Every property in your portfolio review has one thing in common. It's still in your portfolio.

That sounds too obvious to matter. It might be the most important fact in the room. The report in front of you was built from the assets you still own, the residents who still pay, and the deals that actually closed. Everything that failed on the way here is missing. The building you sold at a loss and quietly took off the report. The tenants who left. The deals that died in diligence. Not down-weighted in the numbers. Absent from them.

So when you read that report to learn what works, you're really asking the winners why they won. Their answer will always sound convincing, and it will always leave out the one thing that could have told you what actually kills a property.

The Armor Goes Where The Damage Isn't

The sharpest version of this comes from Abraham Wald, a statistician who worked for the U.S. military during the Second World War. The Air Force was losing bombers and wanted to add armor. Armor is heavy, so it had to go only where it would do the most good.

The natural approach was to examine the planes coming back, find where they were riddled with holes, and reinforce those spots. When the damage was mapped, it clustered in some parts of the aircraft and barely touched others.

Wald's insight was to reinforce the parts that came back clean.

The reasoning is the whole idea in a sentence. Every plane he could study was a plane that made it home. The damage they carried marked the places a bomber could be hit and still fly. The areas that looked untouched on the survivors were not areas that never got hit. They were the areas where getting hit meant you didn't come back to be counted. The evidence that mattered most had already removed itself from the sample.

If you want the primary source, the American Mathematical Society has a detailed account of Wald's original memoranda.

That's the line to sit with. The damage you can see is, by definition, the damage that wasn't fatal. The fatal damage left before anyone started counting.

Your Data Was Filtered By The Thing You're Trying To Measure

This is more than a wartime curiosity, and property is unusually exposed to it.

Your reporting is a sample, and survival chose the sample for you. The properties that went worst are the ones most likely to have been sold, written down, or dropped from the standing report. The residents with the worst experience are the ones no longer on the rent roll. The deals that went wrong are the ones that never reached the pipeline you now study. At every step, the process that produces your data quietly deletes its own failures before handing you the summary.

This is worse than an average hiding a bad asset. When an average buries a struggling property, that property is still in your data, blended into the mean, and you can sort the distribution and find it. Survivorship bias takes the evidence out of the room entirely. No amount of sorting recovers a tenant who moved out eighteen months ago and left no row behind.

The result is a confident kind of wrong. Run any analysis on surviving data and it will tell you that your survivors' traits are the traits of success. They aren't. They're the traits of survival, which is a narrower and much less useful thing to know.

The Winners Are Quietly Doing Your Thinking For You

Look at how this plays out across the business and the same pattern keeps surfacing.

Your five-year portfolio performance is really the performance of the properties you decided to keep. The weak assets you offloaded in year two stopped dragging the number down the moment they left, so the portfolio looks as though it was always this healthy. What the figure actually rewards is your willingness to cut losers, which is a real skill but not the one the report claims to prove, and not one you can find anywhere in the line.

Retention and satisfaction scores have the same problem in a sharper form. They're collected from the people still living there, so the resident who left after a bad maintenance experience isn't in the survey. They're the reason it reads clean. A satisfaction number built only from current residents measures the people who haven't left yet, which is close to circular.

Acquisitions are no different. Your instincts are trained on the deals you did, not the ones you passed on that would have been excellent, and not the ones that fell apart under scrutiny. The pattern you trust grows more confident over time without ever being tested against the cases where it failed.

"Let's Learn From Our Best Properties"

Survivorship bias walks in the moment someone says let's study our top performers, or asks what your longest-staying residents have in common. It sounds like rigor. It's the opposite.

Studying only the winners guarantees you'll never see the trait that sinks a property, because any property with that trait is no longer a winner, and may no longer be yours. You can examine your best assets forever and learn nothing about what fails them, for the same reason Wald couldn't learn about fatal damage by staring harder at the planes on the runway. The lesson was never in the survivors.

