Michaels
genAI personalization of marketing messaging (copy) at scale, tuned to the brand voice
Craft retailer Michaels used Persado's genAI platform, tuned to its brand voice, to personalize the language of its email, SMS, and Facebook campaigns, moving from 20 to 95 percent of personalized email campaigns and gaining 25 percent email CTR and 41 percent SMS CTR (2022 case study).
Key points
- genAI personalization of marketing messaging across email, SMS, and Facebook, tuned to the brand voice.
- Persado Motivation AI platform that generates, tests, and predicts the message.
- Email personalization taken from 20 to 95 percent, +25% email CTR, +41% SMS CTR.
- Evidence B, mixed-signals status.
Objective
Send each customer the message wording most likely to make them click, at the scale of millions of customers with different preferences, without relying on manual A/B testing of copy. The goal is to re-engage and retain the customers Michaels calls Makers across its CRM channels.
The deployment
Michaels, a US craft retailer, partnered with Persado in 2019 to personalize the language of its marketing campaigns. Persado's Motivation AI platform generates wording variants, tests them on a sample to feed predictive models, then predicts the best-performing message for the following campaigns. A language model is tuned to Michaels' brand voice and applied across three channels: email, SMS, and Facebook. The share of personalized email campaigns rose from 20 to 95 percent. The reported results are a 25 percent lift in click-through rate on email campaigns and 41 percent on SMS campaigns. The Persado case study detailing these figures was published in March 2022; the partnership was covered by the retail press (Chain Store Age), which quotes Michaels' VP of CRM and repeats the same figures.
Results Proof B
Quantified Persado platform case study (personalization from 20 to 95 percent of email campaigns, +25 percent email CTR, +41 percent SMS CTR), corroborated by established retail press (Chain Store Age) that repeats the same figures and names Michaels' VP of CRM. The figures are vendor-sourced, hence B despite the press concordance.
How it works
Inferred typical approachThe internal detail is not public. Here is a proven approach that leads to the same result, to adapt to your stack.
The stack in detail
- plateforme Persado Motivation AI Platform Generates marketing language variants, tests them to feed predictive models, then predicts the best-performing wording per campaign. A language model is tuned to Michaels' brand voice and deployed across email, SMS, and Facebook.
How it runs, concretely
For ops teams-
1Language variant generation AI / platform
Persado produces wordings tuned to Michaels' brand voice for a given campaign objective.
-
2Sample testing marketing
The variants are deployed to a subset of recipients over email, SMS, or Facebook to measure their performance.
-
3Measurement and model feeding platform / data
Click-through rates per variant come back and feed the predictive models of message performance.
-
4Prediction and generalization AI
The message predicted as best-performing is applied to the following campaigns, with personalization extended to up to 95 percent of email campaigns.
-
5CRM team oversight marketing
The team chooses the campaigns and segments, and keeps control of brand voice compliance.
The click-through rate per language variant. The model learns which wordings make each segment click; if the engagement feedback (clicks, conversions) stops coming back, the prediction of the optimal message degrades.
How your customers perceive this type of use
Sourced studiesLe paradoxe est documente des deux cotes : 71% des consommateurs attendent des interactions personnalisees et 76% sont frustres quand elles manquent (McKinsey, 2021), mais 75% declarent ne pas acheter aupres d'organisations auxquelles ils ne confient pas leurs donnees (Cisco, 2024). La « creepy line » est localisee : messages recus quelques secondes apres une recherche et suivi de localisation sont les pratiques qui mettent le plus mal a l'aise (Periscope by McKinsey, 2019).
Acceptance conditions
- La confiance dans le traitement des donnees precede l'achat : 75% ne achetent pas sans elle (Cisco 2024)
- Un cadre legal protecteur rassure : 59% des consommateurs disent que des lois fortes sur la vie privee les rendent plus a l'aise pour partager des informations dans des applications IA (Cisco 2024)
- La personnalisation elle-meme est attendue quand elle est consentie : environ la moitie des consommateurs (US 55%, UK 52%) disent s'inscrire souvent ou parfois a des services personnalises (Periscope by McKinsey 2019)
Red lines
- Le message declenche quelques secondes apres une recherche ou un achat : deuxieme ou troisieme cause de malaise selon les pays (Periscope by McKinsey 2019)
- Le suivi de localisation percu comme de la surveillance : 40% de malaise en Allemagne et au Royaume-Uni (Periscope by McKinsey 2019)
- Le mesusage des donnees personnelles par l'IA, devenu la premiere inquietude des consommateurs, a 53% et en hausse (Qualtrics 2025)
Sources: McKinsey & Company 2021 · Periscope by McKinsey 2019 · Cisco 2024 · Qualtrics 2025
How to replicate
Inference, not sourcedData prerequisites
- First-party engagement data per channel (opens, clicks, conversions)
- Campaign history to frame the variants
- Usable customer segmentation
- A brand voice reference (tone, vocabulary, mandatory disclosures)
Org prerequisites
- A CRM/marketing team that owns the campaigns
- Brand voice governance to validate the generated wordings
- An accepted test-and-learn culture on the CRM channels
Possible stack
- A genAI platform for message generation and optimization (Persado-type)
- An ESP for email and an SMS sending platform
- A connection to engagement data as an optimization signal
- A language model tuned to the brand voice
The plan, step by step
- Step 1Frame the brand voice and guardrails (tone, vocabulary, legal disclosures) so the model generates in the brand register.Deliverable: A voice charter usable by the model and validated by marketing.
- Step 2Connect engagement data (clicks, conversions) per channel as the message optimization signal.Deliverable: A feedback flow of CTR per variant.
- Step 3Launch language variant generation on a sample of email campaigns and measure the lift against a control.Deliverable: A first measured lift, control versus AI variant.
- Step 4Extend personalization to SMS and social campaigns once the email lift is confirmed.Deliverable: Multichannel personalization in production.
- Step 5Generalize personalization to the majority of campaigns and monitor brand voice drift.Deliverable: A high share of personalized campaigns and ongoing quality control.
First step: Pick the highest-volume channel (email) and run language variant generation on a subset of campaigns, with a control group, to establish a measurable CTR lift before any generalization.
The same pattern, proven in other industries
Meta
AI feed and video recommendation (large-scale models / LLMs)
Netflix
content recommendation engine
Nike
personalized exclusive offers and access based on member engagement and history
Sources
- S1 How Michaels Transformed Its Personalization Strategy: Unlocking Greater Loyalty & Engagement Interested party archive pending
- S2 Exclusive: Michaels improves personalization via email program Established press archive pending
An error, newer info, a source?
This page lives on its accuracy. If a figure has moved, if the deployment has changed, or if you have a higher-quality source, tell us. Every sourced correction is verified before publication.