The gap between what was promised and what happened I've been looking at the IFS analysis on Help to Buy, and it confirms something I've suspected for years: well-meaning policy can do the opposite of what it's supposed to do. The scheme launched in 2013 with a clear goal. Get more people onto the property ladder. Particularly those who couldn't afford it otherwise. Instead, it mostly helped people who could already afford to buy. Higher-income households captured the majority of the benefit. That's not a minor implementation detail. That's the entire point of the scheme failing. This matters to anyone running their own operation because it shows something crucial: systems have second-order effects. You can't just design something for outcome A and assume outcome A is what you'll get. The real world is messier than that. How good design can accidentally exclude the people it's meant to help Help to Buy worked like this: the government offered an equity loan covering up to 20% of the property price (in London, up to 15%). You needed a 5% deposit. For someone earning £25,000 a year trying to buy in most of the UK, this should have been transformative. Except it wasn't. Here's why. The scheme worked best for people buying new-build properties in areas where prices were rising. That meant South East England, London, and affluent commuter belts. Poorer households, concentrated in different areas and different property types, didn't have access to the same stock of Help to Buy eligible homes. The scheme didn't exclude them on paper. Geography and property availability did it instead. Secondly, mortgage affordability tests meant lenders still needed to see stable income and good credit history. A household on £25,000 with patchy employment history or previous credit issues found the mortgage harder to get, even with the government backing 20% of the property. The scheme removed one barrier but left the others standing. Thirdly, and this is the one I see happen in my own work all the time, the scheme became a tool for people who were already close to buying. Someone earning £60,000 who was saving for a deposit? Help to Buy meant they could buy sooner and buy more. Someone earning £25,000 with no savings and no family support? The scheme didn't solve the fundamental problem: they didn't have the deposit buffer in the first place. The mechanism that made things worse Here's the unintended consequence that really matters. Help to Buy injected demand into the housing market. More buyers, backed by government guarantees, meant more competition for limited properties. Prices went up. When prices go up, the people already in the market benefit. Those trying to enter it face higher thresholds. The scheme meant someone on £60,000 could now access properties at £350,000 instead of £300,000. But someone on £25,000 was now competing for properties that had just become more expensive, not less. The IFS found that the scheme had little effect on social mobility. That's economist-speak for: it didn't help the people it was supposed to help. It helped people who were already on their way up. This is a pattern I've noticed across policy and product design. You create a system to solve problem X. The system works, but it attracts people who weren't actually your target. It changes the market conditions. And suddenly the people you were trying to help are worse off than before because the terrain has shifted. What this teaches us about building anything I work with people building their own thing, and they face similar challenges. You design a product for one user. You launch it. A different user finds it more useful. Your original target gets squeezed out by market dynamics you didn't anticipate. The Help to Buy case is instructive because it had unlimited resources and government backing. If government can't predict second-order effects, what chance do the rest of us have? The answer isn't to give up trying. It's to test your assumptions before you go all-in. When I'm helping someone at ALIRA work through a new offering or policy, we use a 5 Whys approach to trace the actual chain of cause and effect. Not the intended chain. The real one. What actually happens when you change this variable? Who actually benefits? Who gets left behind? With Help to Buy, the questions should have been: where do lower-income households actually want to buy? What's the real barrier stopping them? Is it deposit size, or is it income verification, or is it something else entirely? You can't design a solution without answering those questions properly. The scheme assumed the barrier was deposit size. It turned out to be multiple things, stacked together. By solving only one, you didn't help. You just changed which problem was most acute. The broader point I'm not saying Help to Buy was a disaster for everyone. Some people did buy homes who wouldn't have otherwise. But the IFS analysis shows it didn't narrow the gap between rich and poor. It widened it slightly. The higher-income households got the benefit. The lower-income households watched house prices rise faster. This happens because policy and product design don't work in isolation. They work inside systems. Those systems have feedback loops. Change one thing and three other things shift in response. You need to think through those chains before you launch. If you're building something, whether it's a product, a service, or a policy, trace the consequences. Not just the intended ones. The ones that actually happen when real people interact with your system. That's where the real learning is. What to do this week If you're designing anything new, spend 30 minutes this week mapping out what happens after your solution works. Use a Decision Matrix or 5 Whys tool to trace the actual chain of cause and effect, not the intended one. Ask someone who isn't your target user to poke holes in your logic. They'll see the second-order effects you've missed. Then decide whether those effects matter enough to change your approach before you launch.