Article:

Success Always Starts With Failure Five Techniques for Prospering Amid Uncertainty

Tuesday 31 March 2020
By Looking at Data and Forecasting, Can We Accurately Predict Outcomes? Or Do We Need to Adopt a Practice of Preparing for Multiple Outcomes Simultaneously?
flower in sand

Forecasting is a rough game. The consultancy Oliver Wyman predicted up to 75,000 job losses in the City in the wake of the Brexit vote. So far, that figure looks to be out by at least 90 per cent. Equally, who foresaw the problems the Irish border would cause in the Withdrawal Agreement negotiations? Or that the UK’s expulsion from the Galileo satellite network would be a thing? Again and again, the forecasters have been blind-sided.

To understand why, the economist Philip Tetlock looked at 20 years of forecasts from 1984 to 2004. His conclusion is one of the most famous in the profession: “The average expert was roughly as accurate as a dart-throwing chimpanzee”.

The thing is, managers need to plan and prepare for the aftershocks of Brexit. Which raises the question: how can we plan when advice is unreliable, and the experts rival chimps when it comes to their accuracy? One answer, put forward by David Weinberger in Everyday Chaos (shortlisted for this year’s CMI's Management Book of the Year), is to adopt “strategies of unanticipation” instead.

Rather than relying on predictions, unanticipation means being ready for multiple outcomes. It requires cooperation with third parties, iterating ideas as new information comes to light, and staying agile. As the UK moves closer to the unknown and unknowable post-Brexit world, here are five proven strategies based on the logic of unanticipation. Each one could help your organisation cope with the future, no matter how wild it proves to be.

1. Quantify Uncertainty With Probability Curves

Will it rain tomorrow? Look at the Met Office forecast and you’ll get a probability expressed as a single number, something like 68 per cent. But this is pretty limited information, a binary yay or nay.

When it comes to understanding the full spectrum of potential outcomes, experts use a “probability curve”. Thus, there might be a two per cent chance of 1mm of rain tomorrow, an eight per cent chance of 2mm, a 15 per cent chance of 3mm and so on – usually in the shape of a bell curve.

Probability curves are a part of daily life for supermarkets. How many cabbages will be sold in a day? The forecast is expressed as a probability curve. Managers can then look at the most likely number of cabbages sold and order accordingly.

Interpreting the numbers isn’t a science. “There is a strategic decision to be made at a supermarket,” says Professor Michael Feindt, founder of Blue Yonder, a consultancy that crunches numbers for a variety of European supermarkets. “Are you OK with running out of a product and showing empty shelves? This might be fine for a cheaper brand. But what if you are a luxury supermarket that always has goods in stock?” The curve will help managers quantify the risks of over- or under-stocking.

The probability curve can be used in conjunction with an individual cost function curve. This plots the costs of over- or under-stocking. If there is over-stocking, this equates to wasted produce plus tied-up capital. The greater the error, the greater the cost. Under-stocking implies costs in the form of missed sales. Again, the greater the deviation, the greater the cost. If there is neither and the final product is sold on the bell of closing time, the cost is zero and you’ve achieved perfection!

Curiously, probability curves are underused in management. “Honestly, I don’t know why,” says Feindt. “In finance, it is common, but in other businesses it is simply not done. The mathematics may be too hard. People think they can use their gut instinct better.”

2. Open Your Platform

When Steve Jobs developed the iPhone, he knew it was impossible to forecast the needs of users. What would they want? Games? Spreadsheets? An alarm clock that wakes users up with the hits from the Top 100? The solution was the App Store. This way, Apple did not need to focus on the future. It outsourced creativity to third parties, who’d invent anything and everything users might want.

Today, open platforms are flourishing. The workplace chat application Slack offers more than 1,500 apps, from surveys to work productivity trackers. The apps relieve Slack of responsibility for predicting the future.

Customer relationship management platform Salesforce runs a huge App Exchange, created after founder Marc Benioff discussed the concept with Jobs. It now has more than 5,000 applications, offering Salesforce’s users a conveyor belt of innovation.

Platforms often evolve in ways the creators never imagined. Although GitHub was originally built for software developers, it has turned out to be a platform for collaboration in general. Today, people use the platform to share knitting patterns and cocktail recipes – developing them freely just as they would with software. “GitHub began by allowing developers not to anticipate how their code could be used,” says David Weinberger in Everyday Chaos. “By refusing to anticipate even what type of projects might benefit from unanticipated upstream sharing, GitHub’s utility has reached far beyond the world of software development.”

Even the government is embracing open platforms, making bus data publicly available via an API, for example. Using this, data scientist Tom Forth has developed a ‘Bus Tracker’ app to record and report performance and, in turn, improve service quality. While Forth’s approach is now widely used, it was pooh-poohed to begin with. “I had about half a dozen people assure me that they were already tracking all the buses, they already understood congestion, that my techniques were not fit for purpose. And another half a dozen said it was useless anyway and their models were much better.” Here, an open platform meant that a maverick could advance a vital service, despite the cluelessness of the officials in charge.

Navigate Uncertainty With a/b Testing

When forecasting customer behaviour proves impossible, the best strategy can be to give up. It’s time to stop guessing and start testing.

A/B testing allows organisations to compare a huge variety of different options to find the one that works best.