Go Looking For The Planes That Didn't Come Back

The fix isn't more analysis of what you have. More scrutiny of the survivors only makes you more certain about survivors. The work is to rebuild the part of the picture your systems are built to forget.

Start with the assets you exited. When you report performance over any long stretch, keep the properties you sold in the record, or at least show the number both ways, with and without them. The gap between those two figures is one of the more honest things you can put in front of a board, because it separates owning good assets from being good at dumping bad ones. Those are different businesses with different futures.

Then talk to the people who left. Exit conversations with departing residents are the least biased data you can gather, precisely because survival hasn't filtered them. The leavers are the only group who can tell you why people leave. A renewal survey can't, because everyone in it renewed. If you run a single satisfaction study, run it at move-out.

And keep the failures on purpose. Track the deals you passed on and the ones that collapsed, then look again a year later to see where you were right and where you weren't. Nothing in your normal reporting will assemble this for you. Left alone, every system you own will keep tidying away its own mistakes and handing you a clean sheet with the lesson removed.

You can sense whether this is already happening with three questions. When someone says portfolio performance, does the number still include what you sold? When you look at a retention score, where's the data on the people who left? And for any profile of a good tenant or a good deal, was it ever tested against the ones that fit the profile and failed anyway? When those answers go missing, survivors are running your analysis, and they'll keep telling you everything is fine.

The Takeaway

The report in your portfolio review isn't the story of your portfolio. It's the story of the parts that survived, and everything that didn't work has already been shown the door. That's why it reads so calm. The failures aren't being downplayed in the numbers. They were never allowed into them.

Wald's discipline was to treat the missing planes as the real information and armor against the damage he couldn't see. In property, the equivalent is to stop learning only from what remains. Keep the sold assets in the record. Interview the residents who leave. Hold on to the deals that died. Your reporting is built to forget all three, and if you let it, it will hand you a clean, confident picture of a business you don't actually run.

Holding the full history of every asset and resident in one place, the ones that left alongside the ones that stayed, rather than a report that silently drops whatever is no longer current, is much of what RIOO's dashboards and reporting are built to do. And when the problem isn't just missing history but a system that can't surface a straight answer at all, that failure has its own cost, one we covered in the cost of a system that can't answer a simple question.

FAQ

1. What is survivorship bias?
It's the error of drawing conclusions only from the things that made it through a selection process, while ignoring the things that didn't. Abraham Wald's WWII bomber analysis is the classic case: the returning planes showed where a bomber could be hit and survive, so the armor belonged where the survivors weren't damaged. The planes that were shot down, the most important data, had already removed themselves from the sample.

2. How does survivorship bias show up in property management data?
Your reporting is built from what survived: assets you still own, residents still paying, deals that closed. The properties you sold, the tenants who churned, and the deals that fell through leave little or no trace, so any analysis of "what works" is really an analysis of what lasted. The failures aren't hidden in the numbers. They're absent from them.

3. Why is tenant retention data especially vulnerable to it?
Because most retention and satisfaction studies survey current residents, who are by definition the people who haven't left. A strong score can sit right on top of serious churn, because the residents unhappy enough to leave are exactly the ones no longer available to survey. The only clear view of churn comes from the people leaving, not the ones staying.

4. How do I correct for survivorship bias in portfolio reporting?
Deliberately rebuild what your systems delete. Keep sold and written-down assets in long-run performance figures, or report the number both with and without them. Collect exit data from departing residents, not just renewal data. Track the deals you lost or passed on and revisit them later. The correction is never more analysis of the survivors. It's restoring the failures the reporting was built to forget.

5. Isn't focusing on top performers just good practice?
Studying your best assets helps you understand excellence, but it can't teach you what causes failure, because anything that caused failure has already left the group you're studying. If the goal is to stop losing properties and residents, the top performers are the wrong place to look. You have to examine the ones that didn't make it, which means you first have to stop throwing their data away.