Koka Sexton

TL;DR

• AI is only as good as the data feeding it. Most B2B marketing teams are sitting on fragmented, inconsistent, and incomplete data spread across a dozen tools. Throwing AI on top of that does not fix the problem. It automates bad decisions faster.
• The marketing data stack needs three layers: a single source of truth for customer data, a clean enrichment pipeline, and an AI orchestration layer that routes signals to action. Most teams have none of these.
• First-party data is your moat. Third-party data is degrading. AI crawlers are vacuuming up public content. The only data your competitors cannot access is the behavioral data your own customers generate. Build around it.
• You do not need a CDP to start. You need a CRM that is not a disaster, a consistent enrichment process, and a commitment to never letting dirty data accumulate again. The foundation is discipline, not technology.

I have been through this with my own operations. Multiple properties. Multiple data sources. Contacts coming in from LinkedIn, from website forms, from enrichment tools, from manual imports. At one point I had six different data sources feeding into my CRM and zero consistency across any of them. Some contacts had company names but no titles. Others had job functions but no industry. The data was technically there, but it was useless for segmentation, scoring, or any kind of automated action.

I fixed it. Here is how, and why this matters more than ever now that AI is in the picture.

The Real Cost of Bad Data (It Is Bigger Than You Think)

Most marketing teams treat data quality as a nice-to-have. Something you get to after the campaigns are running, after the tools are configured, after the quarterly targets are set. This is exactly backwards. Bad data is a tax on every downstream marketing activity. It makes your segmentation wrong, your personalization generic, your lead scoring unreliable, and your attribution meaningless.

Here is the math I have seen play out repeatedly. A mid-market B2B company runs a $50K campaign. Using their current data, they segment by industry, send targeted emails, and route leads to sales. If 20% of their data is wrong (wrong industry, wrong title, wrong company size), they are effectively lighting 20% of that budget on fire. That is $10K per campaign. Run six campaigns a year, and bad data is costing you $60K annually in wasted spend alone. Add the opportunity cost of deals lost because leads were misrouted or never followed up on, and the real number is 2-3x that.

Three Layers Your Data Stack Needs Before You Add AI

Layer 1: Unified Customer Record

You need one place where every contact lives with consistent data. For most teams, this is the CRM. Not the email platform, not the analytics tool, not the spreadsheet someone on the sales team maintains. The CRM.

The CRM is where enrichment happens, where scoring happens, where segmentation is built, and where sales and marketing agree on what a qualified lead looks like. If your CRM data is inconsistent across sources, everything built on top of it (campaigns, scoring, reporting, AI agents) will be unreliable. Fix this first. Everything else depends on it.

Layer 2: Automated Enrichment Pipeline

Manual data enrichment is a bottleneck. Someone on your team is spending hours looking up company sizes, verifying titles, and filling in missing fields. That person costs you $80K+ per year and could be doing work that actually generates revenue.

Automate enrichment at the point of entry. When a new contact enters your CRM, an enrichment tool should automatically append company data, firmographics, and relevant signals. No human involvement. The contact goes from raw to enriched in seconds, not days. I use Apollo for this, but the tool matters less than the principle: enrichment must be automatic, consistent, and applied to every single contact. No exceptions.

Layer 3: Signal Routing and AI Orchestration

Once your data is clean and enriched, AI becomes useful. Not before. An AI agent sitting on top of bad data will confidently make wrong decisions. An AI agent sitting on top of clean, enriched, consistent data can route leads to the right sequence, surface buying signals before a human would notice, and automate the low-impact parts of pipeline management.

This is where the real value of AI in marketing lives. Not in writing blog posts. Not in generating ad copy. In processing thousands of signals across your contact database and routing the right actions to the right people at the right time. I have built this layer into my own operations and it has eliminated the manual triage work that used to consume hours every week. The full architecture is documented in my piece on building an AI-native marketing stack.

“AI on top of dirty data is a liability. AI on top of clean data is a force multiplier. The difference is the foundation — and most teams skip the foundation.”

Here is what automated enrichment looks like in a well-built system. A new contact enters your CRM from any source. Within 30 seconds, an enrichment tool appends their company name, industry, employee count, revenue range, job title, seniority level, and LinkedIn profile URL. If any field enrichment fails, the contact gets flagged for review. If all fields populate successfully, the contact immediately enters the appropriate scoring model and is routed to the right sequence or assigned to the right rep.

This entire flow happens without a human touching it. The marketing ops person who used to spend Monday mornings manually enriching leads now spends Monday mornings analyzing conversion data and optimizing campaigns. That is a structural reallocation of talent from maintenance to growth. The enrichment pipeline also creates a feedback loop: every contact that moves through the system teaches you something about your data quality and which sources produce the most complete records.

First-Party Data Is Your Moat

Third-party intent data is getting worse, not better. Privacy regulations are tightening. Cookie depreciation continues. AI crawlers are vacuuming up public content and making it available to everyone. The only data advantage you can build that your competitors cannot replicate is the behavioral data your own customers and prospects generate on your properties.

Every website visit. Every email open. Every content download. Every LinkedIn engagement. Every demo request. These are your signals. They are unique to your business. They tell you who is interested, what they are interested in, and when they are ready to buy. No third-party data provider can give you that.

Build your infrastructure around first-party data. Own it. Enrich it. Route it. The companies that do this now will have a structural data advantage in 2028 that late adopters will never close.

Think about what happens when a prospect engages with your content. They visit your website. They open your emails. They download your resources. They engage with your LinkedIn posts. Each of these actions is a data point that tells you something about their intent. Aggregated across your contact base, these signals form a picture of who is in-market, what they care about, and when they are ready to buy. No third-party data vendor can give you this picture because it is built from interactions that are unique to your business.

Start Here: The Weekend Data Cleanup

You do not need a six-figure consulting engagement or a CDP implementation to get started. You need a weekend and the discipline to clean what you already have.

1. Export your CRM contacts. Every single one. Look at the field completion rates. I guarantee you will find that 15-25% of contact records are missing critical fields like industry, company size, or job function. Those records are invisible to your segmentation and scoring. They might as well not exist.
2. Delete duplicate records. Merge them. Every CRM has duplicates. Most teams just ignore them because deduplication is tedious. But duplicates fragment engagement history. A lead who opened five emails looks like they opened two because their activity is split across duplicate records. Your scoring is lying to you.
3. Set up automated enrichment. Connect an enrichment tool to your CRM and set it to run on every new contact automatically. No manual steps. No human decisions. Every contact gets enriched within minutes of entering the system.
4. Define your essential fields. What are the 10-15 fields that actually matter for segmentation and scoring? Industry, company size, job function, seniority, recent activity, source. Everything else is noise. Enforce completion on these fields. Do not let incomplete contacts into your CRM without flagging them.

This is not glamorous work. Nobody writes LinkedIn posts about cleaning their CRM data. But it is the foundation that every AI initiative, every campaign, and every pipeline forecast rests on. Ignore it and everything you build on top of it collapses. Fix it and you produce compounding returns across every marketing activity you run. This is the natural next step after a marketing stack audit. I covered the full stack consolidation framework in my piece on why your automation stack is costing you pipeline.

Ready to build a data foundation that actually supports AI instead of undermining it? Let’s talk. I help B2B teams clean, enrich, and operationalize their marketing data.