What can machine learning teach us about optimising customer experience?

Brain machine
Share this content

Bots have been called the most overhyped and unimpressive technology of 2016. Facebook’s efforts have been branded spammy and Microsoft’s own attempt had to be shut down after adopting some questionable views on Twitter.

In consumers’ eyes we’re a long way from becoming best friends with robots. But behind the scenes, artificial intelligence and machine learning are being used to improve the customer experience without us even knowing.

I’d like to look at one of these areas today and explore where machine learning is really taking off. Known as conversion rate optimisation, it has been a subject of much debate for years but often overlooked as “too complicated” or “too costly”.

According to Forrester, just a 0.2% improvement on optimising conversions could increase revenue by 10%. And currently ecommerce sites are only spending 1% as much on converting customers as they do on attracting them in the first place.

So what has been holding us back? Why hasn’t conversion rate optimisation been a focus for retailers before?

The answer lies in machine learning, and in this article I’ll be explaining why we can expect conversion rate optimisation to become a central, achievable part of an effective customer experience in 2017.

Machine learning increases accessibility

Between manual solutions, informal service-based approaches and claims from existing vendors that they could build it into their offering, buyers were hard-pressed to know who to trust and which areas to focus on with available budget. At the same time, the costs were still prohibitively high for most.

Now, with a machine learning based approach, conversion rate optimisation is much more clearly defined, and within reach to even smaller retailers.

This is crucial. To take full benefit of optimisation opportunities, you ideally want it to happen invisibly within each customer experience. Conversion rate optimisation no longer has to be so costly, time-consuming and arduous, if machines can help carry the load.

You take all your data relating to conversion rates, feed it through incredibly efficient neural-networks and then, in real time, your customer experience can be updated.

So this is the first big shift - the right kind of automation helps us do more with less.

Customer data volumes are bigger than ever

The second shift we’ve seen to make real time the norm is a massive increase in speed and systems available to crunch data.

On Black Friday, brands like Puma can now quickly access large quantities of data to inform optimisation. But even with computational power increasing faster than even Moore’s Law predicted, you can’t just flick a switch to apply that extra power and solve the problem. However quicker it is to crunch numbers, it still takes intelligent design and the right tools on the back end to get the results you need.

The exercise is really one of organising data. For this, we are big believers in the role of neural networks to identify insights quickly and efficiently. Using machine learning and tools like Google BigQuery, we have reached a tipping point for applying these techniques to daily challenges.

This rate of change is not going to slow down. This time next year, we should have even more power to play with, machine learning tools will have had even more datasets to learn from and we can expect to progress even further.

Cost and customer experience

Cost per action (CPA) has been a fairly standard business model for this kind of optimisation. But the fact is, this doesn’t work to the customer’s advantage and mostly results in users throwing money at the problem without seeing any tangible results.

A CPA model means every marketing technology you use ends up competing to play their part in the customer journey. And nothing undermines a great customer experience like the technology behind it intruding without offering any further value.

Because automation makes the process more efficient, the alternative is to pay a monthly subscription fee, like any other software-as-a-service product. No more battles to capture a last click ruining your customer experience - just a simple, fair system.

The other efficiency comes from running optimisation alongside acquisition efforts. If you can convert more of the traffic you attract, then your acquisition cost can be spread across more customers. Naturally, this means you are getting better ROI right across the chain.

A less confusing future

Used properly, technology makes things faster and simpler. That’s what’s really happening here. Conversion rate optimisation is no longer a confusing and ambiguous, disappointing stable of competing disciplines.

It’s a clearly defined area and an opportunity to get more from the customers you work hard to attract. And for the time being much more useful than asking a bot to order you a pizza.

James Critchley is co-founder and CEO of cloud.iq.

About JamesCritchley


Please login or register to join the discussion.

25th Oct 2016 10:07

Great well balanced article, thanks James!

Thanks (1)