Why do some features flop? image

I often hear businesses talk about investing a lot of effort in trying to achieve an outcome that is important to them, but with little success – I then draw a simple diagram that explains exactly why and offers a better way forward. I hope you find it useful too.

Consider a hypothetical online shop – there are three groups of people: the marketplace, sellers, and buyers. Originally, buyers just found a product on the marketplace that they wanted to buy, they provided a shipping address, and paid for it. Then one day, the marketplace decided that it would like to track what individual buyers were searching for across their phones, laptops etc. To do that, the marketplace created a ‘log in’ feature, but it saw very little usage. They tried to incentivize its use with a discount for registering, but buyers didn’t see any reason to log in after the one-off discount, so they didn’t.

Meanwhile, revenue was flat at the company – it was all coming from commission and that was a fixed proportion of sales, which weren’t going up. A team came up with a great new way of making money – sponsored listings: sellers could pay to get the top spots in search results. The marketplace built the feature – they were happy as it brought in a new source of revenue and the sellers that used the feature were pleased too. Meanwhile, buyers found it harder to find the products they wanted, they stopped trusting the marketplace and started shopping elsewhere.

In the Venn diagram below, there is a circle representing the interests of each group and I have also plotted where the two features sat. Login only benefited the marketplace - and that meant it was really hard work to get buyers to do it. Sponsored listings benefited the marketplace with extra revenue, and some sellers got more sales, and it got a bit more traction; ultimately it was bad for the buyers and in the long term that made it bad for everyone.

What is interesting is when there is a feature that is in everyone’s interests and sits in the overlap in the middle of the diagram. An example is improving the relevance of search results - that helps buyers find things more quickly, they then purchase more which is good for sellers and the marketplace.

When considering a feature, think about who the stakeholders are for your service, draw the diagram and plot where your feature sits in terms of their interests (you might also want to plot out where some previous successes and flops would go). If a feature doesn’t sit in any overlap, it is going to be an uphill struggle to get adoption – if it sits in the overlap of all the groups, it is much more likely to be successful.

If something doesn’t initially land in the overlap of all the groups, you can often nudge it there. The marketplace wanted more logins - if logging in allowed delivery time to be tracked and changed then buyers would log in more. At that point, buyers benefit (getting the delivery when they can receive it), the seller benefits (fewer redeliveries) and the marketplace gets what it wanted: logged in buyers (plus happier buyers and sellers).

As a bonus, we can think about the metrics the platform aims to optimise and where those sit on the diagram (see diagram below). Marketplace revenue is important for survival, but increases in revenue rarely reflect benefit for other stakeholders (e.g. increased commission hurts sellers, sponsored results hurt buyers). A metric such as conversion rate (the proportion of people who search for a product and then buy one) reflects everyone’s interests – it means that buyers find what they want, sellers make sales, and the marketplace gets commission. Similarly, the value of sales through the platform reflects all interests – if people are buying what they want, sellers are getting paid, and the marketplace gets commission (although arguably there is a tension in terms of price for buyers and sellers). When you are considering metrics that reflect your business, again, plot where they would sit on the diagram of stakeholders.

While the marketplace is an artificial example, I have seen these issues occur again and again – including in online retail, consumer goods, and B2B services. Each time it helps to draw out the groups of stakeholders and even if the initial motivation is to help the business, finding ways that it benefits everyone involved with the business.

What about the National Digital Platform?

In health, there can be similar product failures to those experienced by the marketplace. This time, the narrative is – “we are trying to improve health, if people did X their health would improve, we should produce a product to help people to do X.”; followed by: “why don’t people use our product?” and the inevitable attempts to force adoption “Maybe we need to market it harder or get clinicians to recommend it.” The challenge being that the product is not helping the person do a job they want to do (whether they are a clinician or a patient) and instead it is focused on the system’s needs, to the exclusion of other needs.

Applying the approach described above, you could start by thinking about three groups in public health (although there are more):

  • Care recipients and their representatives (e.g. parents, guardians)
  • Care Professionals – e.g. social care workers, nurses, doctors
  • Care Organisations – e.g. NHS, councils, social care providers

Just as in the example above, features will be most likely to succeed if they benefit all the groups using the National Digital Platform. For example – while a feature that allowed citizens to view their own medical records initially sounds like a sensible thing to do, I suspect that it would see little use and not improve health. That is because, beyond the initial novelty of logging in to view their own records, citizens are unlikely to keep logging in to see them, and they are unlikely to take any action as a result of just reading a record. Health will only improve if actions are taken, and that is only likely to happen if the record is accompanied by information that helps people take action.

Similarly, just giving access to records does not help a doctor if it doesn’t change the actions of a patient and it could potentially create confusion and unnecessary concern if the notes are not straightforward to understand. Therefore, for such a feature to be successful we need to consider what would need to happen to make it in the interest of all the groups using the platform – and therefore improve wellbeing. That means it needs to do something that helps people make changes.

Another example is allowing patients to upload data from a home blood pressure monitor – they are only likely to do that if there is a clear benefit for them. That benefit could be avoiding a thirty-mile round trip to a GP practice for a reading or taking a morning off work to go to a surgery. A GP is only going to benefit if the measurements are easily accessible, and quick to interpret. A social care provider would benefit if it didn’t have to arrange for an accessible taxi for two hours and could use those funds for other services. If the blood pressure upload feature is going to be used, it needs to be implemented in a manner that delivers benefit to all the constituencies of the platform – patient, GP, and social care provider.

As for metrics that reflect that unified benefit for the platform, I am uncertain what the metrics here should be, ultimately this is about improving quality of care and better health – that sits in the middle in terms of what all parties want to see. We need to do more work in terms of how we measure that and in a way that enables us to measure our specific impact on it.