Making Sense of AI and Customer Data and Getting Executive Buy-In

A woman on a laptop against a city's skyline

“Brands who fail to leverage more of their customer data also fail to retain relevance and loyalty with their customers.”

Forrester, VentureBeat, and Blueshift joined together in a webinar titled “AI-Powered Marketing: Put Your Customer Data to Work” to discuss how brands can make the most of all their customer data across all marketing channels. The crux of the conversation revolved around using Artificial Intelligence to make the most of your customer data and the inherent problems and myths marketers today face when trying to cut through the hype and really make AI work for them.

In this article, I will address three hotly discussed questions that our viewers asked on the topic of AI and Customer Data. For AI, we’ve moved from a state of novelty to a state of necessity, and these questions revolve around how to get Artificial Intelligence into an organization and how long should it take. So you don’t have to read all of it, I summarized the responses from analysts at Forrester, VentureBeat, and our own co-founders.

What is the best way to “sell” the need for an AI strategy or investing in AI to the C-level?
TL;DR: To get the most use of your customer data, you need AI (there’s just too much of it) — and getting buy-in for AI is a complex process, but when approached with a solid measure for success and a “crawl, walk, run” approach, it gets very doable.

How do you differentiate Machine Learning and AI (deep learning, etc.)? A lot of content presented today has more to do with machine learning than AI, don’t you think?
TL;DR: they’re not exclusive topics…but it’s time to cut through the AI-hype-cycle, and understand what AI and Machine Learning really are and how they help you.

Once you sign the dotted line for an AI platform, how long would one allocate toward implementation and planning and executing campaigns?
TL;DR: Implementing an AI platform gets better with time and shouldn’t be a “one-click solution”, however you can start seeing results in weeks. It takes partners who will guide you through the process and help you plan/organize/execute on the data and systems you have.

Full, more thought provoking answers below…

And if you would like to watch the full webinar, you can check it out here.


Manyam Mallela, Chief AI Officer at Blueshift
Crawl, walk, run: Start with a specific use case to show results and build from there.
“The best way to get executive buy-in for newer artificial intelligence technology platforms is to show how AI improves immediate KPIs. Start by targeting and building personalization in one or two campaigns using segment-of-1 marketing, and do this without having to hire more data analysts/engineers/scientists,” says Manyam Mallela “Once a specific KPI improves, it’s easier to make the case for a wider roll out of AI in the organization.”


Rusty Warner, Principal Analyst at Forrester Research
Focus on how it will improve the customer experience in a specific way and test it.
“There are two levels to selling the need for AI into the C-level. First, start with the benefits AI will bring to the organization as a whole. Focus on what it will do to improve the customer experience (or, digital experience), then look for a specific use case so you can prove the value of the technology.” says Rusty Warner “Second, many of the barriers have been lowering as executives have begun to understand more what it takes to bring in AI and the the value it can surface.”

Before you touch your marketing platforms, understand what your KPIs are (Marketing 101) – what are you trying to accomplish and how will you measure this. Have a plan! In addition, marketers must be able to easily activate customer data quickly in their campaigns across all channels. Selecting a “single-point solution” that only solves one issue backs you into the current situation most marketers face today of a “frankenstack” of band-aided technologies with overlap in functionality and disconnection between their data sets.


Vijay Chittoor, CEO of Blueshift
For marketers, it’s the difference between manual, “driver assisted” automation, and truly autonomous “self-driving” automation
“When looking at the difference in Artificial Intelligence and Machine Learning, think of the evolution of “smart cars”/driverless cars.” says Vijay Chittoor “Machine learning would be similar to “driver-assisted” cars where the machine learning is in the passenger seat. It’s helping in the driving but not actually doing any of the driving for you. For instance when backing up, you might get an alert of nearby objects from machine learning – you, as the diver, must still make the decision of what to do. Contrary to “driver assist” (machine learning), a “self-driving” car would process all the incoming data, assess the situation, and decide on the the best action to take…autonomously. This is what Artificial Intelligence is — the “self-driving car” that ingests, analyzes, and takes action on its own using predictive scoring/models that would have the best outcome.”


Stewart Rogers, Analyst of VentureBeat
It’s the difference between finding and applying a pattern, or, taking a pattern and actually improving upon it
“Machine learning is finding patterns in the data, and by finding patterns in the data, we can apply that pattern to other data. This is useful for predictive analytics, for example.” says Stewart Rogers “With Artificial Intelligence (AI), on the other hand, we have a system that takes the pattern and then makes the best version of it order to optimize or improve upon the data so you get a better result.”

Most legacy marketing platforms use a form of marketing automation that’s more driver-assisted where it’s a simple triggered response to an event being fired or an action being taken, like a yes/no logic in a workflow or a simple “if/then” logic that must be manually built for every outcome by a marketer. There’s no “thinking” involved, no decisions based on continuous learning. Sure, it’s automated, but it’s time consuming, very manual, full of list pulls and exports, and has no real intelligence behind once it is executed. You’re just applying a pattern to a problem and letting it run.

New platforms use the “self-driving” model, where artificial intelligence runs the show — essentially making the marketer smarter and able to really scale their efforts by making informed decisions at remarkable scale across channels. Modern technologies continuously learn and optimize for better performance.


Manyam Mallela, Chief AI Officer at Blueshift

It’s not something you get by just pushing a button, however you can see value and results in weeks, not months
“We have seen over 90 percent of our clients go live with high impact campaigns in less than 4 weeks with a continuous rollout after that based on their priorities.” Manyam Mallela, Chief AI Officer at Blueshift “Blueshift’s unique architecture allows faster on boarding by not having rigid data schemas or limits on size and types of data.”

Activating and getting value out of AI doesn’t come from just a push of a button… But, at the same time, seeing value from AI shouldn’t be a year-long effort. Modern platforms built from AI help marketing and product departments initiate complex campaigns with ever-changing data quickly. And most of all, these campaigns get better. Older, bolt-on solutions introduce lag at every step, and anyone who has the battle-scars from trying to bolt AI onto their “Frankenstack” knows the struggle and countless months spent just trying to get a single campaign out.

The more an organization thinks intelligently about their customer data (from how it’s structured, to where it’s stored, and how it is used), the better prepared they are to truly put their customer data to work using AI. No need to fret though, even if your customer data isn’t all housed in a massive data lake or in one system, there are platforms that will help you unify your data, create a true Single Customer View, and then give marketers the autonomy to actually run campaigns quickly and with the most up to date data.