Leaders Insights
Analysis & insights
Daily articles in marketing, data, finance and AI, for leaders and those aiming for the top.
Marketing strategy
Only 42% of advertisers can see their creator agency fees, and that number explains the week
Digiday's briefing on hidden creator agency margins, SPUR's new AI content telemetry standard and publishers selling GEO all landed within 48 hours of each other. They are the same story: every intermediary between a brand and its audience is being asked to disclose the unit it bills on.
OpenAI's measurement gap is capping ChatGPT ad budgets
Seven months after ads went live inside ChatGPT, agencies say conversions go missing, arrive late or fail to match their own tracking, and brands are keeping spend at test level. For marketing leaders, this is a live decision about how much money moves into a channel that still grades its own homework.
Three meetings with a consultant won't move the needle: how fund selectors actually change their approved lists
Winning a slot on a platform approved list or consultant buy list is the distribution chokepoint that determines whether an asset manager grows or stagnates. This playbook sets out the specific steps, common failures, and quick wins for marketing leaders who need to move gatekeepers from aware to allocated.
The bind rate your PCW traffic never shows you
Most direct-to-consumer insurers can tell you their quote volume. Far fewer can tell you why 60-70% of those quotes never convert to a bound policy, or which funnel stage is eating the margin.
Everyone tracked clicks, Forrester tracked preference: why performance marketing lost the plot
Forrester's research found that B2B marketers were drowning in engagement data while remaining blind to whether buyers actually preferred their company. The signals that fill dashboards and justify budgets turn out to measure activity, not advantage.
Creators told MrBeast's model to go further: why equity deals are replacing flat fees
Creators are pushing brands for equity stakes instead of one-time fees, reframing themselves as co-founders rather than media placements. CMOs who treat this as a negotiating tactic will miss the structural shift underneath it.
Data & AI strategy
lag_tolerance is a budget decision now, and most teams have not made it
dbt State went generally available in September 2026, turning "rebuild everything on a schedule" into "rebuild only what changed." The savings are real, but they only land if you decide, model by model, how stale your data is allowed to be.
SB 947 makes human review of AI firing decisions a data problem
California now bars employers from firing or disciplining workers on the say-so of an automated system alone, and from July 2027 a human has to corroborate the output in writing. The compliance work lands on data teams, who have to prove which model touched a decision, what personal data fed it, and who reviewed it.
How did Telefónica build a churn model that actually moved retention numbers?
Telefónica's data teams spent years accumulating subscriber signals before their churn models started producing revenue-grade predictions. The mechanics of what they built, and where other telcos consistently fall short, carry direct lessons for any CDO running a retention program in 2026.
AI slop detectors make your training data worse before they make it better
Filtering AI-generated text from training datasets sounds like straightforward hygiene, but a recent experiment shows the cure can degrade model performance more than the contamination itself. CDOs who treat this as a tooling problem will miss the governance question underneath it.
The shared data infrastructure problem that quietly breaks every cross-agency program
When two agencies cannot agree on what a "household" means, no amount of technology fixes the mismatch. This article explains how interoperability standards actually work in government data environments and where the traps are for CDOs who underestimate the governance layer.
Everyone assumes the flywheel spins itself: Amazon's data advantage took twelve years of deliberate engineering to compound
Amazon's data flywheel is cited constantly as proof that more data automatically produces better outcomes. The reality is that the compounding happened because of specific architectural decisions, feedback loop designs, and organizational choices made over more than a decade.
Finance strategy
AI debt passed $450bn and now your funding costs more
The Bank of England's Financial Policy Committee published the record of its 25 September meeting and flagged debt-financed AI investment, opacity and "circular arrangements" as a growing threat to funding markets, while Reuters reported that AI borrowers in the riskier parts of US credit are now paying up for money. For finance leaders, the repricing has already reached spreads, concentration limits and the terms on your next refinancing, whether or not your company has anything to do with AI.
KKR flags AI concentration risk: what the credit binge means for bank NPL ratios now
KKR's warning about overexposure to AI-related borrowing is not just a private credit concern. For bank CFOs managing credit portfolios, it reopens a familiar and uncomfortable set of questions about NPL ratios, coverage adequacy, and whether today's cost of risk accurately prices tomorrow's defaults.
$100/kWh and falling: the battery cost math every automotive CFO must own
The $100 per kilowatt-hour threshold has long been treated as the point at which electric vehicles become cost-competitive with internal combustion equivalents on a per-unit basis. But in a year when consumer confidence has hit a 12-year low and EV demand is softening across major markets, CFOs need to understand exactly what that number means for their break-even models, and where it breaks down.
TotalEnergies is flooding shareholders with cash: is the Iran war windfall a repeatable playbook?
TotalEnergies has sharply increased both its buyback programme and its dividend as oil prices surge on the back of the Iran war, generating exceptional cash flows. The case forces a precise question: when a geopolitical shock fattens your balance sheet, how much do you return and how do you structure it?
Pharma partnership terms that protect your pipeline when the science goes sideways
Milestone-based deals are how most biotech companies survive long enough to see their drug approved, but poorly structured agreements can leave a CFO holding the downside while the partner captures the upside. This playbook shows how to build deal terms that align incentives across a decade-long development arc.
Consumer sentiment's four-month low and what it means for every hurdle rate on your desk
Consumer sentiment has dropped to a four-month low on persistent inflation, and the 10-year Treasury yield is sitting at a 19-year high. For CFOs, the combined signal is straightforward: the discount rate assumptions baked into your capital decisions six months ago are probably wrong.
AI & LLMs in practice
OpenAI dots put always-on agents on a model AISI flagged
OpenAI launched dots, always-on ChatGPT agents with their own cloud computers, one day after shelving the model that was meant to succeed the one dots run on. Anyone deciding whether to switch agents on for a team now has to weigh a vendor's own safety veto against a UK government lab's finding about the model already in production.
FTC probe of OpenAI, Anthropic and METR puts safety claims on trial
The Federal Trade Commission confirmed a sweeping investigation into OpenAI, Anthropic and the independent evaluator METR one day after those labs signed a voluntary safety pledge at the White House. Anyone shipping AI agents now has to assume their safety claims, vendor system cards and third-party evaluations are discoverable evidence, not marketing.
Anthropic warns its own models might resist shutdown, and the IPO pitch is where it said so
Anthropic's IPO documentation warns that its own Claude models could resist human attempts to shut them down and cause catastrophic harm. This week's developments show every major frontier lab shipping power faster than governance can respond, and the gap is no longer theoretical.
Read how Duke Energy cut turbine failures using sensor AI
Duke Energy built one of the most operationally consequential AI deployments in U.S. generation by wiring sensor data into machine learning models that flag failures weeks before they happen. The mechanics of what they did, and what transfers to your assets, are worth examining closely.
Shopify wired AI agents into checkout, and that changes how you catch errors before money moves
Shopify's expansion of WebMCP support to checkout lets browser-based AI agents complete purchases on a buyer's behalf. That convenience compresses the window between a model's confident mistake and a real financial transaction.
Did Blue Cross Blue Shield just prove that hospital AI raises costs by $942M?
Blue Cross Blue Shield's claim that hospital AI tools added $942M in spending over two years has handed every CFO a reason to pause. The number deserves scrutiny before it reshapes your capital allocation decisions.