AI: What to do before you even start

"AI needs a baseline,” says Stephen Tompkins, the VP of media activation for Essence in APAC. “It's only as good as the inputs it has at the start. I think many people believe that they can flip a switch and AI will magically solve all their marketing issues.”

Any marketer considering AI needs to start with a tightly defined set of parameters and enable an AI to build up its delivery model, or they risk not only wasting resources during development but also failing to yield business objectives after all.

“At Essence, we have been running AI tests and have seen encouraging results and are constantly building upon it to deliver business results,” says Tompkins. “It's an ongoing process, and I believe the industry is just at the beginning of this phase. It will take some very deliberate and thoughtful rationale within the marketing organizations to fully realize the true potential of AI and mature to the next phase.”

Plumbing investments

Of the three key AI technologies—machine learning, natural language processing, and robotics—Daniel Hughes, the head of data international at DigitasLBi, and his team have helped businesses embrace digital transformation with machine learning as the initial starting point. According to Hughes, companies that are ready to make AI the main driving force of transformation need to take the following steps.

1. Gathering, cleaning, and organizing data

Hughes shared the example of a large consumer-electronics company that gave itself a mission to “fix their data plumbing”. The bulk of the work entailed deploying an extract, transform, load (ETL) system and processes to migrate data from many systems into a single location.

In doing so, the company joined data across a complex tapestry of disparate systems, both online and offline, to achieve the vision of a single, granular view of the customer and the business. 

“The project is entirely internal, it’s operational in nature and it’s not the least bit sexy,” said Hughes. “But it is positioning this company to leapfrog its peers. While others are tinkering with chatbots, this client has gotten down to the serious business of preparing for a future world that is AI driven.”

2. Tearing down silos and outdated processes that impede digital transformation

Simply put, companies that intend to embrace AI as a core value addition to their digital strategy will need to reconsider any value that silos deliver while being open to the possibility of reorganising or erasing them.

Hughes recommends that companies prepare to set new rules for how data uses are governed across functions and more importantly, how teams are empowered to make decisions and effect change without being held back by the existence of silos.

An airline client that Hughes worked with addressed the silo issue by creating an internal API (application programming interface) across the enterprise that democratised access to data and empowered teams to innovate more rapidly.

3. Customer experience digitisation, wherein AI partakes in meaningful decision-making

The integration of systems across functions, departments, and business divisions is critical for a data strategy that allows AI to pool knowledge from all sources at once, in order to offer an informed real-time correction. Unfortunately, most businesses use a variety of software systems and ERP (enterprise resource planning) systems that are locked in to only integrate with their own software extensions.

This means that businesses can either opt for a single software vendor, create their own systems, or invest in middleware to translate between systems. All of these workarounds carry costs. There’s immense risk in relying on one vendor, but developing a self-serving system unique to the needs of one business can be tremendously expensive, and middleware tends to require incessant maintenance to integrate with new versions of the systems it ‘speaks to’, not to mention new tools that the business brings into play.

For a large hotel client, Hughes recommended the second option, creating a proprietary property management system (PMS), with the aim of bringing the systems that were core and strategic to the company in-house.

Hughes said the client considered the PMS system as being on the front line in shaping the customer experience; it was the focal point where crucial decisions could be made. 

“It was the system with the highest degree of leverage, where a future introduction of AI could have the most impact,” Hughes said. “So, they decided to place a strategic bet on creating their own proprietary PMS software and investing the engineering and product management resources required to make the project successful.”

A bold move that is not without risk—precisely the kind of decisiveness and commitment that will be required for traditional businesses to transform digitally.

“Time will tell if the bet is a successful one,” said Hughes. “If it is, they will have differentiated themselves in an industry that is increasingly under siege.”