over year. The median enterprise product retains 118%. That 21-point gap is not a rounding error. It is the difference between a business that quietly shrinks inside its own customer base and one that grows without signing a single new logo. It is also the reason nearly every successful bottom-up product eventually faces the same question: should we move upmarket?
The pull is obvious. The cost is not. Moving upmarket looks like a pricing decision, a matter of adding an Enterprise tier and a sales team. It is actually a decision to rebuild the product, the roadmap, and the company around a different customer. And once you walk through that door, it is very hard to walk back.
The math that makes upmarket almost irresistible
Start with why the gravity pulls one direction. Enterprise customers churn at roughly 0.3% to 0.8% per month, which works out to somewhere between 4% and 9% a year. SMB customers churn at 3% to 5% per month, which compounds to 22% to 39% annually. One analysis put enterprise retention at about 5.8 times better than SMB. Worse, SMB losses front-load: something like 43% of them happen in the first 90 days, before you have earned a dollar of margin on the acquisition.
Stack expansion revenue on top of that. Larger accounts add seats, add modules, and renew on multi-year contracts negotiated by procurement teams who make switching genuinely painful. That is why the net revenue retention curve bends upward as account size grows. The SMB product at 97% is treading water. The enterprise product at 118% is compounding.
Christoph Janz framed the strategic version of this a decade ago in his well-known “five ways to build a $100 million business.” You can get there by hunting elephants, deer, rabbits, mice, or flies, where each animal is a band of average revenue per account. Every path can work. But the elephant hunters need a few thousand customers, and the fly catchers need tens of millions. When you are staring at that chart, the elephants look a lot cheaper to feed.
So the CFO runs the model, sees the retention curve, and asks the product team why you are not already up there with the elephants. It is a fair question. It just has a more expensive answer than the model shows.
What actually changes when you go up
Here is where I have watched teams get hurt. A few years back I did a fractional COO engagement with a B2B software company that had built a genuinely loved self-serve product. Small teams signed up with a credit card, got value in an afternoon, and told their friends. The whole machine ran on simplicity. Then a handful of larger prospects showed up waving contracts, and the company decided to serve them.
Nobody made a bad individual decision. Each larger deal asked for something reasonable: granular permissions, single sign-on, audit logs, a security questionnaire answered, an admin console, custom roles. The team said yes, because yes was worth six figures. Eighteen months later the product had a settings page with forty options, an onboarding flow that now assumed an IT administrator, and a self-serve conversion rate that had quietly fallen off a cliff. The customers who made the product famous found it heavier every quarter. The company had not decided to abandon them. It had just optimized, one enterprise request at a time, for a different buyer.
That is the part the spreadsheet hides. Moving upmarket is not adding a tier on top of what you have. It changes what the product is:
- The buyer changes. A user who adopts in an afternoon is replaced by a committee that evaluates for two quarters. You are no longer designing for the person who gets value; you are designing for the person who signs, plus the security reviewer, plus the admin who deploys it.
- The roadmap changes. Enterprise deals arrive attached to feature commitments. A meaningful slice of your engineering capacity shifts from “what makes the product better for everyone” to “what unblocks this one large contract.” Your roadmap starts getting written in sales calls.
- The definition of done changes. Compliance, provisioning, uptime guarantees, and support SLAs are not features users see. They are table stakes that consume real capacity and return nothing to your original base.
- The cost to serve changes. Self-serve scales because humans stay out of the loop. Enterprise scales because humans stay in it: implementation, customer success, dedicated support. You are trading a high-margin motion for a high-touch one, and hoping the larger contracts more than cover the difference.
None of this is an argument against going upmarket. HubSpot is the reference case for doing it deliberately. It has pushed toward larger accounts since roughly 2019, building Enterprise tiers with custom objects, advanced permissions, and business units aimed at companies with 200 to 2,000-plus employees. The result shows up in the numbers: average subscription revenue per customer climbed from about $9,669 in late 2020 to roughly $11,414 in 2025, an increase the company attributes primarily to an upmarket shift in customer mix and demand for its Professional and Enterprise products rather than raising prices on existing customers. That is the move executed well: the mix shifts up over years, and the company builds the go-to-market muscle to support it deliberately, while staying a credible option for smaller buyers.
