Your small customers churn, your average deal is too small to support a salesperson, and the obvious answer is larger buyers. You update the target list and the deck.
The list was never the constraint. A larger buyer does not evaluate you, they place you in a class, and everything about how you look decides which one.
Why these three models
The decision is whether to move upmarket, and what has to change first. The features that fire are a buyer who screens by category rather than by assessment, a sales cycle long enough that the evidence arrives after the commitment, and a change that is a shift of state rather than a shift of emphasis.
Three lenses. Statistical discrimination produces an equilibrium answer about how you are sorted. Delay produces a complex answer about why the move will be over-corrected. Regime states produce a cycle answer about what actually changes. The second is the one that turns a reasonable strategy into an expensive one, because it makes the failure and the success look identical for two quarters.
1. They are sorting you, not assessing you
A large buyer’s procurement process exists because individual assessment is expensive. So they substitute the average of a group, and the whole exercise is about establishing which group you belong to before anyone examines what you do.1
The class markers are mundane and they are not your product. Do you have a security questionnaire already answered. Is there a data processing agreement. Can you name three customers of comparable size. Do you have an audited financial statement, a stated uptime record, insurance, a legal entity in the right place, a support commitment with hours attached.
Each of those is checkable and cheap for a competent vendor to hold, which is exactly what makes it a class marker. Their absence does not suggest you are worse. It places you in the class of vendors who have not done this before, and that class has a price and a probability attached that no product demonstration overrides.
Which reframes the first move. Before changing the target list, spend a quarter assembling the artefacts, because they are the entry condition rather than the follow-up. Most of them take days rather than months and they are the difference between being evaluated and being sorted.
The list is regional as well as generic. A large buyer in Kampala, Nairobi or Dar es Salaam adds items a Delaware checklist does not contain: a tax clearance certificate that is current rather than expired, registration on the buyer’s supplier portal before the tender opens, and a bank account in the buyer’s currency so paying you does not require an exception from their finance team. A ministry or a donor-funded programme adds one more, which is evidence that you were not created last month. A certificate of incorporation dated eleven weeks ago is itself a class marker, and the only thing that overrides it is a customer of size who will take the call.
One distinction the checklist blurs. A reference and a class marker do different work. The marker gets you past the filter and it is binary: the questionnaire exists or it does not. The reference decides the price and the terms once you are through, and it is comparative: three customers at the buyer’s scale beat thirty at a tenth of it, because the question being asked is whether anyone with their problems has survived you. Founders answer with volume when the buyer asked about scale. If you have no customer of that size, say so and offer the closest one with the numbers attached, because a gap you declare costs less than a gap they discover.
You are not being rejected on the product. You are being filtered before anyone looks at it.
One honest caution. Some of these markers are genuinely expensive, particularly formal security certification, and a founder who pursues all of them before any upmarket revenue has spent a year on procurement readiness for a market they have not tested. Assemble the cheap ones first, and let a real buyer tell you which expensive one is actually binding.
2. The evidence arrives after the commitment
The second lens is where the money goes, and it is a timing problem rather than a strategy problem.
An upmarket sales cycle is long. Six to twelve months from first meeting to signature is ordinary, and that delay sits between every action you take and any information about whether it worked. A control loop with a long delay does not simply respond slowly; it overshoots, because the operator keeps pushing while nothing comes back, then corrects hard when the accumulated result finally arrives.2
The company version is recognisable. Two quarters of upmarket effort produce nothing, because nothing was ever going to close in two quarters. The founder concludes it is not working and returns to the small segment. Three months later two of the enterprise conversations close, arriving into a company that has just repositioned away from them. Then the correction reverses.
The remedy is not more patience, which is unenforceable when cash is tight. It is to measure something that moves early. Meetings held with the right buyer title. Security questionnaires received, which is a strong signal because nobody sends one casually. Proportion of conversations reaching a second meeting. Those move within weeks and they tell you whether the class-marker work is landing, long before any revenue does.
