Why Most Companies Are Less AI-Ready Than Their Dashboards Suggest

Front Lines is a podcast network built for the people who build and run technology companies. Its Marketers of Technology show sits down with the operators running marketing inside tech businesses: the people who own pipeline, attribution, and the systems underneath them, not just the campaigns on top.

That made it a natural fit for this conversation. My work lives in the layer most marketing teams inherit and rarely get to fix: the GTM systems, the CRM, the data flowing between them, and now the AI being bolted on top. And my angle is one most AI conversations skip. AI readiness isn't about what you bought. It's about whether your data and processes can actually carry it.

90% of organizations aren't ready for AI

In my experience, about nine out of ten organizations aren't ready for AI, and the software they picked isn't the reason. The reason is the foundation. Three patterns show up again and again.

Fractured data pipelines. Leads come in through forms, ads, events, and partners, and each source lands in the CRM a little differently. When you point AI at that, it inherits every gap. Scoring models score half a record. Routing rules route on fields that were never filled. In practice it looks like "the AI isn't working," when the AI is doing exactly what the data told it to.

Attribution that tells the wrong story. Most teams have a dashboard that credits the last touch, or the channel that's easiest to track. AI trained on that story doubles down on it. You end up automating budget toward whatever your tracking happens to see, not what actually drives revenue.

Duplicate CRM records under a clean dashboard. This is the sneaky one. The top-line report looks fine, but underneath there are three versions of the same account, two owners, and conflicting lifecycle stages. Put an AI agent on that and it emails the same buyer twice, or skips them because it thinks someone else has it.

Teams usually judge their readiness by the stack they paid for. That's the wrong yardstick. If the inputs are messy, AI just moves faster in the wrong direction. If you want the longer definition, I broke it down in what AI readiness actually means.

Why diagnose before you sell a fix?

Before you buy an AI tool, you need to know whether your team can actually use one. That's what the AI Readiness Assessment is for. It takes about two minutes, and it measures two things: how fluent your team is with AI today, and how much manual work is slowing it down. You walk away with one clear next step, not a 40-page report.

Readiness is a people and process question first. A team that doesn't trust its data or doesn't know where AI fits will stall no matter what platform you hand them. Diagnosing that up front saves the budget you'd otherwise burn on a tool nobody adopts.

The second diagnostic is the one I built as Stack Finder's flagship product: the revenue agent. Any business can run it against its own website in under two minutes (on the episode I said "thirty seconds" as that was before I bolstered its capabilities by analyzing your home page and top 10 high intent pages), and it sorts what it finds by department. You get a full checklist and a step-by-step workflow plan of fixes to your revenue leaks and cracks before anyone asks you to buy anything.

That flips the usual sales motion. The value shows up before the first call, so the conversation starts with proof instead of a pitch. The revenue agent is also great for external teams to do competitor analysis on areas your competition is leaking revenue, or for signal-based intent outreach. 

Is data hygiene the real bottleneck?

Yes. Every time I dig into a "we need a better AI tool" problem, I find a data problem underneath it. Duplicate records, broken handoffs, and attribution nobody trusts don't get fixed by adding another layer on top. They get fixed by cleaning up the records and processes the AI is supposed to run on. Otherwise you get AI that moves fast in the wrong direction, and it's much harder to unwind than a slow manual process.

If you're not sure where to look, check these first:

• Duplicates. How many accounts and contacts exist more than once, and which one is the source of truth?
• Handoff fields. When a lead moves from marketing to sales to customer success, which fields are required, and how often are they actually filled?
• Attribution source of truth. Is there one agreed answer to "where did this deal come from," or does every team have its own report?

That's why I treat data decay and revenue leaks as AI problems, not ops chores.

How much hands-on help should a customer get?

As much as they want, and no more. After the diagnostic, teams pick one of three paths:

• Fix it in-house. Take the checklist and run with it. This fits teams with an ops person or a marketer who already knows the stack and just needs the priorities.
• Automate the fix. Use the in-CMS fixes from the CRO Agent. This fits lean teams who want the website issues handled without opening a ticket.
• Hand it off. Bring in a partner from my execution network. This fits teams who know what's broken but don't have the hours to fix it.

That split came straight from customer feedback. People wanted the diagnosis, but they didn't always want to do the implementation work themselves. So the offer has three tiers: the free diagnostic, a 90-day GTM strategy engagement, and hands-on tech stack optimization for teams who want me in the room.

Why give away the playbook?

I used to treat my expertise like proprietary gold. I figured if I gave away how I diagnose and fix GTM systems, nobody would need to hire me. About a year ago I reversed that and started posting my actual operator experience on LinkedIn instead of generic advice.

The opposite happened. Real stories from real systems landed in a way polished thought leadership never did. That shift took me from a standing start to close to 18,000 people reached in about a year, and it's now a steady source of qualified inbound.

It's also a big part of how I run Stack Finder as a one-person business backed by a network of partners. Giving the playbook away does the qualifying for me. The people who reach out already know how I think.

Why build the education layer before more tooling?

Most teams don't yet know what a given AI action costs to run, let alone how to architect around it. They're buying seats and credits without a mental model of where the cost comes from or where the value shows up.

So alongside the paid work, I run a free Stack Finder Academy with beginner, intermediate, and advanced tracks covering AI foundations, workflow design, and token economics. The beginner track gets people comfortable using AI day to day. The intermediate track is about designing workflows around it. The advanced track gets into what it costs to run and how to build for that.

More tooling won't fix a knowledge gap. Teaching people how the pieces fit will.

Where should you start?

Start with the plumbing, not the purchase order. In order:

1. Take the AI Readiness Assessment to see where your team stands on fluency and manual work.
2. Scan your site with the revenue agent to find the cracks by department, which gives you your revenue leaks.
3. Take the Revenue Bottleneck & Leak Audit Assessment to find the broken systems.
4. Fix your broken processes and revenue leaks first (duplicates, handoffs, attribution) before looking to add AI.