How to validate your idea or feature before you code

Validate your learning

I recently received an email from a potential customer who has an idea for a business and required funding. Though the money was not a large amount, I decided to take a leaner approach to the situation to try and understand how we can cut away wastage and provide a cheaper, more effective solution. I suggested that she starts small and with what she had, but the person wanted to go big or go home.

Many people don’t believe in the power of learning, yet disruptive innovation is filled with learning being valued more than money:

• Airbnb had to learn which part of their system is not complimenting the stay-at-home experience. This included discovering handing over cash, relooking the quality of the venue photos and targeting people outside of special events
• Netflix learnt about customer behaviour and building up a relational database of which shows would be more appealing after watching certain shows

Where did validated learning come from?

Eric Ries first coined the term validated learning in 2011.

The idea is that you test whether a feature, product or enhancement will add the value that you think it will.

This can be achieved by firstly realising that we are making assumptions about how much value a feature will add to a system. Secondly, we need to be able to measure the effect that it has on the user – will this feature move the business forward? Are we able to validate this without writing the feature?

The validation process

Though this is not a one size fits all, the typical steps in validated learning can be seen as:

1. Specify a goal – add details about what you want to learn
2. Specify a metric that represents the goal – define how you will be able to achieve it. Include details such as software or analytics trackers to be used, methods etc.
3. Act to achieve the goal – do what needs to get done to get the metrics in
4. Analyze the metric – do the results align with the initial goal? Will this give you the result you need?
5. Improve and try again – what did you learn? How are you using that?

Examples of validated learning in software development

Validated learning makes sense in small businesses and entrepreneurship, as one can test it directly with the customer.

In most companies, the developer never sees anyone other than the immediate IT team, it can be challenging to change the mindset from a cost centre to a team of enablers.

Yet, with a bit of creativity, the developers or business analysts are able to work through the validated learning process without interrupting business as usual.

Some of these include gathering analytics through Azure Performance Monitoring, Google Analytics, HotJar or similar software. This can be combined with A/B testing and other methods to see if a button is clicked on or not.

The case of the vital feature

An accounting online software as a service (SAAS) solution received feedback from clients that they require a fairly large feature. They decided to add a menu item with the name of the feature. When the menu item as clicked, it navigated to a page saying “Coming soon”.

Using Google Analytics, they realised that no one clicked on the page – it wasn’t something that added value to their clients.

The large app requirement

A client required a complicated mobile app. To simplify the requirements, we analysed the features that we believed to be most important. We built this into an MVP to pitch to clients.

The client was able to demo the app with the basic features and receive feedback about what the users expected. The developers were also able to get feedback via analytics software to understand which features were clicked on most and which were not seen as important.

Conclusion

Validating your feature as valid, value-adding and worth the effort means that the developer codes knowing that his code will make a difference.

Creating MVPs, prototypes, using analytics to track user activity can give us insight into what is actually important.

Simply be effective.