It is worth knowing how stubborn this failure is before assuming discipline will cover it. In a controlled setting with demand held nearly constant, participants including experienced executives still produce large swings, run costs about ten times the achievable optimum, and afterwards describe an external cause that never occurred.4 The over-correction is not impatience, it is what a long delay does to anyone operating inside it.
There is a cheaper version of the whole test for a company that cannot fund four quarters of it. Run one upmarket conversation at a time while the small segment operates unchanged, and treat that conversation as a purchase of information rather than as a pipeline. One buyer, carried to the point where they either send paper or stop replying, tells you which markers bind, how long your cycle actually is, and what their contract does to your payment terms. That costs a few weeks of founder attention instead of a repositioning, and it produces the numbers you need to size the real attempt.
Plan for the counterparty who does neither thing the section assumes. A large buyer is expected to progress or to decline. The common outcome is that they go silent for two months and then resume as though no time passed, because your contact was reassigned or the budget year closed. Do not read the silence as a decision and do not restart the sale when they come back. Ask at the second meeting who signs, when the budget cycle opens, and what has to be true by which date. A cycle you cannot shorten is at least one you can put on a calendar.
And size the commitment against the lag deliberately. If the cycle is nine months, a two-quarter trial cannot produce evidence about revenue, so either fund four quarters or run the test on the early indicators and say that is what you are doing.
3. This is a state change, not an adjustment
The third lens is about what you are actually deciding, which is larger than the sales conversation suggests.
A business alternates between states with different rules rather than sliding along a continuum, and the useful questions are which state you are in and what a transition costs.3 Upmarket is a different state, not a stretch of the current one, and most of the change is not in sales.
Support gets a response-time commitment. The product acquires roles, permissions and an audit log. Pricing moves from published to negotiated. Contracts get redlined by someone else’s lawyer. Payment terms lengthen, which changes working capital before it changes revenue. Roadmap decisions start being made by whoever shouts loudest among four large customers.
Each of those is survivable. Together they are a different company, and the failure mode is attempting both states at once: a small-customer product and pricing with an enterprise sales motion, which serves neither and costs more than either.
The working-capital item deserves its own sentence because it surprises people who have modelled everything else. Enterprise payment terms of sixty or ninety days mean that a successful upmarket move makes your cash position worse before it makes it better, and the size of the gap scales with how well the strategy is working. A company that wins four large contracts and cannot fund the receivables has succeeded its way into a crisis.
Two things follow from that and only one of them is financial. Price the delay into the deal rather than absorbing it, because a discount for payment in thirty days is cheaper than invoice discounting at East African rates, and a buyer whose finance team can actually move will sometimes take it. The other is that your own obligations do not move to ninety days because your customer’s terms did. Staff are paid monthly, a landlord quarterly, a supplier on delivery. The gap between those and collection is not a forecasting problem, it is a fixed sum you have to be holding before you sign. Compute it as one quarter of expected upmarket revenue at the terms written in their contract, not at the terms printed on your invoice.
So the honest version of the decision is whether to transition, with a date, rather than whether to try. And the specific thing to check first is whether your existing small customers can survive the reduced attention, because they are funding the transition and they will notice.
What the three say together
- Assemble the cheap class markers before changing the target list. Security questionnaire, data processing agreement, three comparable references, stated support hours.
- Measure early indicators rather than revenue: right-title meetings, questionnaires received, second-meeting rate.
- Fund the trial against the actual cycle length, or state openly that you are testing indicators rather than revenue.
- Decide whether you are transitioning states, with a date, rather than trying both at once.
Where they disagree
The class-marker argument and the regime argument recommend opposite sequences.
Sorting says do the preparation first: the artefacts are cheap, they are the entry condition, and going to market without them wastes the conversations you get, which are the scarce resource. The regime argument says the preparation is a guess about which markers bind, that different buyers care about entirely different ones, and that a quarter spent assembling documents nobody asked for is a quarter of runway spent on a market you have not yet met.