For example, when Kia Cars wanted to find out the best layout for its website, rather than trusting designers to get it right, Kia ran A/B tests to optimise every aspect. The digital marketing team tested five areas:

  • Colour of the buttons;
  • Shape of the buttons;
  • Size of the buttons;
  • Order of the buttons; and
  • Duration that each vehicle and promotion panel was shown.

Kia ran A/B tests over five weeks to see which variations worked best. An A/B testing consultancy called Evolv automated the experimentation, using artificial intelligence to manage 2,592 possible combinations. As Tyler Foster of Evolv explains, “gut feeling is unreliable and subjective. A/B testing gives you objective evidence about what works.”

All elements of a website can be tested: checkout structure, keyword advertising, responsiveness. “You can keep optimising,” says Foster. “I’ve never known a company run out of things to test.”

Philosophically, the strength of A/B testing is the ability to discover the best course of action, while not knowing why. Barack Obama's first presidential campaign tested three sign-up buttons on his website. “Learn More” earned more clicks than “Join Us Now” or “Sign Up”. Why? The answer is irrelevant. A/B testing illuminated the optimum strategy, without necessarily revealing the logic of user preferences.

This method isn’t entirely blind. You need a hunch to direct the parameters. But A/B testing can illuminate the path ahead when forecasting falls short.

3. Tim Harford Says Adapt!

As the Financial Times’ Undercover Economist, Tim Harford is renowned for his alternative takes on everyday phenomena. When it comes to coping with uncertainty, he realised that a totally new approach was needed. His solution was published in his bestseller Adapt.

His starting point is that forecasts are terrible. “The world, given our current state of knowledge, human affairs, politics, economics, is not forecastable at all,” he says. So what should we do? “You need to plan ahead, and think about what could go wrong. Think of the worst case scenario, and then work out in a detailed, quantitative way how you might work to prevent it.” He advises creating multiple plans, each tailored to a different outcome. That way, no matter what happens, you are prepared.

Over time, the key is iteration. The subtitle of his book is Why success always starts with failure. The first attempt at anything is likely to be pretty terrible, so he advises you stop trying to perfect a product, launch it early, and then adapt. This way, it doesn’t matter if your forecasts are wrong. You’ll find out what customers think early, and have the time and budget to respond and relaunch.

Listening to that feedback will be vital. A curious by-product of erroneous forecasts is the tenacity of their followers. Harford cites a judge who rejected DNA evidence in a trial, owing to his prior belief that the defendent was guilty. Changing our minds is hard. A commitment to changing tack multiple times during the life of a project can help us avoid this pitfall.

Above all, learn by trial and error. This isn’t easy, Harford admits. “Experimenting can be a frightening process. We are constantly making mistakes, not knowing whether we are on the right lines,” he says.

However, Harford is under no illusions. Most people will ignore his advice, he explains. “People are addicted to forecasts.”

4. Abiy Ahmed’s Ethiopian Strategy

The recent history of Ethiopia is one of trauma and conflict. Relations with neighbours are fractious. The 12-year war with Eritrea, ending in 2000, claimed 100,000 lives. There is still sporadic warfare with militias in Somalia, and internal ethnic violence has displaced 2.9 million people. Ethnic federalists demand self-governance, while unionists demand rule from the centre. Journalists claim harassment and persecution. All told, the nation was regarded as ungovernable.

In April 2018, a new prime minister took office. Abiy Ahmed understood that his role was incalculably complex. In an effort to break with the past, he coined the term “Medemer”. He explains: “Medemer, an Amharic word, signifies synergy, convergence and teamwork for a common destiny… I like to think of Medemer as a social compact for Ethiopians to build a just, egalitarian, democratic and humane society by pulling together our resources for our collective survival and prosperity.” In a 280-page book, he spelled out his philosophy of Medemer, essentially asking Ethiopians to forget the past and take control of their destiny by pulling together. Incredibly, it seems to have worked.

Ahmed managed to lower ethnic tensions by apologising for past mistakes. He released political prisoners. He announced that Ethiopia would “no longer be a jail for journalists”. He asked for rival groups to focus on “civic nationalism”.

In 2019, Ahmed won the Nobel Peace Prize for his work. But what are the lessons? Medemer is many things, including the belief that trust must be at the heart of any project to shape the future. A business without Medemer would be rigid, as workers wouldn’t co-operate. It would obsess over old grievances. Ideas would be locked away, for fear of ridicule. Niall FitzGerald, former chairman of Unilever, put it this way: “You can have all the facts and figures, all the supporting evidence, all the endorsement that you want, but if you don’t command trust, you won’t get anywhere.”

The crux of Ahmed’s Nobel Prize speech was this line: “Together, we must neutralise the toxin of hatred by creating a civic culture of consensus-based democracy, inclusivity, civility, and tolerance based on Medemer principles.”

Trust is an old-fashioned concept in business. Without it, the future will be unmanageable, irrespective of what it holds. A philosophy of Medemer may be what your company needs. A strong sense of shared purpose can help an organisation cope with the future, no matter how uncertain it seems.

This article was published in our most recent edition of CMI’s magazine – if you’re not a CMI member, check out the many benefits we can offer you.

Charles Orton-Jones

Charles Orton-Jones