How To Get The Most Out Of AI In 2025

Aggressive goals for 2025 seem to be more of the norm than the exception.?

I was catching up with a prospective customer just before the holiday break, and they were sharing their goals.

New strategic areas of focus, efficiency requirements across the entire funnel from acquisition to retention, etc. Top-line expectations with bottom-line mandates.

On the technology & data side, this customer seemingly had everything - a smart data team with top-tier data infrastructure, an end-to-end marketing stack, and an A-grade team.?

But despite all this, she was incredibly concerned over her ability to hit goal next year.

We finally got to talking about her AI strategy, and some of her frustrations at this point started to come out around lack of achievement.?

“I’ve tried a whole bunch of these AI-based marketing tools, and none of them drive the results that I need.”?

While incredibly frustrating for the customer, her experiences aren’t at all unique. Despite being surrounded by an influx of data and inundated on a daily basis by AI advancements - it’s still incredibly hard to drive truly leveraged, revenue-generating AI-enabled outcomes.

What are the core problems at play - and what can you do to drive AI success in 2025??

Problem #1: Your AI doesn’t understand business context

Let’s start with a simple example: pretend you’re a shoe retailer ShoeCo and recently started offering socks. With this as a sole point of context, you can bundle socks with shoe purchases, you can create cookie-cutter post-purchase offers, or you can blast mail anyone who’s purchased shoes in the past year with relatively non-descript CTAs.

But if you double-click into some of the reasons why your shoe brand starting offering socks, strategies might change significantly:

• ShoeCo’s socks are specifically designed to wick moisture and provide maximum comfort for their running shoes
• ShoeCo’s socks feature vintage designs that are color matched to their high end Italian leather dress shoes
• ShoeCo’s socks are a loss-leader for their discount brand. ShoeCo’s biggest competitor SockCo recently started selling shoes, and ShoeCo needed a sock offering to compete

The campaigns and strategies corresponding to any of the above strategies would result in a completely different set of audiences and personalization strategies.

But most likely your AI doesn’t at all understand any or this context.

Problem #2: Your campaign production workflows have a giant AI hole in them

For so many teams, AI is applied upstream with data science initiatives to analyze customer intent, and then it’s applied downstream to optimize channel mixes and calls to action.

Standard marketing workflows look like the following - and some of the biggest points of leverage are in the middle of the process where AI is underutilized or not used at all.

This tracks with Problem #1 - your AI doesn’t understand marketing or business context, so workflows that touch this part of the production process don’t properly leverage AI either.

Problem #3: Your AI is optimizing only a subset of the problem

Your AI strategy is fundamentally limited by what it’s able to understand.

If you setup an algorithm to optimize a specific set of discounting strategies for ShoeCo, and if it doesn’t understand other core buying reasons - then it’s going to stick to a narrow lane of optimizations.

To make a skiing analogy, to fully leverage the power of AI, you need to be constantly exploring new trails and new ways down the mountain.?

If your algorithm only knows one blue square trail then you can only get really good at running down that one blue square.

Takeways

There’s a massive AI opportunity that still lies ahead of us - and it’s all about understanding both data and marketing contexts.?

Today, your data team can do smart things by analyzing broad data to understand trends, customer preferences, and beyond. But this work is mostly done at the enterprise level and doesn’t actually understand what your marketing goals are or any campaign specifics.

On the other side, marketing AI workflows focus on optimizing specific strategies, but do so via more rigid views of what’s going on. Optimizing channels, timing, etc - this all requires measuring campaign engagement but doesn't require any sort of deeper understanding of the customer.

The Untapped Potential lies in using AI in a way that both understands deep data context and can then use it to optimize highly strategic marketing outcomes:

• Which customers might actually be interested in your new offering?
• How do you capture their attention and educate them around the benefits?
• How do you create sufficient motivation to get them to move forward and purchase?

So what’s the takeaway for your team? There are two paths forward - and they aren’t necessarily mutually exclusive:

• Align your teams. To close the gap between marketing strategy & data science, get your teams working more closely together. Embed a data scientist into your team and get them to live and breathe the same applications as the rest of the team.
• Sit tight for a next-generation tooling. The gap is closing, and in full disclosure this is a top area of focus at Simon. We’re always happy to chat with you.

To win with AI in 2025 and to hit aggressive 2025 goals, businesses will need demand more than incremental optimization. Bridging the divide between marketing strategies and data strategies is the key piece to crack. Those that do so successfully will have a real advantage in 2025.