AI Pushes Personalization From Guesswork to Growth
Personalization is welcome, but only when it hits the mark. A recent study conducted with AWS found 83% of consumers are open to personalized offers, yet just 44% say those offers are very relevant. That gap is exactly where growth is leaking.
Nearly half of consumers say they would switch merchants for more relevant offers. Personalization is no longer a nice-to-have - it's a competitive lever that impacts revenue, loyalty and acquisition costs.
Consumers Want Personalization - On Their Terms
Relevance beats raw discounts. When offers miss, customers tune out, even if the price looks good. That was a clear theme from a CES panel moderated by Melissa Harrison of the Consumer Technology Association.
Marketers are walking a line: give people what they want while still surprising them with something new, without crossing into "too personal." That's a brand trust issue as much as a data issue.
From Prediction to Context
Najoh Tita-Reid, growth chief officer at Mars Petcare, put it plainly: prediction matters, but anticipation wins. With half the population owning pets, relevance means knowing the breed, life stage and likely moments - even when a family brings home a new pet.
Angela Zepeda, global head of marketing at X, said AI tracks where conversations move in real time so brands can place timely ads as topics heat up. Speed and fit, without extra handoffs, is changing how teams plan and execute.
Physical and Digital Converge
Toby Espinosa, vice president of ads and growth services at DoorDash, noted that expansion beyond restaurants opened up new signals across categories. About 25% of users now order outside restaurant delivery, which means a richer view across consumers, merchants and Dashers.
He called out Smart Campaigns helping smaller merchants prospect across DoorDash and other channels. As he said, bringing physical and digital closer together reveals new pockets of growth.
Break the Silos or Pay the Tax
Allison Stransky, VP and CMO at Samsung Electronics, shared how separate business lines can act like standalone companies. Samsung's Connected Experience Center was built to see how customers actually use products, backed by a media team that tunes search and social as AI reshapes performance.
Peggy Roe, executive vice president and chief customer officer at Marriott International, said data has finally caught up to what the industry wants: anticipating needs and presenting the right mix at the right time. Trust sits at the center - long-term brands earn it through relevance and transparency.
The Personalization Playbook for Marketers
- Define the job to be done: Pick one outcome per journey moment (repeat purchase, AOV, retention, app adoption). Anything else is noise.
- Fix the data basics: Unify IDs, consent and events across web, app, email, POS and support. Standardize taxonomy. Stream data; avoid batch delays where possible.
- Model real context: Blend product, price, inventory, location, recency, session behavior, and sentiment. Capture zero- and first-party data with clear value exchange.
- Act in real time: Trigger offers within minutes, not days. Use feature stores and streaming triggers for cart, browse-abandon, back-in-stock and store proximity moments.
- Measure relevance, not volume: Track relevance rate, acceptance rate and incremental lift. Cap frequency. Rotate creatives to fight fatigue.
- Protect trust: Give a clear preference center, easy opt-outs and plain-language "why you're seeing this" notes. Avoid sensitive categories and small-audience targeting.
- Bridge channels: Tie POS and eCommerce records. Use app or SMS for curbside and in-store prompts. Equip associates with context that helps, not creeps.
- Scale smart for SMBs: Use automated campaigns with budget guardrails and proven creative templates. Report on incremental sales, not clicks.
- Align the org: Form pods across media, product, analytics and CRM with shared KPIs. Give them the same dashboard and the same budget target.
- Close the loop: Feed outcomes back into models weekly. Prune features that don't move lift. Keep a human QA step for creative and brand safety.
Practical KPIs and Cadence
- Relevance rate: % of recipients who say the offer fits (target 60%+).
- Offer acceptance rate: Click or redemption rate adjusted for holdout.
- Incremental ROAS or contribution margin: Lift vs. holdout, not last-click.
- Time-to-trigger: Median minutes from signal to offer (target under 5).
- Frequency cap: Max 2-3 offers per person per week unless intent spikes.
- Opt-in and preference completion: Make consent a growth metric.
- LTV/CAC trend: Monitor quarterly; personalization should improve both.
- Model refresh: Weekly feature updates, monthly re-train, quarterly audit.
Tools and Resources
If you're formalizing your stack, explore how managed services approach real-time recommendations and offer ranking via Amazon Personalize. For context on the industry conversation, see the Consumer Technology Association behind CES here.
Marketers building skills around AI-driven personalization can find curated paths for practice and leadership needs in the AI Learning Path for Training & Development Managers and the AI Learning Path for Business Unit Managers.
The Bottom Line
Personalization stops being a gamble when data flows across channels, models act fast and messages feel helpful, not noisy. Focus on context and consent, measure lift, and keep the feedback loop tight. That's how relevance turns into growth.
Your membership also unlocks:
