“How do we collect better data to build better products? “
This may seem like a simple enough question, because data in itself is a base form. Data is either valid or invalid, but how do we then optimise to collect, and make use of data?
My idea of better data in the context of product design and marketing is based on experience trying to build a technology business in Nigeria. In developed markets, data has been collected for decades at least, which allows creators build better products. Where data doesn’t exist, it’s arguably easier to collect than in Nigeria.
An example is something as simple street/house numbers. These are things taken for granted in the developed world, and even in parts of the developing world. But in Nigeria, house numbering is pure chaos. Yet, this is an integral part of the logistics and fulfilment system. While Amazon is trying to launch drone deliveries in America, depending largely on existing numbering and GPS systems, Konga literally had to build their own postal service. (What would be interesting is Konga’s data on how many times it takes their delivery guys to identify a home to deliver to, and how they do that.)
This inherent chaos is our system is not an accident. If I may borrow the word, anyhowness is a real thing. But we need to build products and services, irrespective of this chaos. And we need data to make these products/services work well, because without data, our products and services cannot operate optimally. So how do we get better data?
I was recently speaking with a technology team at a top 4 bank, helping redo their service design for Internet banking, and we were stuck on whether certain fields on their interface should be displayed at all. Bankers are fun people to work with – they had filled their software with fine banking language, like ‘liens’, ‘facility’ etc, and we’re unwilling to make changes.
So I asked: why are we making these changes?
They answered: because we want our online banking services to appeal to younger customers.
Me: How many young customers understand these big words?
Banker: They do!
Me: I don’t. And I’m not very stupid.
Banker: Well…you should.
Me: Ok, but more importantly, how many of your users, statistically click on those sections of your website, or use the service?
It emerged that it was less than 5 percent, maybe 2%. Less than 5% of users accessed a feature, yet we were fighting over it being front centre.
This is the problem. It’s a bigger problem than not having data. Understanding the place of data in the design process is the bigger problem.
The era of assuming when you build, customers will use it just because you have built it never existed. Only Apple gets away with that, and that’s because Apple understands great design better than all of us combined. Since we’re not Apple, we must behave like we’re not Apple.
A website design for example, is more than just wireframe and new CSS tricks. It’s not carousels and new HTML5 widgets. Designers must see themselves as the forerunners of intelligent service design. This means they can’t work alone.
Design is a destination which describes the entire customer journey in one location. Hence, if the design doesn’t not understand and properly capture the customer journey, and how the customer interacts with the brand, then the design product is a failure.
We must remember though, that collecting data for product design is not rocket science. Especially for people who already own their platforms, the data is there already. You probably have already collected enough, or could with minor adjustments. The challenge is in asking the right questions, and being honest to self. “What am I currently doing right? Where am I failing, and how can I improve?”
Over the last couple of months, we’ve designed various data collection models. The most successful ones have been those disguised as products.
It is very interesting that if we look at things differently, we could completely change the way we see data collection (and their applications to design), mostly from sources which are not very obvious to the user. How do viewing patterns on iRokoTV help Nollywood create better films (which score in the market)? This was an unstated, but critical observation, casually dropped in a recent blog by Jason Njoku.
The gaming industry is an example of where data is currently being harvested, with meaningful real life applications. The new Kim Kardashian game could deliver valuable user behaviour data to hotels, TV and entertainment producers, Candy Crush could help study addiction and compulsive behaviour, and law enforcement could do very interesting things with Grand Theft Auto data. It could be argued in gaming that building a great product is not an end in itself, but the beginning.
This is indeed the experience we have had over the last couple of years at Anakle, but the last 15 months have been the most interesting. While I was out being notorious for building the Bride Price app (again, I didn’t build it, Ofure’s team did), our servers were quietly collecting various kinds of data (which was one of the reasons the app was designed in the first place). We were lucky to have a few million interactions to play with, and when the various data points were reviewed, the emerging patterns provided very interesting feedback.
For example, returning female users tended to score higher marks in the quiz. Does that tell anything? Imagine that a user had scored N200,000 in the first taking of the quiz, then returned to take it – the pattern was that most returning users ended up with higher scores. What were the most popular skin tones in a given location? What image of themselves did users see in their minds? But these are the more obvious pieces of information users left behind, without filling any forms.
Most of the data we collected is currently helping our team design a more serious, real life services, and we’re approaching the product design with a lot more confidence than we did months before.
For our advertising team servicing clients in various industries, it is an advantage that we are able to predict user behaviour. While hypotheses are good, real data allows us to walk into battle with a more significant amount of confidence.
Users don’t have the time to fill our surveys, and most times, when people actively fill surveys, they are more inclined to put their best foot forward. I believe we would find more valuable data by hacking passive user activity to collect real life data. Of course nothing I am saying here is new. Silicon Valley perfected this model decades ago. But in Africa, we need to get started too, as a mainstream model.