PlaybooksMost marketing teams spent 2023 experimenting with chatbots. A December 5, 2023 McKinsey article by Lisa Harkness, Kelsey Robinson, Eli Stein, and Winnie Wu tries to move the conversation from novelty to a sequenced plan. The authors argue that generative AI is what finally makes hyperpersonalization at scale feel plausible — not as a slogan, but as faster content, tighter targeting, and quicker learning loops.
The economic claim is large and should be handled carefully because it comes from McKinsey’s own earlier research. A June 14, 2023 McKinsey report, cited here, estimated that generative AI could add up to $4.4 trillion in annual global productivity, with marketing and sales among the functions that could capture most of that value. For marketing alone, the authors say productivity could rise by 5% to 15% of total marketing spend.
The useful part of the piece is the ladder of ambition. First come off-the-shelf pilots inside existing workflows: draft copy and images faster, personalize campaigns, and mine customer feedback. The article cites Michaels Stores moving from personalizing 20% of email campaigns to 95%, with SMS click-through rates up 41% and email up 25% (via a Persado case write-up dated March 30, 2022). Stitch Fix and Instacart appear as examples of using models to interpret unstructured feedback or to generate recipes and lists.
The second rung is customization: fine-tuning models on a company’s own brand guidelines and historical creative. A European telecommunications company is said to have moved from four macrosegments to 150, using GPT-4 and DALL·E with human review, and to have seen a 40% lift in response rates and a 25% reduction in deployment costs.
The third rung is transforming the function so that most tasks start with a model draft, under guardrails for privacy, copyright, hallucinations, and bias. The authors recommend an accountable leader and a technology oversight board, and they advise picking two or three priority use cases rather than launching everything at once. Their suggested timeline is a six-week pilot roadmap, a 90-day “win room,” and a six-month plan to measure impact.
Why it matters
Keep the operating sequence: pilot on two or three workflows you already run, then train a model on your own brand data, then redesign the handoffs between marketing, legal, service, and product. Those case numbers are vendor-adjacent illustrations, not a guarantee. Copy the governance habit immediately: human review for anything a customer will see, and a short list of topics a model may not touch.
Read the original: How generative AI can boost consumer marketing (Lisa Harkness, Kelsey Robinson, Eli Stein, and Winnie Wu, McKinsey, December 5, 2023).
Format: Curated note. Original analysis based on the cited source; we do not reproduce the report.
Keep reading
More in Playbooks →
PlaybooksThe 30-Day Marketing Measurement Audit
You do not need a new attribution vendor to know whether marketing is working. You need four weeks of honesty: an inventory of spend, a single business outcome, one incrementality test, and written rules for moving money. This playbook is that sequence.
PlaybooksSeasonal hires are regular employees as far as payroll taxes are concerned
IRS Tax Tip 2026-53, posted July 1, 2026, said seasonal and part-time workers follow the same federal withholding rules as other employees and explained the Form 941 seasonal-employer box.
PlaybooksA profitable hobby is still taxable — the IRS cares how you run it
IRS Tax Tip 2026-45, posted June 2, 2026, said the line between a hobby and a business is a facts-and-circumstances test, and that 1099-K and 1099-DA payments are taxable either way.