Does this problem really require AI? Brands should have a defined problem set or desired outcome in mind before considering AI. Their challenge or objective should guide their AI journey -- and reveal whether they actually need AI or not.
For example, if creative optimization is the goal, consider whether adding an AI vendor into the mix will change results significantly. For brands with a strict, relatively small creative set, no AI system will help them better understand if changing “click now” to “buy now” will have impact.
On the other hand, for brands with hundreds of thousands of creatives, AI can help sort the performers from non-performers by revealing the signal in the noise.
Vendors should also be able to explain why traditional methods have failed until now, and why AI is necessary to succeed.
What does the machine do -- versus what does your marketing team do? Do you want the AI to step-analyze data and provide you with recommendations you can execute on your own? Or do you want it to take actions on your behalf in pursuit of KPIs?
An AI that helps with decisioning vs. an AI that makes and acts on the decisions in real time operates at two very different altitudes. For organizations that aren’t particularly wedded to one option, an alternative question to ask is, “What level of automation is sufficient to solve our workflow and scaling challenges?”
Will it play well with other systems and data? AI requires massive data sets to perform best, so marketers will want to be able to integrate it with other systems, giving it access to many datasets created by their efforts. The two questions to ask here are: Can the AI at hand be integrated with your brand’s other systems? And, straight to the point, can this AI be enriched with external data sources?
If the AI can be integrated with existing systems, how long does the vendor’s typical integration take?
Weigh time to market and ability to integrate large datasets against the end benefits among the vendors under consideration.
Who’s building it? AI is very challenging to build, so it’s important to understand the DNA of the vendor’s team. How much of the team is dedicated to research and development? If, for instance, they have 100 people but only five of them are in R&D, it's likely they have relatively simple technology.
Also consider who is on the R&D team. Do they have backgrounds from high-profile research universities and have on-the-ground practical AI experience? Or, are even the most senior members of the team learning AI on the job?
AI comes with a price tag, so it’s important for marketers to know exactly what they’re paying for.