https://alli.pmg.com/blog/ Analysis from PMG on data, AI, and growth. en-us https://alli.pmg.com/blog/openai-ads-early-results/ https://alli.pmg.com/blog/openai-ads-early-results/ Thu, 12 Nov 2026 09:00:00 -0600 Seven of eight brands earned more per OpenAI dollar than on their average media dollar, at very small spend. What PMG measured, and how far it goes. https://alli.pmg.com/blog/synthetic-research-when-to-add-real-people/ https://alli.pmg.com/blog/synthetic-research-when-to-add-real-people/ Tue, 10 Nov 2026 09:00:00 -0600 AI survey estimates landed about four points from real results on average. When that's close enough, and when to add a small real sample. https://alli.pmg.com/blog/model-drift-bug/ https://alli.pmg.com/blog/model-drift-bug/ Sat, 07 Nov 2026 09:00:00 -0600 Marketing models lose accuracy quietly as platforms, signals, and customers change. Why drift goes unreported, where to look for it, and five questions to ask. https://alli.pmg.com/blog/ai-search-visibility/ https://alli.pmg.com/blog/ai-search-visibility/ Fri, 06 Nov 2026 09:00:00 -0600 Why one AI answer tells you little about your brand, how mentions differ from citations, and how to measure AI visibility across repeated runs. https://alli.pmg.com/blog/bayesian-thinking-without-the-equation/ https://alli.pmg.com/blog/bayesian-thinking-without-the-equation/ Thu, 05 Nov 2026 09:00:00 -0600 Bayesian thinking without the math: start with what you know, weigh new evidence, and adjust. Worked through one brand search question. https://alli.pmg.com/blog/agent-evaluation-failures/ https://alli.pmg.com/blog/agent-evaluation-failures/ Tue, 03 Nov 2026 09:00:00 -0600 AI agents that pass their tests still fail in production. Four ways agent evals go wrong, what to test in a marketing agent, and five questions before launch. https://alli.pmg.com/blog/why-teams-ignore-ai-insights/ https://alli.pmg.com/blog/why-teams-ignore-ai-insights/ Sat, 31 Oct 2026 09:00:00 -0600 AI insights often need a price, segment, or budget change the person who sees them can't make. What the 2026 data shows, and a 90-day plan. https://alli.pmg.com/blog/mmm-range-matters/ https://alli.pmg.com/blog/mmm-range-matters/ Thu, 29 Oct 2026 09:00:00 -0600 A marketing mix model's ROAS is one plausible answer among several. How wide MMM uncertainty runs, how to decide with a range, and five questions to ask. https://alli.pmg.com/blog/rag-chunking-problem/ https://alli.pmg.com/blog/rag-chunking-problem/ Mon, 26 Oct 2026 09:00:00 -0600 Most wrong answers from document assistants start with retrieval. How chunking and search cause them, what the evidence shows, and five checks to run. https://alli.pmg.com/blog/marketing-agent-circuit-breakers/ https://alli.pmg.com/blog/marketing-agent-circuit-breakers/ Thu, 22 Oct 2026 09:00:00 -0600 Marketing agents are being built faster than their limits. Four circuit-breakers, the evidence, and five questions to ask before an agent gets a budget. https://alli.pmg.com/blog/cannibalization-loop/ https://alli.pmg.com/blog/cannibalization-loop/ Mon, 19 Oct 2026 09:00:00 -0600 Bidding, affiliate, and CRM systems can each claim the same sale. What incrementality tests show about branded search and affiliate, and how to check yours. https://alli.pmg.com/blog/marketing-finance-trust/ https://alli.pmg.com/blog/marketing-finance-trust/ Fri, 16 Oct 2026 09:00:00 -0600 Marketing and finance expect similar payback but disagree on proof. What 2026 research shows about definitions, timing, and uncertainty, and what to fix. https://alli.pmg.com/blog/closed-loop-marketing-telemetry/ https://alli.pmg.com/blog/closed-loop-marketing-telemetry/ Tue, 13 Oct 2026 09:00:00 -0600 Ad platforms optimize on checkout value, but returns arrive later. How return-adjusted signals change bidding, with 2026 evidence and a checklist.