One look at the Gartner hype cycle shows there's no shortage of available technologies. Many businesses have rolled up their sleeves to experiment with emerging technologies. The problem is, most lack an understanding of use cases that deliver real business value. As a result, the majority of experiments fail to deliver actual value to the business.
Knowing where the shoe pinches: why pilots tend to fail
Here's what happens in many organizations. After discussing an emerging technology, management launches an initiative to leverage its possibilities. Suppose it's AI they wish to explore. In that case, an AI (data science) team is created, and it's given lots of freedom to experiment. The rationale behind it: learning will create value in itself, and the team will figure out a way to make the most of the available data. Typically, people are in good spirits at the start of the experiment, and expectations are high.
This results in the team focusing on the most advanced (or, fun) technologies in an attempt to master them. The team decides to operate in those areas where data or business access is easiest. Working products or pilots are created, promising results are reported. But at some point, reality kicks in: once a pilot needs to be accepted by the business, it fails miserably. Why? Because focusing on technology and ease of implementation hardly ever results in products that meet urgent business needs. The business doesn’t get involved in the development process – basically, the product is created in the technology department's ivory tower. And by the time it's finished, the business won't prioritize its incorporation because it’s not all that useful.
The results: money wasted, no added value for the business, and disillusioned technology experts who are inclined to leave your company (after you've gone out of your way to attract and hire them).
A different approach: from promising pilot to actual product
You can go about it in a much more constructive way. How? We'll break it down for you.
1. Align your experiment with business needs
Start your experiment with a quick scan involving business and technology experts. Identify the best opportunities for experiments, considering:
• Business value
• Ease of implementation – maturity of technology, complexity of the solution, and availability of data, among other things
Based on this quick scan, create a list of prioritized business issues the experiment team can work on. Serving as a starting point, this list should be updated through a 'learning-by-doing' approach.
2. Make your solution work in the real business environment
Whenever you pick a new product from the list to develop, make sure you build it in such a way that it will work in the real business environment:
• 'Definition of done:' clearly describe what it means for a product to be done, which is when the business uses it in ‘business as usual.’ Typically, this goes way beyond running a prototype based on historical data in an experimental IT environment.
• Business value: roughly define business value and check it with the business.
• Involve business users during product development.
By ensuring the business is on board from the start, you'll tackle issues with real business value. This doesn't mean all experiments will be successful – some might be too complex, and you might occasionally misjudge business value. But most finished prototypes should deliver value. Moreover, your data scientist will know their work does, in fact, matter – a unique selling point which makes it more likely they'll keep working at your company.
Do you like this approach? And could you use some support in your endeavors? Please don't hesitate to contact us.