In an industry where brands, agencies and technology providers are always chasing the new shiny thing, generative AI has perhaps created the most buzz in recent memory. Beyond ChatGPT, Google and Microsoft, Salesforce made the most recent splash in AI with the introduction of generative and predictive AI capabilities across commerce and marketing.

Many brands are now grappling with how AI can help them create more personalized shopping experiences before, at and after the “buy” button. To do so, they’re considering how to implement this tech to support promotions, online chat, on-site product descriptions, one-click checkout capabilities, email content, product images, enhanced audience segments and journeys, all in less manual ways. Customers expect brands to anticipate their needs, and AI can help them do so in a scalable way.

But the chatter around AI is just the bait. The real conversation must first focus on the data foundation. AI models are powerless without clean and unified data sets. “Data foundation” isn’t sexy; it may not grab all the headlines or set the premise for the next episode of Black Mirror. But before leveraging AI, it’s critical for brands to focus on creating an actionable data foundation. In other words: garbage in, garbage out.

Data as the Foundation

Many brands still struggle with harmonizing distinct data sources into a singular customer view. Beyond that, they must make the view clean, accurate and accessible to marketing and business teams, allowing them to surface insights for decision-making, planning and optimization. Siloed systems, complex processes and billions of data points compound these challenges, which are greatly magnified by the advent of AI. Brands shouldn’t rush into AI without tackling the data foundation first.

From a tactical perspective, foundational steps should include:

• Assess: Identify data sources and plan how you will ingest them in a scalable way, no matter where the data sits (e.g. CRM, Salesforce Commerce Cloud, Salesforce Marketing Cloud, Google Cloud, Amazon, Azure, POS).
• Prep: Examine source data to understand what cleaning or filtering steps might be required in the harmonization phase. Prepare to clean the data, make sense of it and separate it as needed to later power insights (e.g. by region, brand, business unit).
• Model: Design the unified data model and storage layer to deliver the expected derived insights.
• Unify: Data must then be mapped and ingested into the model for harmonization across all sources. AI can be trained to operate on your data, but the outputs will disappoint if the data isn’t of high quality.
• Visualize: Performance analytics like sales conversions, email and media outcomes, customer LTV and propensity must be visualized to surface insights for optimizing models against business goals.
• Optimize: Data work is not only foundational, but also perpetual. ML/AI models will learn to the best of their ability if they’re continually optimized, especially with a human in the loop to validate quality.

Marketing Technology News:  MarTech Interview with Anil Malhotra, Chief Marketing Officer at Bango

As a final reminder, don’t underestimate the importance of the data prep phase. Oftentimes, brands aren’t fully aware of the types and quality of data they are (or aren’t) collecting. Spend extra time in this phase specifically. Guidelines include:

• Define sources and data types located at those sources: Document methods of extraction: structure, volumes and schedules. Design transport mechanisms.
• Analyze and identify gaps: Plan the unified data schema by understanding the structure, and by defining identifiers and relationships. Avoid structure errors, including missing keys, missing required data and value issues. Uncover irregularities like irrelevant or illogical dates and invalid ages.
• Determine data clean-up rules: Create fixes for bad data, align data values to correct types and ensure data is consistent.
• Review data governance practices: Ensure complete and accurate documentation, as well as review data definitions.

Unlike many new shiny things that marketers have gravitated to in the past, AI is likely here to stay. It will change the way we market to consumers, build and optimize on-site commerce experiences and even live our daily lives. All brands should be looking to tap into the potential of AI; just be sure to look past the hype and take a data-first approach.

Marketing Technology News: Is Virtual Production The Holy Grail of Advertising?