AI IRP sme

The problem to be solved is that of unprofitability for online retailers, manufacturers and distributors. Unprofitability is the primary reason why e-commerce businesses fail.

In online trading, the cumulative effect of recurring, identifiable problems contribute significantly to the problem of unprofitability: unmanageable amounts of diverse and
distributed data, the need for a multiplicity of tools to monitor and analyse data, inadequate revenue-focussed direction generated by analytics software, a lack of immediately available, real-time information, imprecise attribution and accountability regarding digital marketing, constantly leaking revenue and inefficient business workflows.

The IRP Trading Terminal (ITT) was designed from the ground up to address every one of these issues. Initial releases of the ITT incorporated complex data mining and statistical analysis techniques to generate insights for IRP clients. This capability was combined with benchmarking and performance ranking analysis to address the problems of attribution and accountability within e-commerce.

This technology was extremely successful in terms of addressing the key issues contributing to unprofitability for IRP clients. However, in itself, the technology was not sufficient to continually drive up sales and profits – the fundamental requirement for solving unprofitability. To progress further, the RnD team realised that it was essential for the ITT to be capable of accurately predicting consumer purchase probabilities and intervening to close sales. This capability in turn necessitated machine learning and probabilistic learning functionality.

The RnD team selected the Accord.NET Framework to develop the machine learning capability of the ITT. This was chosen because it uses the same development environment and programming language as existing IRP technology (.NET and C#) and also because of its extensive collection of development libraries. In the ITT, the Framework has been used to develop pattern recognition, probabilistic learning, data visualisation, statistical processing, graph plotting and performance measurement functionality.

The RnD team applied the machine learning technology to manipulate and increase each of the variables in the fundamental mathematical equation underlying successful e-commerce: Site Traffic x Conversion Rate x Average Order Value = Total Sales. For sales to keep increasing, the ability to accurately predict the probability of a purchase occurring was essential, as was the ability to continuously learn from previous successful interventions and feed that information back into the machine learning algorithms.

The team had to overcome several difficulties when developing the software, primarily relating to the sheer volume of data involved and the speed and performance of the completed system when running in real time.

Now, the ITT harnesses the power of machine learning and statistics to turn an IRP client’s ‘big data’ into actions that boost sales and profits by:

● Analysing hidden patterns in the primary data generated by the client’s business, extracting the high-value information and advising e-commerce trading team members on the optimal strategy to adopt to accelerate sales.

● Computing the probability that a site visitor or customer will make a purchase, based on many variables including their existing data and purchase history, their location, language, device and current real-time behaviour.

● Working on IRP clients’ behalf as a sales agent to manipulate individual customer sessions, engage directly with customers or entice them with personalised and effective
offers.

● Suggesting very practical actions, adjustments to traffic channels and precision conversion techniques in order to positively and continually impact the client’s growth rate.

The ITT learns the sequences of moves that increase sales, informs users what moves they should make next and who should make them – for example, which performance-ranked service provider should be used for a specific digital marketing channel such as pay-per-click advertising or affiliate marketing. Over time, the ITT learns from the results of its own interventions and becomes increasingly effective at closing sales. The ITT is constantly calculating who and what is providing value, monitoring performance in real time to generate ever-increasing revenues.

This is disruptive technology as it fundamentally alters conventional approaches to online selling. The ITT removes reliance on ‘gut feelings’ about key decisions or trial-and-error guesswork and replaces them with evidence-based direction.

ITT technology increases profitability and growth for Irish online retailers, manufacturers and distributors who come on board the IRP Commerce Cloud platform. It helps to strengthen businesses with outbound export trade and potentially creates jobs in multiple areas, including website administration, customer relationship management, digital marketing, finance, HR, warehousing, logistics and all the other roles associated with prosperous businesses. This is good for the economy and for Ireland’s reputation for e-commerce success and expertise.