Marketing AI & Customer Data

As humans our behavior contains multitudes of patterns. Some of us get up at the same time, use the same toothpaste, eat the same breakfast, work the same hours, and come home to the same number of kids. A big suited man sang about the impact of commercialism at the dawn of cable TV in the aptly titled “Once in a Lifetime,”

“Letting the days go by… same as it ever was.”

Lots of things determine our patterns, interests, schedules, occupations, locations, income, but it is undeniable that those patterns match to the wants and needs of consumers. Starting with the printing press and probably earlier with the town crier, the first appeals to different segments of the population’s wants and needs began. Marketing.

Segmentation Is the Truth

Market segmentation is the undeniable truth about your customers. A truth that aligns with what you are selling (and should be selling more and less of). It’s the fundamental question of “who are we selling what to?” A bedrock concept for any business. The main lever of optimizing business results.

There’s a reason the mythical “1:1” personalization has never materialized into anything other than poor-performing cookie-matches in display advertising. One data point is just that. Growth is only possible at measures of scale. 

Segmentation isn’t a “nice-to-have.” It’s the only systematic way to understand, improve, and ultimately raise the value of your business — by increasing relevance and mindshare with consumers. Segmentation is personalization at scale.

Segmentation Drives Strategy & Informs Execution

Segmentation maps the strategic directions of whom to acquire, retain, and grow from your data. It dictates designs for experimentation and targeting. It guides every tactical decision from creative to channel to cadence.

Most importantly, segmentation aligns investments, resources, and level-of-effort with upside. Meaning it’s tied directly to both top-line growth and bottom-line efficiency across every part of a business — but especially for brands that rely on product and marketing.

So if it’s this foundational then why is it so often overlooked by brands?

Shaky Foundations

Despite segmentation being a first principle and information every board of directors should be strategizing against, many brands outsource their segmentation projects. They hand their first-party data to someone else, who matches that data against their own segment or audience definitions. 

Or they pay $500k–$1M for lengthy segmentation projects with agencies or consultants. This work centers on qualitative customer insights like interviews and surveys resulting in segment personas that are biased and not mapped to KPIs, IDs, or anything that makes them actionable. This is very common in Enterprise B2C.

Because of the cost, time, and lack of actionability historically associated with segmentation, many marketing leaders and exec teams have relegated it to a downstream tactical practice. Segmentation gets limited to a channel, an offer and a single KPI.

From Strategic to Spreadsheet

Segmentation has too often devolved into a bottom-up SQL query or dashboard filter that reduces data into simple pivots and “if/then” rules. One channel or KPI at a time. This has clearly not worked.

Conversion rates have stagnated if not gone down. DTC revenue mix has averaged out. Any win is short-term before regressing to the mean. And LTV? It is impossible to optimize this way.

To paraphrase Sun Tzu: “Tactics without strategy is the noise before defeat.”

As Shiv Singh recently noted, it’s no wonder CMOs are often disregarded. C-level leaders are expected to think strategically and drive business KPIs — not chase surface-level metrics.

Why Rules-Based Segmentation Breaks Down

Traditional rules-based segments rely on reductive heuristics:

• RFM scores
  
• Demographic slices
  
• Single-channel behaviors
  

They work for basic personalization — but struggle in five key areas:

Issue (Limitation)	Consequence
Linear rules assume additive relationships (Discovery)	Miss nonlinear, compounding patterns that drive outsized performance
Fixed boundaries (Speed)	Segment decay as customer behavior changes; personalization loses relevance
Human-defined features only(Precision)	Latent signals are missed; segments skew toward obvious or biased assumptions
Delivery and prioritization (Optimization)	Slow and insufficient feedback; scaled decisions lag behind clear behavior shifts
Manual scaling (Scale)	Costs rise linearly with segment granularity; optimization becomes costly and doesn’t move the needle

As data volume and complexity grow, these limitations increase, not decrease. The more data you have, the further behind you fall when using rules-based segmentation.

Your Board Doesn’t Care About SQL Queries

Your CEO and Board are not strategizing around “Abandoned Cart + Last 30 Days + Value > $100.” That “segment” your team created and those filters & rules mean nothing. Zero. Zippo. Nada to grow your business. At best they are customer fables, at worst they are sucking the lifeblood out of your P&L.

This isn’t to say channel-level or offer-level segmentation doesn’t matter. It does. But without a foundation of customer segmentation that maps to business strategy and operational KPIs, investments can’t be understood, latent opportunities remain invisible, and optimization becomes a game of whack-a-mole.

Segmentation Is Strategy—Just Ask Google and Meta

Those that could invest in strategic AI for marketing matching and targeting already have. Long ago. The biggest platforms in the world don’t guess who to target — they let AI decide. Google and Meta have built their entire ad businesses on segmentation as strategy, using neural networks to group customers not by surface-level rules and filters, but by behaviors that map directly to their objectives: conversion, engagement, revenue. 

When you run a Performance Max or Advantage+ campaign, you’re tapping into neural segmentation that learns, adapts, and optimizes in real time. These systems aren’t guided by rules; they’re guided by outcomes. They find the clusters across their customer base that look like yours but the KPIs they optimize for aren’t yours. They’re Google’s and Meta’s. That means your best customers are already being segmented and monetized by someone else’s model. The question is: why aren’t you doing the same for yourself?

***

In Part 2 I’m going to discuss the Neural Networks and Reinforcement Learning systems behind the platforms ad products. While we can hate on their dominance, they’ve open-sourced the key building blocks of their AI meaning a new generation of Marketing AI products are now in the market for brands to use. I will dive into them as well.