The Scurvy Problem: Why Companies Keep Forgetting What Makes Customers Stay
The Royal Navy cured scurvy in 1747 and had forgotten the cure by 1911.
That is not a figure of speech. James Lind ran his citrus trial in 1747. By the end of the century, lemon juice was standard naval issue, and a disease that had killed more British sailors than combat nearly disappeared from the fleet. A hundred years later, Robert Falcon Scott's men were dying of scurvy in Antarctica, and the expedition's own physician did not believe citrus was the cure. The knowledge did not decay quietly at the edges. It was official policy, and then it was gone.
Every enterprise B2B SaaS company I have worked inside believes it knows what makes its best customers stay. Most of them are wrong in exactly the way the Navy was wrong. They know a rule, not a mechanism. And a rule that nobody understands is a rule that will break the moment conditions change, which in a growing company is constantly. This post is about why retention insight keeps evaporating, why product marketing is the function that should be preventing it, and what owning that knowledge actually requires.
The Cure That Kept Disappearing
The Navy never understood why citrus worked. That is the whole story.
Lind's sailors got better, so lemon juice went into the ration. But the accepted theory held that the cure was the acidity, not anything specific to the fruit. That mistaken belief sat quietly for decades doing no harm, because the fleet happened to be issuing Mediterranean lemons, which are rich in what we now call vitamin C. Around 1860 the Navy switched to West Indian limes, sourced cheaply from the Caribbean colonies and, being more acidic, assumed to be more potent.
Limes carry roughly a quarter of the vitamin C that lemons do. The juice was then concentrated through copper equipment and left exposed to air, which destroyed most of what little remained. The ration was now close to useless, and nobody could see it, because faster steam voyages had shortened time at sea enough that scurvy rarely had a chance to appear.
So the fleet ran an ineffective cure for years while believing it was protected. When scurvy resurfaced on long polar expeditions, the accumulated evidence pointed the wrong way. Fresh meat prevented the disease and preserved food did not, which made a bacterial-contamination theory look reasonable. By the time Scott's men were in Antarctica, a century of "progress" had buried a cure that had once been law.
Here is the distinction that matters for anyone running a commercial motion. The Navy knew the correlation and never knew the cause. Correlational knowledge is fragile in a specific and dangerous way: it cannot tell you which conditions it depends on. The rule "citrus prevents scurvy" carried a hidden clause the Navy never read, which was "as long as the citrus actually contains vitamin C and you do not cook it out." When the hidden clause was violated, the rule failed silently, and no one could reconstruct why because the reason had never been written down. That same fragility governs what your company thinks it knows about why customers stay.
Your Company Already Knows the Rule
Andrew Chen, in The Cold Start Problem, describes the method growth teams use to figure out whether a product is actually working. You group users into cohorts, you track how many are still active after one day, seven days, thirty days, and you watch the shape of the retention curve. You separate the high-value users from the low-value ones and you study what the high-value cohort does that the others do not. Chen treats the cohort retention curve as the foundational instrument for reading a product's health, and the analysis of what differentiates your best users as the work that tells you where the real value lives.
In B2B, that same knowledge almost always exists. It just does not exist as a system. It exists as folklore.
Across the PMM audits I have run over the past few years, the pattern is close to universal. Somewhere in the company, usually inside customer success, there is a person who can tell you the behavior that predicts retention. "Accounts that stand up a second use case in the first forty-five days basically never churn." They are often right. But that sentence is the West Indian lime. It is a correlation someone noticed once, never interrogated for mechanism, and never transferred to anyone who could act on it without asking that person first. Nobody can tell you which forty-five days, or what "stand up" means operationally, or whether the second use case is the cause or merely a symptom of a customer who was already committed. The rule feels like knowledge. It behaves like knowledge, right up until the conditions underneath it shift and it quietly stops being true.
That is the moment product marketing should exist to prevent, and usually does not.
The Function That Leaves the Room at Signature
Ask why retention knowledge stays tribal and the easy answer is that customer success is busy. The real answer is structural.
In most enterprise B2B SaaS orgs, the work of understanding the customer is split across functions that do not share a nervous system. Retention lives with CS. Product usage and the engagement ladder live with product and data. Product marketing's charter, in practice, runs from positioning through launch and stops at the close.
The one function whose actual job is to govern what is true about the customer and make that truth transferable across the commercial motion is not in the room after the deal signs. So the mechanism behind retention never gets encoded by the group that encodes everything else. It stays with whichever CSMs happened to notice it, and it walks out the door when they do.
This is not a competence gap. The CSMs are not failing. They are doing exactly what their role rewards, which is saving the account in front of them, not documenting the general principle underneath it. Product marketing is the function chartered to convert individual observation into institutional knowledge, and it has been pointed almost entirely at the pre-sale motion. Retention is treated as someone else's number. The predictable result is that the most commercially important thing a company can know, why its revenue renews and expands, is the thing least likely to be written down anywhere durable.
