Business hunches are so 1977. Proper data's a lot more grown-up

Fighting the instinct vs analytics wars

Binary data/big data conceptual illustration. Illustration via Shutterstock

Sponsored In business, should you trust your gut, or go by the numbers? Business mythology and Hollywood favour the gut. Both cite stories about maverick geniuses that ignored everyone’s advice and won big.

Sure, it happens, but maybe not as much as you’d think. We’re affected by survivorship bias, concentrating on those that got it right, rather than those who bet against the data and got it wrong. Gut instinct only goes so far. Just ask Homer Simpson about pumpkins.

We shouldn’t ignore the seasoned old board member who has been around several blocks. Their hunches are often surfacing years of experience. But even with those inputs, evidence-based decision making is crucial.

Hunch-based decisions arose in the first place because the data to do anything else just wasn’t there. These days, we’re swimming in an ocean of the stuff, and don’t need to rely on guesswork as much as we used to. Using business analytics to optimize your performance can give you a competitive edge.

Managers can mine this information to ask simple questions: How are we doing? Why aren’t we doing better? What are the key areas for improvement?

They can also use analytics tools to test assumptions using what if scenarios. In the old days, you did this using an Excel pivot table and some directly related data. In modern analytics environments, you can adjust one parameter and see what that stomping butterfly does to key performance indicators half a world away.

This all sounds peachy, but in reality, not enough people are relying on their data. Back in 2014, the Economist Intelligence Unit surveyed 174 executives. Fifty-seven per cent of them would reanalyse data that contracted their intuition, it found.

Why aren’t they trusting their numbers? It may come down to the lack of a clear Business Analytics and Optimization (BAO) strategy. You need to define things before you can crunch numbers effectively: the right data, and the right tools to grind it with.

The barriers to analytics

Getting the right data isn’t as easy as you might think. McKinsey bigwigs point out that you first have to figure out what data you want to use. Collating and digesting that internal data can be enough of a challenge.

Firstly, you have to find it. That can be tricky in many modern companies, which have built up their data architectures over time.

You’ll find a puddle of supply chain performance data over here, and a few gobs of regional sales information over there. The chances are that they’ll be in different systems that don’t talk to each other, in different formats. Different people will probably own them, both of whom are confirmed server huggers and refuse to set their information free. If you can wrest their data from them, you must work out which of it is useful.

Then you have to repeat the process for external data sets, which may also be relevant. Environmental factors like weather, average rental prices, traffic congestion or interest rates might all affect your business. Understanding which data to consume takes domain expertise and vision.

Finding the right skills

Once you've collected the data you need and put it into an appropriate format, the next problem is developing or sourcing the smarts to make sense of it.

You’ll need a few components to build out an analytics team that can deliver actionable results. The first set of skills lies on the technology side. You’ll be dealing with new kinds of data, and more of it. You’ll probably be breaking up queries and processing then concurrently to deliver faster results. That requires technology skills to manage networks and clustering, in addition to software tools like Hadoop.

Then, you have to think about the data science. This is a discipline of its own, and you’ll need people well versed in the mathematical techniques to mash the numbers together. Depending on your business area, you may find that these vanilla data science skills need some guidance. Finding data scientists with your specific domain expertise may be a challenge, in which case you’ll have to find an internal businessperson who can talk their language.

The other big challenge, according to McKinsey, is getting companies to act on the data. Companies often amass insights from the data but then refuse to do anything with them, it seems.

This is more of a cultural problem than a technical one. The best analytics program will fail if the company running it doesn’t plug it into the business, providing an avenue to digest the information it’s providing and act on it.

If no one owns the findings from the data, then it’s unlikely to result in real business change. You’ll have some pretty dashboards, though.

It isn’t just inertia or poor company structure that gets in the way of evidence-based decision making. Sometimes, people will fight to maintain the status quo. Almost half of the executives in that EIU survey (44%) said that company politics trump evidence when it comes to making decisions.

How to make a friend of your data

So, how can you stop flying by the seat of your pants and begin making some evidence-based decisions? Begin with baby steps.

Target a single business process that you think could benefit the most from a data-driven approach. Find the data that affects it and use this project to begin building a data warehouse or data lake if you don’t already have one.

Look for quick wins. Begin the project with pointed questions that you are trying to answer, and make sure that the findings can lead to change in the business. Hold someone responsible for taking the data and actioning it.

Just as you base your business decisions on data, so you must base the success of your analytics program on hard numbers. Look for ROI by measuring the results of your data-driven decisions.

In six months’ time, you should be able to tell the managing director that you have increased your Net Promoter Score by 20% on average thanks to evidence-based changes in your customer support, say.

From there, expand to other use cases, gathering data sources and breaking down silos along the way. Be sure to do it with one eye on the company culture, though. It’s important to get buy-in from the mid-level managers who own the data, and from the senior execs who can actually do something with the results.

Sweeten this deal by offering to add value for the stakeholders. Bill in field engineering may be more inclined to give you his data if you show him a way to shave percentage points from his fleet petrol budget and slash paperwork back at the office.

Encourage different departments to begin working with the analytics team to find new efficiencies in the business. One option may be an ‘analytics challenge’ that brings together employees who wouldn’t normally interact. See if they can collaborate on analytics projects. We hear data bake-offs and hackathons are all the rage these days.

Proper business analytics optimization isn’t something you can do overnight. It involves technical change, but unless you include the cultural part too, your business analytics and optimization program may fall flat.

A gradual approach is a smart way to approach things, though. How do you drink a data lake? One glassful at a time.

Biting the hand that feeds IT © 1998–2018