UpTrajectory Review

Mike Pastore's column opens with a confession that doubles as a diagnosis: the AI research industrial complex is producing so many reports, surveys, and predictions that even someone paid to read them is drowning. The immediate occasion is the 2026 holiday shopping season, which various analysts are calling either the first or second AI-driven shopping cycle, with AI agents allegedly doing some or all of the purchasing depending on which vendor's whitepaper you open. Pastore's real point is that the data is contradictory, the ROI claims are inconsistent, and his own read comes less from the studies than from off-the-record conversations with marketers and martech leaders. The piece as excerpted cuts off before his actual conclusions, which is a shame, because the setup is the most honest framing of AI marketing discourse we've seen from a trade publication this year.

For a small-business operator, this framing matters more than any single statistic. You are the target of a relentless stream of AI marketing claims, and this column confirms what you probably suspected: the people generating those reports disagree with each other about adoption rates, ROI, and workforce impact. If the analysts can't align on whether AI agents are shopping for consumers this holiday season, you should be deeply skeptical of any vendor who tells you their tool is the one that will transform your business. Pastore's decision to privilege practitioner conversations over survey data is a useful model for how you should gather your own intelligence: talk to peers, not pitch decks.

What's genuinely new here is the explicit acknowledgment that AI agent-driven commerce has moved from speculative to contested. Two years ago, the question was whether AI shopping agents would ever matter. Now the argument is over how much they matter, which is a different and more commercially significant debate. If even a fraction of holiday purchases are being routed through AI agents, that changes how small businesses need to think about product discovery, pricing transparency, and where marketing dollars go. The column doesn't resolve this, but it correctly identifies it as the question that matters right now.

The second-order effect worth watching is what contradictory ROI data does to internal decision-making. If you're a small-business owner trying to justify or kill an AI tool subscription, conflicting research gives you cover for whatever you already wanted to do. That's dangerous. The more useful takeaway is Pastore's implicit point: the variance in reported outcomes is real, which means the difference between AI that works and AI that doesn't is likely execution, not the technology itself. Businesses with clear use cases and disciplined implementation are probably seeing returns; businesses buying tools because they feel behind are probably not.

Watch for Pastore's full analysis, which presumably continues beyond the excerpt provided here. In the meantime, the practical move is straightforward: before renewing any AI marketing subscription in Q1, ask the vendor for specific case studies from businesses your size in your industry, and ask two peers what they're actually using versus what they're paying for. The gap between those two answers is where your budget decision lives. The December 9 master class Pastore promotes on moving beyond prompts to workflows is worth considering, not because a webinar will solve your AI strategy, but because the framing is right: prompts are table stakes, and workflows are where value compounds or dies.

“my sense of where AI stands here in Q4 2026 is based more on the conversations I've had with marketers and leaders in the martech space, on- and off-the-record” — MarTech

Takeaway: Conflicting AI marketing data means outcomes depend on execution, not the tool; audit your AI subscriptions against actual workflows before renewing in Q1.