UpTrajectory Review

Scott Gillum at MarTech flags a pattern that should make every small-business operator pause before chasing the latest AI productivity promise. Marketing is now in its third major wave of bringing capabilities in-house, and the first two—post-2008 recession budget squeezes and the mid-2010s transparency backlash against programmatic media—both followed the same arc: seductive cost savings upfront, then talent drain, cultural friction, and technology debt that eroded the gains. The ANA's finding that in-house agency adoption jumped from 42% to 78% between roughly 2010 and 2018 sounds like vindication until you notice what happened next: creative talent fled back to agencies for variety, and CMOs discovered that competitive salaries alone couldn't replicate the cross-pollination of working across dozens of clients.

For small-business operators, this matters differently than it does for Fortune 500 CMOs. You don't have boardrooms demanding AI efficiency metrics, but you do face the same vendor pitch: buy these tools, bring marketing in-house, save money. The trap is identical and arguably more dangerous. A large company can absorb a failed in-house experiment and still pay McKinsey to write a post-mortem. A small operator who hires a 'marketing generalist' to run AI tools, then discovers that person can't strategize or that the output plateaued after month three, has burned runway and opportunity cost that doesn't appear on any spreadsheet. The hidden costs Gillum identifies—talent, culture, technology—hit harder when you have no bench to rotate in.

What's genuinely new this time is the speed and scale AI promises, which is also what makes the warning more urgent. Previous waves moved at human hiring cycles; AI tools can be procured in a credit-card transaction and deployed over a weekend. That compresses the 'honeymoon period' and accelerates the moment when someone asks whether the output is actually better, not just faster and cheaper. Gillum's framing is sharp here: the test comes when boards ask if efficiency produced better results. For small operators, substitute 'when you ask yourself why revenue didn't move.' We're skeptical of the assumption that AI changes the fundamental equation of in-house versus outsourced marketing; it may just compress the cycle from optimism to disillusionment.

The under-reported casualty in this pattern is creative quality itself, especially in B2B contexts that small businesses inhabit. Gillum notes that agency talent got bored in-house, and brands couldn't replicate the variety that sharpens creative instincts. AI doesn't solve this; it amplifies the risk. Tools that generate volume without variety produce a different kind of boredom—market fatigue among audiences who see the same prompts, the same structures, the same uncanny-valley copy across competitors. The second-order effect is a leveling effect: everyone uses the same AI tools, so differentiation collapses precisely when the tools promise to help you stand out. Meanwhile, the operators who kept relationships with human strategists—freelancers, boutique shops, specialized agencies—gain relative advantage because their output isn't homogenized.

What to watch: whether the AI in-housing wave produces a counter-wave of 'AI cleanup' specialists—editors, strategists, and creative directors who rescue brands from their own tool stacks. Also watch whether platform vendors shift pricing to capture the value they currently leave on the table; today's cheap AI marketing tools are loss-leaders for market share, not sustainable price points. For operators, the actionable move is to run a six-month controlled test before restructuring any marketing function: keep existing relationships running parallel to AI-in-house experiments, measure revenue and brand metrics not just output volume, and build in a deliberate 'boredom audit' at month four when the novelty fades. The history Gillum cites isn't destiny, but it is expensive tuition that someone else already paid.

The deeper question this piece raises but doesn't fully explore: what is marketing actually for in a small business? If it's lead generation with measurable attribution, AI tools may genuinely suffice. If it's differentiation in crowded markets, trust-building with skeptical buyers, or narrative creation that justifies premium pricing, then efficiency and effectiveness diverge sharply. The operators who thrive will be those who know which game they're playing before they pick their tools.

“The promise of efficiency can outpace the reality of building and managing the capability yourself.” — MarTech

Takeaway: Run parallel AI and human-led marketing for six months, measuring revenue not output, before restructuring your team.