The reconciliation is to split the markers by cost. Do the cheap ones unconditionally, because they take days and are demanded by almost every buyer of size. Let the first real conversation tell you which expensive one binds, and be willing to lose that first deal in exchange for knowing. Founders usually do neither, arriving with none of the cheap markers and then pursuing an expensive certification because one prospect mentioned it.
What none of them contain
None of the three prices what upmarket does to the people who work there. Enterprise selling is a different job with different rhythms, and a team assembled for volume and speed will find long cycles and procurement processes demoralising in ways that show up as attrition rather than as feedback.
None of them handles the possibility that the small segment was the better business. All three treat upmarket as the destination and the current state as the problem, which is the framing the decision arrives with. A high-churn small segment sometimes indicates a product problem that a larger buyer will also have, discovered more slowly and at greater cost.
None of them prices the cost of doing this twice. A company that moves upmarket, retreats, and moves again has spent the class-marker work, the sales-cycle patience and the small-segment attention, and arrives at the second attempt with a team that has learned the strategy does not hold. The models describe a single transition and the expensive version is the oscillating one.
And one property the ensemble will not produce: your first large customer will shape the product more than any decision you make deliberately. Whoever signs first sets the roadmap for a year through the accumulation of reasonable requests, and choosing them for fit rather than for speed is worth more than the terms you negotiate with them.
The one action that survives the ignorance: list the cheap class markers you do not currently have and put a date on each. If you cannot produce a security questionnaire answer, a data processing agreement and three comparable references, the upmarket conversation you are planning will end before your product is discussed.
Who has to move
The founder, and the first quarter of work is administrative rather than commercial, which is why it does not happen. The instinct is to change the target list and the deck, because those feel like the strategy, while the artefacts feel like paperwork someone should do later. The cheapest first test is to ask one friendly large company to send you their standard vendor onboarding pack before you pitch anyone. Reading it takes an hour and it converts the entire question from a judgement into a checklist.
Sources and notes
- Jeffrey Carpenter and Andrea Robbett, Game Theory and Behavior, MIT Press. Statistical discrimination, in which decision-makers substitute the average of an observable group when individual assessment is costly, is treated alongside signalling and screening. Used in section 1. The corollary drawn here is that the class markers are valuable precisely because they are cheap for a competent vendor and therefore separate the vendors who have done this before from those who have not.
- John D. Sterman, Business Dynamics: Systems Thinking and Modeling for a Complex World, McGraw-Hill, and Donella H. Meadows, Thinking in Systems: A Primer, Chelsea Green Publishing. The behaviour of a balancing loop containing a delay, including overshoot when the correction is sized as though the delay were zero, is developed in both. Used in section 2, with the standard remedy: measure closer to where the state changes rather than responding faster.
- Sanjit S. Dhami, The Foundations of Behavioral Economic Analysis, Oxford University Press, and the regime-switching literature it draws on. The property used in section 3 is that a system alternating between persistent states with different rules is poorly described by a continuum, so the useful questions concern which state you are in and what a transition costs. No transition probability is claimed for any market.
- John D. Sterman, The Beer Distribution Game, MIT, https://web.mit.edu/jsterman/www/SDG/beergame.html. Participants in a four-layer chain with nearly constant demand produce large oscillations anyway, at average costs about ten times the optimum available from the information they hold, and afterwards attribute the swings to external causes that did not occur. Cited in section 2 as evidence that the over-correction under delay is structural rather than a failure of the individuals running the strategy.
A note on a number this article does not give. There is no deal size at which upmarket becomes correct, and no standard sales-cycle length to plan against. Both are specific to your buyer and your category. What transfers is that you should measure your own cycle from your last three long conversations and fund the trial against that number rather than against a quarter.
Joshua Agonya Pi’Rwot, Founder.