Efficiency isn't a relationship strategy. Most CMOs are confusing the two. Martech Futurist | May 15, 2026
The most consequential insight in this week's research comes wrapped in a single line from Forrester: "Efficiency isn't a relationship strategy." It was written about retail banking — but it could just as easily be the headline for nearly every AI-driven marketing transformation happening right now. Across new research from HBR, Gartner, and Forrester this week, the same pattern keeps surfacing: marketing leaders are deploying AI to make existing workflows cheaper and faster, and in the process are quietly accelerating commoditization, transparency loss, and customer detachment.
The deeper story is about control. CMOs historically owned three things: who saw the message, how the message landed, and how the customer relationship was kept. AI is eroding all three at once. AI shoppers don't respond to the persuasion tactics that have powered ecommerce for two decades. AI-mediated paid media is concentrating spend in three platforms and making advertising harder to defend internally. AI assistants are inserting themselves between brands and customers, capturing the advisory layer while the brand gets pushed back to the role of back-end execution. The companies pulling ahead aren't the ones moving the fastest — they're the ones sequencing the work correctly and refusing to confuse activity with progress.
Three themes from this week's research that CMOs should sit with:
You're now marketing to two distinct buyers. HBR's research is unambiguous: scarcity, countdown timers, vouchers, and bundles do not reliably influence AI shopping agents, and can actually reduce selection. The buyer you've optimized for over the last 25 years is no longer the only buyer making the next purchase. AI models are a distinct segment with a different decision logic — and most ecommerce stacks are still pointed at human psychology.
Your paid media is becoming a black box. Gartner's diagnosis from London is that AI is concentrating spend in a handful of platforms, optimizing for ad-seller margin rather than advertiser outcomes, and asking marketers to share proprietary data with algorithms they can neither audit nor influence. The result: paid media that's harder to govern internally and harder to defend at the board level.
Efficiency-led AI accelerates commoditization. Forrester's banking research is the loudest signal. When AI is used to deflect calls, automate journeys, and remove cost, it doesn't deepen customer relationships — it trains customers to treat the brand as a commodity and to switch to whichever third-party AI assistant gets good enough first. "Agentifying" a hollow process doesn't make it valuable. It just makes the hollowness scale faster.
Three strategic decisions CMOs need to make now:
Re-segment around AI buyers. If a meaningful share of your purchase volume now flows through AI agents — and it does — you need a testing infrastructure that continuously measures how different models respond to different signals, and an ecommerce stack that competes on what AI agents actually weight (price, structured product data, authentic reviews, competitive specs) rather than what humans respond to.
Audit your advertising attribution and governance, not just your spend. If you can't explain to your board where AI-mediated paid media is going, why, and what the alternative would be, you don't have an advertising strategy. You have a budget allocation to platforms you don't control.
Stop applying AI to existing workflows. Sequence the deployment. BNY's playbook (detailed below) is the most operationally specific example of how to do this right: governed platform first, workforce fluency second, agents last. Most organizations are doing this in reverse and paying the trust tax on every agent they ship.
Traditional Marketing Doesn't Work on AI Shopping Agents
Source: Harvard Business Review | Authors: Jafar Sabbah and Oguz A. Acar | Published: May 12, 2026 Link:https://hbr.org/2026/05/research-traditional-marketing-doesnt-work-on-ai-shopping-agents
The single most important piece of marketing research this week. Sabbah and Acar ran thousands of simulated shopping rounds across four leading AI models and four common product categories, testing whether classic e-commerce persuasion tactics — scarcity, countdown timers, strike-through pricing, vouchers, bundles — influence AI agents the way they influence humans. The answer is mostly no. Only star ratings consistently increased AI choice in the expected direction. Price reliably decreased it. Every other tactic produced unstable, model-specific effects, and more advanced reasoning models appeared actively skeptical of overt persuasion attempts.
The strategic implication is bigger than ecommerce. As OpenAI, Google, and Amazon push AI agents deeper into product discovery and transaction (ChatGPT into merchant apps, Google's universal commerce protocol, Amazon's cross-retailer agents), a growing share of "shoppers" are no longer human. Marketers need to treat AI models as distinct segments, prioritize fundamentals (competitive pricing, authentic reviews, structured product data), and build a testing infrastructure that continuously measures how different agents respond as models and prompts evolve.
If your conversion strategy still assumes a human at the end of the funnel, you are optimizing for a shrinking segment. The B2C and B2B ecommerce teams that move first on AI-buyer segmentation will compound a structural advantage that's invisible in current dashboards but already showing up in the data.
AI Makes Advertising Less Transparent and Harder to Justify
Source: Gartner Newsroom | Author: Q&A with Eric Schmitt, VP Analyst | Published: May 12, 2026 Link:https://www.gartner.com/en/newsroom/press-releases/2026-05-12-ai-makes-advertising-less-transparent-and-harder-to-justify
Live from the Gartner Marketing Symposium/Xpo in London, Schmitt delivers the supply-side counterpart to the HBR research. AI is making advertising harder to justify on three fronts at once: it accelerates the concentration of spend captured by three large platforms; it moves decision-making about targeting, placement, and creative delivery into platform algorithms that are optimized for ad-seller revenue rather than advertiser outcomes; and it asks marketers to share proprietary data with those same algorithms while making the resulting decisions less auditable. The economic effect: paid media becomes both more expensive and harder to govern.
