Survivorship Bias Is a Marketing Strategy Problem

During World War II, the U.S. military studied damage patterns on bombers returning from combat missions. The bullet holes clustered on the fuselage and wings. The logical response seemed obvious: reinforce those areas. That is where the planes were getting hit.

A statistician named Abraham Wald looked at the same data and reached the opposite conclusion. Reinforce the engines, the cockpit, the areas with almost no bullet holes at all.

The reason was simple. The military was only studying the planes that made it back. The planes hit in the engines and cockpit had not returned to be studied. They were shot down over enemy territory and absent from the dataset entirely. The bullet hole map was not a picture of where planes got hit. It was a picture of where planes could get hit and still fly home.

The military, to its credit, listened. They shifted where they added armor. More planes made it back.

Most founders know this story. Almost none of them apply it to their GTM strategy.

The case studies you read are all bullet holes on returning planes

Slack's bottom-up PLG motion. Notion's community flywheel. HubSpot's content engine. Dropbox's referral loop. These are the canonical GTM stories. They get told at conferences, written up in blog posts, turned into frameworks and playbooks, and studied by founders building their first go-to-market strategies.

All of them are planes that came back.

Only 13% of SaaS companies ever reach $10 million in ARR after ten years. The GTM strategies in every case study you have read came from that 13%. The other 87% are not publishing post-mortems. They ran PLG motions that never got bottom-up traction. They built content engines that produced traffic but not pipeline. They launched referral programs that generated noise but not revenue. They are not writing case studies about it. They are not speaking at SaaStr. They are not in your dataset.

The question is not whether a GTM motion worked for a company you admire. It is whether the conditions that made it work for that company exist in yours.

This is not an argument against studying successful companies. It is an argument against the specific error of assuming that what is visible in the dataset is representative of what actually works. The bullet holes on the returning planes looked like they were telling you something important. They were. Just not what everyone thought.

The mechanism: superstitious learning

Researcher Justin Miller studied what he calls superstitious learning in startup strategy: copying features of successful companies under the assumption that those features caused the success, when they may only be correlated with it.

In stable markets, the lesson is usually clear enough that less superstitious learning occurs. In changing markets, the causal links are harder to verify and founders are more likely to copy the wrong things. An example Miller uses: if you read that companies founded by college dropouts become unicorns at a higher rate than expected, and you decide to drop out because of that data, you have committed superstitious learning. The dropout is a correlate of a certain risk profile and founding context. It is not a cause of startup success.

Applied to GTM strategy, the mechanism works like this. A company uses product-led growth and becomes a case study. You read the case study and adopt PLG. What you cannot see from the case study: whether the company's product had inherent network effects that made PLG natural, what their ACV was, how self-serve their buyer journey already was, and whether the timing of their market entry created conditions that no longer exist. The PLG motion was visible. The underlying conditions that made it viable were not in the write-up.

You copied the bullet holes. You did not ask which planes those holes appeared on.

The specific case of product-led growth

PLG is the most widely copied GTM strategy of the last decade and also the one most subject to survivorship bias distortion. The case studies that get published are from companies whose products have specific structural features: network effects or collaboration utilities that make individual adoption viral, low price points that allow self-serve conversion without procurement approval, and buyer journeys that start with personal use before expanding to organizational buy-in.

Slack spread inside companies because every person someone messaged on Slack became a new user. Notion spread because shared documents pulled non-users into the product. Calendly spread because every scheduling link sent was a product demo for the recipient. The PLG motion was not the strategy. It was the natural expression of a product architecture that made adoption inherently viral.

The companies that tried PLG with $40,000 ACV enterprise products, multi-stakeholder buying committees, and procurement requirements do not appear in that case study library. They waited for bottom-up adoption that never arrived at the speed the model requires. They ran out of runway. Their story is the engines on the planes that didn't come back.

This is not a critique of PLG. In 2026, GTM motion should match ACV: product-led for under $5,000, hybrid for $5,000 to $50,000, sales-led for above $50,000. PLG cuts CAC by up to 40% for the companies it fits. The failure is not PLG itself. It is the assumption that the visibility of the case studies means the strategy transfers.

The Wald Question for GTM strategy

Wald's insight was not that the military should ignore the data they had. It was that the data they had was incomplete in a specific and predictable way. The fix was not more data. It was a different question: where are the planes that didn't return, and what do those missing planes tell us?

Applied to GTM strategy, the Wald Question is this: what are the characteristics of the companies that ran this same motion and failed?

The companies winning at GTM in 2026 are not the ones with the most impressive case study library. They are the ones who matched their GTM motion to their specific market conditions: their ACV, their buyer's journey, their product's natural adoption path, and the competitive density of their category right now. Most of those companies are not the ones you read about.

Wald did not tell the military to stop studying the returning planes. He told them to notice what the study was missing and adjust their conclusions accordingly. The planes that came back were still useful data. They just could not tell the whole story on their own.

The GTM case studies you read are still useful. Notion's community strategy teaches you something real about how communities can drive adoption. HubSpot's content engine teaches you something real about organic demand. What they cannot tell you is whether the conditions that made those strategies work for those companies exist in yours.

The job of a Portfolio CMO, before any channel is chosen or any playbook is adopted, is to ask the Wald Question. Not "which strategy worked for a company we admire?" But "what are the conditions of our specific market, and which motion fits those conditions?" That answer does not come from a case study. It comes from knowing your buyer, your product, and your market well enough to build the strategy from the inside out rather than copying it from the outside in.

The reason that question is hard to answer from the internet is that most failures are quiet. Post-mortems are rare. Companies that ran a GTM motion and ran out of runway do not publish retrospectives. The insight about what did not work comes from pattern recognition across many companies at similar stages. That is what working with someone who has seen both sides of the dataset actually gives you: not just the case studies worth copying, but the failure patterns worth avoiding before you repeat them.

That is the work that happens before the first campaign brief. If you want to know whether your current GTM motion was built from your market conditions or from someone else's case study, that question is where a Marketing Audit starts →