Author: Jeff Barnes | demg.ai | May 2026
Seventy-four percent of companies are getting AI wrong. Not because the technology failed them. Because they bought tools before they built strategy. That single inversion — tools before positioning, software before story, automation before audience definition — explains nearly every wasted AI budget, every ghosted campaign, every “we tried AI and it didn’t work” post-mortem you’ve ever sat through. Strategy-first operators outperform tool-collectors. Full stop.
The 74% Problem Is a Sequencing Problem
Optimizely published the number in April 2026: 74% of companies struggle to achieve and scale value from AI initiatives. The headline sounds like a technology story. It isn’t.
BCG’s 70-20-10 rule tells you the real breakdown. Seventy percent of AI problems trace back to people and processes. Twenty percent to underlying technology infrastructure. Ten percent to the AI itself. The tool is the smallest variable. The strategy is everything.
And yet most SMBs run it backward. They buy the tool first. They ask what it can do. Then they retrofit their offer, their audience, and their proof into whatever the software’s dashboard allows. That sequence produces noise, not signal.
Pertama Partners’ 2026 analysis of AI failure rates found that 73% of failed AI projects lacked clear success metrics before approval. They never defined what winning looked like. They had a tool. They had no doctrine.
What Dan Kennedy Taught Me in a Hotel Ballroom
I trained under Dan Kennedy. That’s not a credential I list to sound interesting — it’s context for how I think about marketing sequence.
Kennedy ran a full-day intensive. He had a room of business owners. Every single one of them wanted to talk about tactics: what media to buy, which copy format to use, whether direct mail was dead. Kennedy didn’t care. He opened by asking one question to the room: “What’s your offer?” Silence. Then he asked, “Who’s your market?” More silence. Then: “What’s your proof?”
He hadn’t asked about tools yet. He didn’t ask about tools at all until those three questions had real answers. His doctrine was explicit: market, message, and media. In that order. Always in that order. The medium — which today we’d call the tool — comes third. It is always third.
Magnetic Marketing codifies this as the DNA of direct response. Kennedy’s core principle is clear: craft the right offer for the right audience first, then select the vehicle to deliver it. The sequence matters. Violating the sequence produces expensive noise.
Most AI marketing initiatives start at step three and work backward. That’s the problem.
The FOCUS Strategy Applied
The framework that maps cleanest onto Kennedy’s doctrine in an AI context is The FOCUS Strategy. Five elements, in sequence.
F — Foundation. Define your ideal customer profile before you buy any software. Not a persona. An actual description of the specific human being most likely to buy, stay, and refer. Without this, your AI has no targeting logic. It is optimizing toward nothing.
O — Offer. Articulate the specific result your client gets. Not features. Results. The offer must stand on its own, without a single tool supporting it. If you can’t close a sale on the phone without your AI stack, the stack won’t save you.
C — Content and Proof. Document your wins. Case studies, testimonials, before-and-after numbers. This is your ammunition. AI can amplify proof. It cannot manufacture it. Strategy-first operators collect proof before they build the automation that distributes it.
U — Unique Positioning. Where do you sit in the competitive landscape? What do you claim that no one else claims? This question has nothing to do with technology. It is answered by thinking, not by subscribing to another SaaS platform.
S — Systems. Only here, at step five, do you build or buy tools. The system serves the strategy. The system does not define it.
Most SMBs skip directly to step five and call it an AI strategy. It isn’t. It’s a subscription with no doctrine behind it.
Competence Beats Credentials — The Doctrine in Action
Here’s a corollary worth planting: proof of competence beats claims of sophistication. The operator who can show a clear track record of client results in a well-defined niche will outperform the agency with the longest technology stack and the glossiest capability deck. Every time. In every market. Without exception.
This is why tool-collectors lose to strategy-first operators. The tool-collector’s pitch is “look at what we use.” The strategy-first operator’s pitch is “look at what we’ve done.” One is a feature list. The other is a receipts stack.
Doctrine Connection — Competence beats credentials. Strategy-first operators compete on demonstrated results, not on the sophistication of their tool selection. The receipts win. The software list loses.
The Numbers Don’t Lie
Gartner’s April 2026 findings confirm what Kennedy knew about direct response in the 1980s: clarity of purpose precedes return on investment. Only 28% of AI infrastructure projects delivered promised ROI. One in five failed outright. The projects that succeeded allocated 47% of their budget to foundations — data, governance, workflow integration — before selecting models.
That’s the FOCUS Strategy in budget form. Nearly half your resources go to the first four steps. The tool is where you finish, not where you start.
McKinsey’s research on AI high-performers found organizations seeing significant returns were twice as likely to have redesigned end-to-end workflows before selecting models. Workflow design is strategy work. Model selection is procurement. Strategy comes first.
The gap between AI adoption (88% of organizations using it) and provable ROI (only 41% can demonstrate it) is not a technology gap. It is a strategy gap. Tools cannot close it. Only doctrine can.
What Strategy-First Actually Looks Like
The operator who gets this right does the following before touching a single AI tool:
They define one ideal customer with enough specificity that a stranger could pick that customer out of a crowd. They write out the offer in plain language — the result the client gets, the timeline, the price, and the guarantee. They document three to five case studies with real numbers. They write a positioning statement that differentiates them from the two closest competitors.
Then they ask: which tools would distribute this message, automate this follow-up sequence, and score these leads? The tool selection becomes obvious. The configuration becomes straightforward. The ROI becomes measurable because the success metrics were defined in step one.
That is the sequence. That is the doctrine.
Strategy Beats Tools — Where Each Wins
| Strategic Asset | What It Produces |
|---|---|
| Defined ICP | Targeting precision |
| Written offer | Conversion anchor |
| Documented proof | Trust at scale |
| Clear positioning | Competitive separation |
| Tool selection | Distribution speed |
Speed without direction is just expensive wandering. Tools accelerate the strategy. They do not replace it. Strategy-first operators > tool-collectors. The data confirms it. Kennedy confirmed it thirty years ago.
FAQ
Q: We already have AI tools deployed. Do we stop everything and go back to strategy?
No. Run a parallel track. Keep the tools running. Simultaneously, audit your ICP definition, your offer clarity, and your proof documentation. Within 60 days, reconfigure your tools around the strategy rather than around the software’s default settings. You will see a measurable lift.
Q: Isn’t strategy just a buzzword for “do more planning”?
No. Strategy in this context has a specific definition: a documented answer to three questions. Who is the ideal customer? What is the specific result they get? What proof exists that you can deliver it? If you can answer all three in writing without hesitation, you have a strategy. If you can’t, you don’t — regardless of how many tools you’re running.
Q: How long does it take to build a real strategy before deploying AI tools?
Four to six weeks if you move with discipline. That includes ICP documentation, offer articulation, proof collection, and competitive positioning. Skipping this work costs the average SMB $4.2 million per failed AI initiative, according to Pertama Partners. Six weeks of strategy work is the cheapest insurance you can buy.
Q: What if our competitors are already using AI and we’re not?
Your competitor using a tool without a strategy is not an advantage over you. Their noise does not threaten your signal. Build the strategy. Then deploy tools with precision. You will outperform them inside twelve months.
Q: Does this apply to AI-generated content, or just AI-powered advertising?
It applies to every AI application in your marketing stack. Content without positioning is filler. Advertising without a defined offer is spending. Automation without a defined audience is friction at scale. The doctrine applies universally.
Jeff Barnes is the founder of Angel Investors Network, former Hartford Steam Boiler/Munich Re Innovation Scout, US Navy nuclear-trained submariner, and a direct response marketing practitioner trained under Dan Kennedy. He holds an MBA from the University of Washington.
