(I took his advice by ignoring his advice)
It’s happening again. I’m focusing on metrics and losing sight of the big picture. The more I analyse the individual elements, the more I lose sight of the whole. I’m feeling stuck. It’s time to remember the fishtank.
And as I ponder the Lesson Of the Fish Tank, I see wearable trackers are getting a bad rep for not helping their wearers lose weight. The fish tank again!
The molecular biology graduate in me has always hated the phrase “gut instincts” when it appears next to “data” or “metrics” because the two have always been presented as if they balance each other, usually by marketers. You rarely hear Nate Silver talk about his gut feelings.
When I was a part of Business Information, I could never understand why there was so much scepticism against data from people who’s best interests would be best served by it. Data – my experience had taught me – is absolute and non-judgmental. Data is an un-opinionated compass in many important journeys.
Until the Fishtank. I opened the hood of my 200 litre tropical tank one morning and immediately knew the water was off. This sometimes happened if Thames Water changed the amounts of chemicals in the water and I was too aggressive with my water changes that week. I did a quickie water change and threw some ammonia blocker just to be safe before driving to work. I came home that evening and tested my water to see how much water I might need to change. The ammonia – a chemical highly toxic to fish- was too high. I did a 70% water change, and added water protectors to minimise gill damage from the previous readings- at least I hoped. In the morning I tested again. Despite two water changes, the ammonia was now at a level so high nothing I would be able to do could save my fish from the internal damage. Tears were streaming down my face, knowing I had failed them. Until a teeny, far away voice inside my head said, “but they don’t look sick.”
I stopped crying and really looked at them. They didn’t look sick at all. They looked…pretty damn happy. Beautiful smooth scales. Healthy gills and colourings. Active and playful. I sniffed the water. No ammonia smell.
It turns out the particular brand of ammonia block I used as a preventative gives false positives on water quality tests. Because the test kit and the additive were by the same manufacturer -and because there was no warning on the bottle or kit- it didn’t occur to me they might cross-react. I changed the water again, this time leaving out the ammonia blocker and re-tested. The water was as perfect as it should have been after the first water change.
Data is absolute, unjudgmental and impartial, but data cannot self-aggregate nor self-analyse without a human made model, based on human-made assumptions. Data can be susceptible to factors we aren’t even aware are at play. Our interpretations susceptible to wishful or to fearful thinking.
The problem with the fitness trackers isn’t that step counts aren’t useful for losing weight. It’s that only looking at step and fitness targets made wearers vastly over-estimate how much they can eat after the log big numbers. Such trackers work best for weight loss with integrated with some kind of meal journal, macro counter or calorie tracker. This isn’t restricted to trackers, either. The results of exercise alone, as opposed to in combination with diet gives the exact same results.
By focusing too much one metric, we lose sight of the big picture.
That’s why if you start looking, you’ll see many examples that seem to support “gut” vs “data“, but are really “tunnel vision” vs “big picture” thinking, such as:
- Shrödengers Content SEO
This is a story about one site architecture and backend running two sister brand websites using a variety of backend, server side, and front-end magic to swap out the domains, branding and content as necessary. The main site was older. The new sister site, newly acquired. Both had one master product list, with the majority of products filtered on the siter site, meaning they wouldn’t appear by browsing or searching, but you could still find them if you entered their title in the URL explicitly.
The sister site marketing team did not do content marketing. All new content was purely advertising the release new products. The master site marketers had weekly targets for blog posts. I had noticed that the quality of the content seemed very thin bordering to embarrassing at times, but it wasn’t until the main site was ranking below it’s blog-free, filtered, sister site for ALL content that we could fully appreciate just how bad poor content can be to a site’s SEO.
The very things the marketers were doing in the belief it would make their products become more visible and search-friendly- churning out shallow articles with repeating keywords- we’re getting us de-ranked. Worse, because they had to hit their weekly content targets, they didn’t have time to research and write articles that spoke to their audience.
Target numbers aren’t a metric, but team managers do like giving them to their staffers or freelancers thinking they will boost a metric down a line- such as website traffic, email subscriptions or sales. Targets are good places to start but without keeping an eye on how they are being met, they lose all meaning. For the teams above, literally doing nothing worked better than hitting their marketing targets.
2. Twitter: Metrics over User Experience
As we all know by now, Disney has turned down Twitter for purchase because they fear the way the platform has fostered trolls will hurt Disney’s image. But Twitter started off as a great platform. It started to go wrong once it started optimising interactions around the idea of boosting “engagement”. Applied without context (e.g boosting engagement within your social network, or among peers), trolling can be seen as a roaring success. People are watching hashtags scroll by and replying to multiple strangers all day long and starting fires of discussions in multiple timelines- all signals that tick the boxes of “engagement”. Twitter is a classic example of how designing for a metric over the user experience has lead to many users abandoning it, and not feeling safe to share the same fun, personal thoughts that made it so great in the beginning.
3. The ‘too trivial’ metric
When I started running, I was using endurance and speed as metrics. First, I improved my endurance then I wanted to improve my speed. I tracked my running vs walking interval times and how much distance I covered in each. What I wasn’t measuring was how I felt during and after a run. I felt like my lungs were on fire, and after running my arms, legs and lungs felt like they were on a slow burn all day long. It would be a year before I realised this wasn’t how other people felt. By focusing on endurance and speed (what everyone shared online as “important”), I was ignoring some serious red flags that were blocking my progress. On the other hand, reviewing my inability to reach my goals in my expected timeframe by visiting a doctor got me a nifty flip-top inhaler and a series of physiotherapy appointments. Now, I measure the above plus my heart rate during a run and how I feel after.
The problem with data isn’t the data. The problem is lacking sufficient context in which to understand it.
Data needs context.
So what’s my take home message?
- Just do good work. Focus on the end user or consumer, not optimising for stats, but have a data and business analyst review and ask meaningful questions to understand both what’s important to every team, what’s influencing their internal data, and to provide some holistic, big-picture feedback on what direction to take.
- If you’re not hitting your targets you expected, get an external opinion. Maybe you set a target that makes sense for a similar business model, but not for yours. Maybe there’s something else you need to be focusing on before you can even start worrying about that metric. Before someone starts worrying about content SEO, first they need good, human- readable content. Before they can do that, they need to understand their audience.
- Ignore what other people are doing: People love dishing out simple solutions: You need to be on social networks, you need to use this new technology, that email service provider is no good, change CMS. Back to my running project: A runner friend recently told me that caring about my heart rate range wasn’t important. He then went on to tell me that he didn’t care what his other friends did to improve, as long as what he did worked fine. So I took his advice by ignoring his advice. I’ll run my first 10K race in two weeks. Ignore what competitors and peers are doing and focus on what brings meaningful results for your end users.
So, what about you? Have you had any situations where chasing a metric was driving you off-course, or compromising your vision? Have you seen anyone look at their mixpanel and leap to a strategy you had doubts about to improve something? Drop me a message or email below. I love a good story :)