84 Per cent of marketing organisations are using machine learning, according to industry sources.
One of the most notable areas where machine learning is burgeoning and delivering fast-paced impact is improving conversion and it’s easy to see why, considering the value machine learning can add to the performance optimisation process.
Businesses already on the AI bandwagon are certainly stealing a march of their competitors – and successful machine learning cases are definitely not just the remit of large dot coms.
The addition of AI-powered machine learning can not only take the time and hassle out of poring through masses of data, but also finds new ways to monetise this data and deliver personalisation at scale. By providing faster and more intelligent feedback, it’s allowing more time for performance teams to experiment and apply creativity to greater effect; and this is the stuff that will deliver a fast return and greater impact on the bottom line.
So the question is – what’s stopping more uptake of machine learning in revolutionising digital product performance?
Firstly, many organisations are still in the traditional mindset of focusing on delivering new features and ‘build’ rather than measuring impacts and outcomes as the product is developed, iterating from a MVP. Machine learning goes hand-in-hand with a product thinking approach and agile working to continually evolve a digital product based on real time, real user experiences.
Secondly, effective use of data for improving conversion can only happen with the right data strategy and good data foundation. Of course, joining up data sources is easier said than done, but if you can enrich your data to allow patterns to be drawn, then you will be in a much better place than the majority of your competitors to create a relevant and engaging customer experience and improve your site conversion.
The other barrier to implementing new machine learning tools and practices is, like with many digital transformation practices, simply not knowing where to start.
So what are the three most easy-to-implement machine learning techniques you can use to improve conversion right now and to powerfully personalise your customer experience?
Data mining or ‘opportunity mining’ is the practice of using an algorithm to search for opportunities in the data. The algorithms are designed to ‘mine’ large data sets to find patterns of behaviour – this can either identify issues with current campaigns or indicate opportunities for future growth.
For instance, the data may identify that PPC traffic converts well at weekend and not much during the week, so you might adapt your campaign spend based on that information. Or you may find a campaign isn’t working well in a particular region and may adapt the messaging based on this insight.
Data mining is useful in identifying patterns, but does not provide solutions. Machines can’t make up empathetic experiences. They need to be used in conjunction with experimentation and other conversion optimisation methods to really make the most of the practice.
By applying AI to identify these patterns to greater effect, the insight may not only find opportunities that might be hard to spot but enables greater time to be spent on acting on the insight, achieving better conversion rates based on real-time trends and problem solving.
Qubit and Needl Analytics (with their virtual web analyst tool) are two technology companies that are driving this approach.
Predictive algorithms power recommendation engines such as ‘what others also bought’. They are what most people think of when it comes to personalisation. AI applied to this process can help businesses go much further and become much more sophisticated in their personalisation.
For example, Amazon uses its predictive algorithms to forecast user interest, how motivated a user is to buy, and the sales volume of the individual products. It then provides product recommendations based on this behavioural and operational data (so, for example, it’s likely that you and I will get a different set of search results for ‘green hat’). It is also why you sometimes get search results for a product with little or no reviews above items with hundreds of reviews in the search results.
This personalisation tool accounts for 35% of Amazon’s sales and is obviously very complex, requiring a huge amount of data maturity, but it highlights the opportunities with a strong data strategy.
Nudgr, Sentient Ascend and Fred Hopper are all currently offering machine learning technology worth considering as part of this process. They are all driving different types of predictive tools – to deliver more relevant user experiences.
Best match algorithms
Adding automated intelligence to A/B testing with the use of ‘best match algorithms’ (or ‘multi-armed bandits’) is proving an effective way to capture value sooner during an experiment or campaign.
With best match algorithms, as soon as a winner between two variants being tested becomes obvious, more traffic will automatically redirect to that winning test. This means that the test is more likely to reach statistical significance and the benefits will be immediate. The New York Times website currently employs this technique with its headlines as a fast and highly efficient way to improve its article readership.
This type of algorithm is great for short-term promotions and campaigns but can also be used in long-term tests.
Say you have two versions of a homepage, and over time, one proves much more popular, getting 95% of all traffic. Then a seasonal change occurs – for example, Black Friday – the page that initially delivered much lower performance might suddenly start performing better. The best match algorithm would detect this and start directing more traffic towards the second page, and then automatically switch things back once the seasonal change alters traffic figures again.
As an indication of the speed at which the addition of machine learning can achieve change, Optimizely claims that its Stats Accelerator product enables companies to accelerate experimentation and reach statistical significance up to 300% faster with intelligent traffic optimization. Using machine learning, Stats Accelerator automates the flow of traffic to your experiments, so you can drive learnings and impact more rapidly.
So is machine learning a silver bullet for conversion best practice?
Technology alone will not achieve the most impactful customer experience and digital excellence. Machine learning offers a new level of speed and accuracy to performance optimisation but it can’t be used in isolation or be thought of as a replacement for the skills of a data scientist. Any use of AI needs to be balanced with human intelligence and closely managed, there is still a need for investigation and interpretation.
For many organisations it will be a matter of finding a solution that is right for them, that fits in with their capabilities and goals. While many large in-house software teams are undoubtedly developing their own solutions, the application of machine learning into performance is certainly something that is gaining traction within agency / client partnerships.
Within our own client base at Code Computerlove, for some businesses like Hillarys and Missguided, getting on the front foot with machine learning has been a high priority. These businesses already have a strong data driven strategy that we’re now able to take to a higher level with new tools and benefits that ‘greater, faster’ insight enables.
Machine learning applied to performance is a hugely exciting space – offering lots of potential, and new technology and solutions are entering the market every month. We expect more businesses to be benefitting from the efficiency and optimisation that it can drive as a fast paced area of growth this year.
If you want to learn from those who have already implemented their AI and machine learning strategies, why not attend Supercharged, Econsultancy’s AI and marketing conference, 1st May, London.