The whale you didn't chase

Six months of redlines for a deal that was never yours is a real and measurable failure mode: but going bottom-up does not let you skip enterprise sales, it lets you do it after the seats are already in the building

Almost every founder has a version of this story, and it always has the same shape.

A large company gets in touch. The calls go well. You mention it at the all-hands, possibly with the logo on a slide. Then the paper cuts start: redlines, a security questionnaire, three custom features, a SOC 2 report you do not have yet, and someone asking whether any of this could run on-prem. Each request is small. Together they quietly become the company's roadmap. And then, right at the end, the hammer: they want monthly billing after all, or a competitor turns up with the playbook you spent six months building with them and quotes 30% less.

I wanted to know how much of that is real and how much is the thing founders say to each other for comfort. Mostly it is real, and the numbers are worse than the anecdote.

The failure mode has a name and a base rate

The part that feels most unfair, that the deal was never yours, is the part with the strongest evidence behind it. Matthew Dixon and Ted McKenna ran machine analysis over 2.5 million recorded sales conversations for The JOLT Effect and found that 40 to 60% of qualified deals are lost to no decision. Not to a competitor: to nothing. The buyer simply never buys, from anyone.

Their breakdown of those losses is the useful bit. Only 44% were people who preferred the status quo. The majority, 56%, were people who wanted to buy and could not get themselves over the line, usually out of fear of being the person who picked wrong. Which means the champion who kept telling you it was going well was probably telling the truth, and it did not matter.

The time cost is easier to pin down. Benchmarks across several hundred B2B companies put deals above $100,000 in annual contract value at 90 to 180 days and often longer, with strategic accounts over $500,000 now planned around 9 to 12 months as the baseline rather than the exception. Each extra person added to the buying committee stretches it further. A committee that grows from five people to ten adds something like six to twelve weeks on its own, and nobody tells you when it grows.

The roadmap capture is a recognised risk too, with numbers attached. The conventional customer concentration thresholds are that any single customer under 10% of revenue is healthy, 25 to 50% is where investors start asking pointed questions, and past 50% is where companies break. The financial risk is the obvious problem. The less obvious one is that an account that large gets a vote on what you build, and it will vote for things only it needs.

So: the diagnosis holds. Chasing a Fortune 100 logo as a five-person company is a coin flip that takes half a year to land, and the coin is weighted toward nothing happening.

Where the usual prescription goes wrong

The advice that follows is normally some version of: price it so cheap a developer can expense it, bill monthly, let it spread inside the company on its own, and the whales will come to you. The sequencing is right. The promise underneath it, that you get to skip the enterprise part, is not, and the companies held up as proof are the ones that disprove it.

Datadog is the cleanest example because the numbers are in its filings. As of 30 June 2026 it had roughly 33,400 customers. About 4,720 of them spent $100,000 a year or more, and those 4,720 accounted for 91% of ARR. Fourteen percent of the customers, ninety-one percent of the money. The other 28,000 accounts are not the business. They are the mechanism that produces the business.

Figma's IPO paperwork makes the same point from the other end. The S-1 says that 78% of the Forbes Global 2000 used Figma in March 2025, and then, in the risk factors, that only 24% of the Forbes Global 2000 spent more than $100,000 a year on it. Read those two sentences together and you have the entire argument. Figma got inside three quarters of the largest companies on earth without a procurement process. Turning that presence into money is a separate job, done later, by salespeople, and by Figma's own account it was still mostly undone at IPO. Its net dollar retention was 132%, which is what that job looks like when it works.

Slack and Atlassian went the same way. Both are cited constantly as proof that bottom-up beats enterprise sales, and both built enterprise sales teams. This is well-trodden ground for investors: a16z frames it as the $20M to $500M question, and Bessemer has a whole playbook for bolting sales onto a product-led company. Nobody who has done it describes it as optional.

What bottom-up actually buys you

It is not an escape from the redlines, the security review or the on-prem question. Those are still coming. What changes is when they arrive and who is under pressure when they do.

Run the same negotiation twice. In the first, you are a slide deck and a pilot. Every custom request is a condition of getting anything at all, the security review is an exam you might fail, and the competitor who undercuts you at the end is undercutting a promise, which is easy to undercut. In the second, four hundred people inside the building already use the product daily and a manager somewhere is expensing it on a card. Now the redlines are a formality about something that is already load-bearing. The security review is a box to tick so the thing everyone uses can keep being used. And the competitor at 30% off is asking a department to change how it works to save money the department does not spend.

That is the actual trade, and it is worth being precise about it, because the fuzzy version of this advice sets founders up to be surprised twice. First when the enterprise work turns out to still be required, and second when they discover they hired nobody who can do it.

The version I would actually give

Do not run a six-month enterprise process for a logo before you have usage inside that logo. That is the defensible half of the advice, and the no-decision numbers are the reason: you are spending your only scarce resource on an outcome that most often is not an outcome at all.

But do not tell yourself the enterprise motion goes away. Price so a single person can start without asking anyone. Bill monthly so nobody needs a budget cycle. Then watch for accounts where usage crosses some threshold you have decided on in advance, and go and sell those properly, with a real contract and a real security posture, because that is where the revenue is. Datadog's 14% is not an accident of product-led growth. It is a sales team working a list that the product wrote for them.

The line about whales swimming to you is right about the swimming and wrong about what happens next. They swim over. They do not climb into the boat.

If the harder problem is that nobody is swimming anywhere yet, that is the part we work on.