Schmitt's broader point — that audiences are conflicted about AI in customer-facing communications and are aiming to spend less time on AI-powered social platforms — adds another layer. Marketers are increasing ad investment into channels where (a) they have less control, (b) their data is being used to train systems they don't own, and (c) audiences are showing signs of disengagement.
The instinct to lean harder into AI-mediated media to "keep up" is the wrong move. The right move is governance: build the internal capability to evaluate platform AI claims, audit performance independently, and rebalance toward channels and assets the brand actually controls. Otherwise, the marketing budget keeps growing while the marketing function's leverage keeps shrinking.
AI Isn't Fixing Retail Banking's Customer Growth Problem — It's Exacerbating It
Source: Forrester Blog | Author: Alyson Clarke, Principal Analyst | Published: May 14, 2026 Link:https://www.forrester.com/blogs/ai-isnt-fixing-retail-bankings-customer-growth-problem-its-exacerbating-it/
The piece of the week. Clarke's argument is structural and applies to nearly every consumer category, not just banking. Banks are spending heavily on AI in 2026 with productivity-led strategies — call deflection, automation, cost takeout — and seeing the predictable result: no improvement in customer loyalty or primacy, and rising commoditization. The mechanism is simple: when AI delivers convenience without connection, it trains customers to treat the relationship as transactional, then to outsource the relationship to a third-party AI assistant the moment one is good enough.
The threat is not AI assistants per se. The threat is relationship displacement. If customers form the habit of asking an external assistant first, that assistant becomes the interface and the brand becomes the back end — competing on balance-sheet economics alone. "We have an assistant, too" is not a strategy. The win condition is to use AI deliberately to make customers feel guided, understood, and safer in their decisions — measuring relationship health, trust, engagement, and wallet share rather than call deflection and cost-per-interaction.
This is the most important reframe in marketing right now. CMOs whose AI strategy is primarily productivity-led are accelerating their own commoditization. CMOs who use AI to deepen guidance, advice, and trust are building the only competitive moat that will survive the agentic transition.
BNY Built Its Digital Workforce Backward — And It's Working
Source: Forrester Blog | Author: Brian Hopkins, VP, Emerging Tech Portfolio | Published: May 12, 2026 Link:https://www.forrester.com/blogs/bny-built-its-digital-workforce-backward-and-its-working/
The operational counterweight to every other piece in this edition. BNY — the first global systemically important bank to publicly deploy "digital employees" (AI agents with logins, email addresses, and human managers) — built its agentic capability in the order most organizations are doing in reverse: platform first, people second, automation last. The result: 130+ digital employees in production (up from 70 a year ago), 18% adjusted CAGR for 2025, 21% pre-tax income growth, and a 13% dividend bump.
The sequence matters. In 2023, BNY launched Eliza, a model-agnostic platform unifying Anthropic, Google, OpenAI, and other providers under governed access. Today 97% of the bank — ~50,000 people — works on it, with 160+ production AI solutions and 200%+ year-over-year growth. In 2025, the bank put more than 1,400 employees through 40-hour live bootcamps and delivered 170,000+ hours of AI learning. More than one in three BNY employees has built a custom agent. Only after both the platform and the workforce were in place did the bank deploy autonomous agents into core operations — payment validation, code repair, fully auditable, scoped, and monitored.
Hopkins's framing is unusually direct: "If your agentic strategy starts with the agents, you are already behind." Marketing leaders racing to deploy autonomous agents on top of fragmented data, ungoverned access, and untrained teams are about to learn an expensive lesson. The trust tax is real, it compounds, and it eats ROI. BNY's case is the clearest available proof that disciplined sequencing beats speed.
Key Takeaways
1. The buyer base has bifurcated. A meaningful share of purchase decisions is now made or heavily influenced by AI agents that don't respond to the persuasion logic powering 20+ years of ecommerce. CMOs need a deliberate AI-buyer segmentation strategy — different content, different testing infrastructure, different success metrics — running in parallel with their human-buyer strategy. The organizations treating this as one problem are losing both.
2. The advertising control surface is collapsing. Three platforms are absorbing more share, AI is moving decisions inside opaque algorithms, and the proprietary data marketers share to feed those algorithms is being used to optimize ad-seller margin, not advertiser ROI. The CMO who can't independently audit and defend their paid media performance won't be able to defend their budget in 2027.
3. Efficiency-led AI is a commoditization machine. If your AI use cases are dominantly about cost takeout, deflection, and faster execution of existing workflows, your AI strategy is quietly working against your loyalty goals. The strategic question isn't "Where can we apply AI?" — it's "Where does AI deepen the relationship?" The latter is a much smaller, much more valuable list.
4. Sequencing beats speed. BNY's case is the most operationally rigorous proof point in the market right now that disciplined order — governed platform, then workforce fluency, then autonomous agents — outperforms reverse-order deployment. The trust tax on agents bolted onto unready foundations is paid retroactively, every time, and it's the reason most pilots stall. The CMOs winning in 2027 are the ones who finished the unglamorous platform and enablement work in 2026.