I've been learning about trade-off curves recently and they seem pretty useful for learning and making decisions when designing products.
A trade-off curve is a graph that explains what happens to the performance of something when you change something else. For example, the graph below from rimmkaufman.com explains that the effectiveness of advertising through Google increases the more you spend but there comes a point where effectiveness begins to decrease. Given this picture, a marketing manager could cap their spending at a sensible place and put the remainder of their budget somewhere more effective.
A trade-off curve seems like a powerful tool for making good decisions, in certain situations. Any time where we're doing something that we expect is going to create value is likely to have a pattern of diminishing returns. A smart thing to do is to work out at what point this might be. There's no need to make a design that pushes harder than it needs to. Let's push the design just as hard as it needs to be pushed, but no further. Any more pushing and you risk creating waste.
Another example of a trade-off curve is the Phillips Curve
, which shows us that unemployment and inflation are highly negatively correlated.
Designing good experiences is often about trade-offs. When you change one thing, you change something else. There are always a number of different features vying for attention on a page, there are adjectives we use to describe the functionality of a design and there are purposes to every feature we design. Everything we communicate in a design has a negative or positive effect on customer value.
If you can use an adjective to describe something, then you can probably measure it somehow. And if you can measure it then you can probably make a trade-off curve out of it. And if you can make a trade-off curve then you can make some good decisions.
As far as I can see, this could be an incredibly powerful way of optimising the design of an existing product. It seems like the kind of thing you'd do if you had a really clear set of objectives and you're already engrossed in a constant state of improvement in a quantitative way, for example, using A/B testing. It seems like a good next step if you're already enjoying the benefits of A/B testing at scale.
But the thing (I think) that excites me about them is the idea that you might be able to create them for products that don't even exist yet by prototyping with smaller numbers of people. So, could we understand the trade-off between commercial messages v.s. non-commercial messages? deep v.s. wide product range presentation? elaborate language v.s. simple language etc. It feels like this could potentially create much more re-usable knowledge. It isn't just about one idea working and another idea not working, it's much more nuanced than that, it's about the degrees to which a hypothesis might be true. The terms 'validation' troubles me because it suggests a state of either validation or no-validation, on or off, true or false, which is dangerous. The world is never black and white, it's hundreds of shades of grey with black on one side and white on the other.
It seems like a somewhat analytical tool but if you're getting good results with A/B testing then this might be an interesting experiment to try.
I think that what I'm looking for is a trade-off curve that shows the effectiveness of trade-off curves against the difficulty of capturing data to make trade-off curves.
If you're interested, take a look at this Tribute to Dr. Allen Ward
. He was one of the biggest exponents of this approach. He attributed a considerable amount of the success of the lean product development kata at Toyota to this technique.
I'd be fascinated to know if you've had any success with these in product development.