The gap is sharper in sales-led motions, where PMM's charter runs from positioning through launch and stops cold at signature. In product-led motions PMM tends to stay closer to the account, because usage data and in-product messaging keep the function tethered to the customer after the deal closes. Most enterprise B2B SaaS still runs sales-led, which is why the absence is the rule rather than the exception.
The internal alignment problem here is the same one that governs every handoff between the teams that carry the message. Before the market can trust what you say, the people inside the building have to share a single version of what is true, which is the entire premise of treating your own organization as Customer Zero. Retention knowledge is just the version of that problem that shows up after the sale, where PMM has quietly stopped showing up.
Cohort Curves Are an Encoding Tool, Not a Growth Dashboard
The instinct, when a company realizes it cannot explain its own retention, is to buy a dashboard. That is the wrong move, and it is wrong for the same reason the acidity theory was wrong. A dashboard shows you the correlation faster. It does not give you the mechanism.
Chen's method is more useful than the dashboard version suggests, because the important part is not the curve. It is the experiment. Chen describes using A/B tests to determine which behaviors actually improve engagement, which is the step that separates the behavior that causes retention from the behavior that merely travels alongside it. In a consumer product you run that test in the software. In enterprise B2B you rarely can, so you run it the slow way: you find the cohorts that retained and expanded, you find the ones that did not, and you interrogate the difference until you can name the specific value-realization moment that the good cohorts reached and the bad ones missed. Not the feature they adopted. The outcome they experienced, and the sequence of events that got them there.
This is product marketing work. Isolating the causal driver of retention is the same discipline as isolating the real reason a deal was won, and PMM already owns win-loss. Chen's engagement ladder, the idea that users climb through a sequence of deepening behaviors toward the core of the product, is not a product-team artifact when you translate it into enterprise terms. Every rung is a value moment that needs a message, an owner, and a place in the onboarding narrative. Turning that ladder from a chart into governed infrastructure that survives a reorg is precisely the operating-system work that scales a GTM motion instead of letting it reset every time a tenured CSM leaves.
I have watched this exact failure mode before, where GTM scaled fast enough that the founding team's institutional knowledge outpaced whatever got written down. The composite that captures it most cleanly: a vertical SaaS company around $52M in ARR, four years into the market, hired its first CS team early in year one to support its customers by hand. Three years later, as the company entered its first real growth push, it was hitting its first wave of departures among that original CS cohort. These were the people who had been there since before there was a playbook, and who had never written down the instincts the company now quietly ran on. Net revenue retention had held around 120% and slipped to 111% over two quarters, the first sustained down-quarters the company had seen, and leadership read it as a product problem. It was a memory problem, and it was early enough to still be fixable.
When we reconstructed the cohorts that had retained beautifully in the first years, the differentiator was not the feature-adoption metric CS reported on. The real driver was a thirty-day onboarding sequence the original CSMs had run on instinct, built around getting the customer to one specific operational outcome fast. That sequence existed nowhere as a document. The people who carried it were the ones now walking out the door, and the cure was leaving with them.
The Test of Whether You Actually Own It
There is a clean test for whether your company owns its retention mechanism or merely remembers it. Could a CSM who started last week deliver the value-realization moment that drives retention, correctly, without ever having met the person who discovered it? If the answer depends on tribal transfer, on shadowing the right veteran before they quit, you do not own the knowledge. You are the Navy in 1859, protected only by conditions that have not yet changed.
None of this argues that PMM should run CS, own the account relationship, or take over renewal conversations. CS owns execution, and should. What PMM should own is the mechanism: making sure the causal driver of retention gets identified, documented, and fed back into onboarding, messaging, and enablement consistently, so it does not live or die with whichever CSM happened to notice it first. Quarterbacking the mechanism is a different job than running the play, and conflating the two is what turns a legitimate charter question into a turf fight.
Owning it means the mechanism is documented as a mechanism, not a rule. It means the causal value moment is written into the onboarding narrative, the expansion plays, and the positioning itself, so that it is reinforced by the whole commercial motion rather than remembered by a few people. It means product marketing extends its charter past the close and treats retention insight as a first-class output, governed with the same rigor as launch messaging. The companies that compound do this. The ones that keep rediscovering their own retention driver every few years, always after the number has already slipped, are running the scurvy loop on their most valuable customers.
The knowledge is almost certainly already in your building. The question is whether it lives in a system or in a person who has not given notice yet.
If your NRR is drifting and no one can explain the mechanism behind it in a sentence that would survive that person leaving, that is the blind spot BlindSpot exists to close. We run retention and post-sale narrative audits that isolate the actual causal driver of your renewals and expansion, then encode it into governed onboarding, enablement, and positioning infrastructure so it stops walking out the door. Book a retention narrative audit and find out whether you own your cure or just remember it.
