Generative AI (GenAI) is for the last mile. If you want to create something special that resonates with your target audience, you first need an underlying data infrastructure – and other forms of AI working under the surface – to facilitate this innovative approach. Matthew Biboud-Lubeck, General Manager EMEA from Amperity, outlines how GenAI can personalise communications at scale.
The emergence of GenAI has sparked significant excitement over the last two years. The output of GenAI tools is so impressive that investment in the technology increased fivefold in 2023 – with 36 GenAI companies hitting unicorn status. According to Bloomberg Intelligence, the market is now expected to grow rapidly and be worth US$1.3 trillion by 2032.
The potential is clear. But, there is also growing awareness around the practical realities of applying the technology. As businesses have scrambled to implement GenAI in various ways, they have also realised that this is not a simple plug and play solution.
For instance, we know the content creation capabilities of GenAI can be remarkable, but, in truth, the results can only ever be as good as the data it is based on. GenAI can be deployed to generate personalised customer experiences at scale – but it can only do this if brands hold accurate, comprehensive information on each individual and their preferences.
So, before any investment in GenAI can pay off, organisations must lay down an underlying data infrastructure to deliver the right information to these applications. This is one of the key reasons why almost half of business leaders say they are now actively driving forward data modernisation programmes – and this is likely to include an investment in other forms of AI.
Tackling the data challenge
While many organisations are awash with data, it’s often unstructured and siloed. It’s now more critical than ever for businesses to make this data usable. As businesses invest in their data infrastructure, we’re starting to see a seismic shift in the data management landscape, with many major vendors now adopting a ‘lakehouse’ architecture approach. This combines data lakes, which are repositories for raw data, with data warehouses, where more structured data is stored.
This, in effect, creates an architecture that allows businesses to access, read and make use of data, wherever it resides – and there are many benefits to be gained from this approach. For instance, it can save businesses a huge amount of time copying data from one system to another, as they no longer need to extract, transfer and load data to make it usable.
A lakehouse architecture also enables AI to view across multiple systems using a ‘zero-copy’ approach. This method allows companies to quickly consolidate data to build accurate customer profiles, enhancing their ability to understand and serve their clientele.
Resolving messy data
The ability to use AI in this way is helping to resolve a common problem that many large companies face – having customer data spread all over the place. Companies that want to build a unified customer profile to help personalise their communications often need to draw data from multiple sources – such as POS, e-commerce platforms, email campaigns, customer surveys, CRM systems, etc.
Added to this complexity is the fact that data is often highly inconsistent. For instance, it is not unusual for customers to use nicknames or abbreviations when interacting with companies via different touchpoints. It is possible that a business could have the same person recorded as Mr B. H. Jones, William Jones and billyj@hotmail on three separate systems.
When companies have several versions of the same person, with a history of their sales and interactions spread across duplicate profiles, it becomes difficult to personalise communications effectively.
How AI generates insight
It requires a different type of AI model to GenAI to unify customer profiles. But, when it’s possible to read data across various systems, these AI models can view every piece of information that the business holds on a customer and stitch that together.
These trained AI models can also find connections between multiple data points, so companies can determine whether William Jones and billyj@hotmail are one and the same person.
This provides a 360-degree customer view that enables businesses to see every touchpoint in an individual’s journey, from the first interaction to the last purchase. A unified view makes it possible to draw insights from past behaviour and predict what each customer is likely to buy next – which is a clear advantage when building out marketing campaigns.
Personalise communications, at scale
It’s at this point that GenAI can start to have a real impact. In the past, a marketer may have spent weeks gathering the information they needed to activate promotional campaigns. This process may have required the assistance of data engineers to use code to draw relevant insights from a SQL database.
With GenAI, however, marketers can quickly do this for themselves. They can now ask questions using natural language and find out what their most valuable customers purchase most often, what form of content they are most likely to engage with and on which channel.
This is also enabling marketers to produce highly effective personalised campaigns, at scale. When there is a unified customer profile, GenAI models can quickly determine the best way to engage with each individual and produce relevant content based upon their past purchases and current preferences – which will ultimately help to drive loyalty and sales.
Delivering ‘over the last mile’
The customer communications produced by GenAI can be highly effective. But, before we get to this point, organisations need to put a data infrastructure in place that will allow these applications to access the information they need.
It’s the same as when we receive a parcel at the front door. We only ever see the delivery person who carried that parcel over the last mile. We don’t see the global logistical infrastructure that comes before that.
In a similar way, when it comes to AI investments, we can’t just look at ‘the last mile’. We need to think about the data foundation that is enabling GenAI to generate insights that can, ultimately, increase customer retention and revenue.