The failure mode is not the destination. It is arriving there by accident, one reasonable yes at a time, without ever deciding to.
The door swings one direction
There is a structural fact about this move that gets left out of the pitch: it is far easier to climb than to descend.
David Sacks, who built Yammer on exactly this bottom-up-then-upmarket path, describes the asymmetry plainly. If you have customers paying $10,000 a year, you can find customers who will pay $100,000, and your organization will figure out how to sell to and support them. The reverse almost never works. Once you have built the expensive, high-touch sales and support motion that enterprise demands, you cannot profitably run it at a lower price point. The infrastructure that wins a $100,000 deal makes a $1,000 deal lose money on contact.
So a company that has gone upmarket and then wants to reopen the SMB market usually cannot. The cost structure will not allow it. The org has been reshaped around long sales cycles and human-heavy delivery. This is why the decision deserves more scrutiny than its individual steps ever get. Adding SSO for one big customer feels reversible. The cumulative drift it is part of is not.
That asymmetry cuts a useful way, though. If you are going to be trapped at one end of the market, the top is a better place to be stuck than the bottom. It just means the choice should be made with your eyes open, because you are unlikely to get to unmake it.
The question strategy has to answer before the tier gets built
The mistake is treating “should we go upmarket” as a pricing question, when it is a positioning question with pricing attached. Pricing follows from who you have decided to serve; it is one of the clearest expressions of product strategy, not a lever you pull independently of it. So before anyone builds the Enterprise tier, the strategy has to answer a harder question: who is this product for, and what are we willing to make worse for everyone else to serve them?
A few things worth deciding out loud, in that order:
Decide the direction on purpose, not on inbound. Larger deals will find you before you go looking. If every big prospect that emails in bends the roadmap, you have not chosen to go upmarket; the market has chosen for you. That is the accidental version, and it is the one that erodes the base. Choosing means naming the segment you are moving toward and accepting the tradeoffs, not just cashing the biggest checks that arrive.
Protect the base or knowingly release it. You can serve both ends, but only if you defend the self-serve experience with the same rigor you apply to enterprise requests. That means a real owner for the simple path and a bar that enterprise features have to clear before they are allowed to complicate the default experience. If you are not willing to staff that defense, be honest that you are letting the base go, because you are, whether or not you admit it.
Watch the leading indicator, not the lagging one. Revenue per account will look great while it is happening, because the mix is shifting up. The signal that you are hollowing out the base shows up earlier and quieter: self-serve activation, time to first value, and conversion from free to paid. When those slide while ARPA climbs, you are not adding a segment. You are trading one for another, and the trade may not be the one you intended.
Understand what upmarket does to concentration. Fewer, larger customers means each one carries more weight, and a single account’s demands start steering the roadmap for everyone. That is a different flavor of the same risk in letting one big customer own your roadmap. Bigger contracts are more durable, but they are also louder, and they do not hesitate to use that volume.
The reason this matters is that a product strategy is a set of choices about who you serve and what you will decline. When the answer is “everyone, at every size, with every feature they ask for,” you do not have a strategy. You have a feature list that grew until it stopped meaning anything, and a product that got heavier for the people who loved it first.
Going upmarket can be the best decision a company makes. The retention math is real, the expansion revenue is real, and for many products it is the only path to the kind of durable business the spreadsheet is dreaming about. Just treat it as what it is: a deliberate, largely irreversible bet on a different customer, made once, with full knowledge of what you are agreeing to build and who you are agreeing to leave behind. The companies that get hurt are not the ones that chose to climb. They are the ones who looked up one day, realized they had already climbed, and could not remember deciding to.