We recently launched our DeepAgent™ technology which has been under development for the last two years.
The aspects of DeepAgent™ software that attract most attention involve Artificial Intelligence (AI) and machine learning, however the project really has only one goal — to increase sales for companies on the IRP platform.
In ecommerce this simple goal actually proves extremely difficult to achieve. The main block to sales is that companies often work against their own interests in a way that you would never encounter in any other form of selling. This happens because ecommerce is a very abstract sales channel and is often misunderstood. In fact the deep complexity of ecommerce is the opposite of the markets’ perception of the channel.
If we compare ecommerce to chess, some of the moves that companies make on the ecommerce board are really bad. Companies do this because they do not know the game well enough to understand the outcome. The DeepAgent™ project has a remit to make sure that companies make the right moves to increase sales.
How DeepAgent™ Works
The context within which the DeepAgent™ project exists is the reality we all face that many services that are currently being carried out by people are eventually going to be controlled by computers. In fact Microsoft has suggested that in 10 years’ time up to 50% of service jobs in business may be automated by software and machines. In the ecommerce arena this will mean that areas such as Pay Per Click (PPC) advertising, Email Marketing, Remarketing and many other areas will continue to trend towards automation. In the end, many business functions will be operated by software rather than people.
The DeepAgent™ project marks a beginning to the process of automating key moves that need to be made to increase sales in ecommerce and influencing those moves to occur at the micro level and at the macro level:
- As an example of micro-level influence, DeepAgent™ can analyse the data of an individual customer browsing a website and learn what offer or intervention might be required to pop up on their screen in order to close the sale. DeepAgent™ bases this decision on its knowledge of what margin there exists to play with and the customer’s likelihood of purchasing. DeepAgent™ can also learn from the results of its own interventions and so becomes increasingly effective at closing sales.
- As an example of macro-level influence, DeepAgent™ might compare a company’s Google PPC performance with the rest of the market and conclude that the company is massively underperforming. DeepAgent™ will then reach out via email or phone call, perhaps initially to the ecommerce manager but subsequently to the company owner, to provide details of the revenue that the company is losing on PPC and also which service provider could potentially run their PPC channel better than their existing provider.
The DeepAgent™ project can be summed up with two points:
- DeepAgent™ aims to know the price of everything and the value of everything that it sees in ecommerce, turning things into numbers it can understand.
- DeepAgent™ will then automatically take corrective action to increase sales — at the micro level of the individual customer or at the macro level of suggesting to the business decision makers what moves they should make to increase sales.
DeepAgent™ is a project that involves creating intelligence in ecommerce software. In due course the project will cover a huge range of areas and is going to run and run as we improve how the ecommerce game is played.