Personalization Beyond First Name at a Meal-Kit Brand
A meal-kit subscription business emails 60,000 active subscribers a weekly menu email. The only personalization is 'Hi {{contact.firstname}}' in the greeting; 4% of contacts lack a first name. Open rate is 18% and click rate 0.6%. The team wants to add the subscriber's city to the subject line and insert their last ordered meal into the hero image. Subscribers provide dietary preference (70% have a value), delivery ZIP, and a post-delivery survey. Legal is concerned because the email is sent to EU and UK subscribers, and the team once inserted a customer's full address into an email by mistake. The ESP supports contact properties, active lists, and dynamic content by list but not true real-time product recommendations. The company has a 'Dietary preference' multiple-checkbox property and a 'Last order date' property.
1. Critique the proposed first-name, city, and last-order personalization. Recommend a more reliable personalization strategy using HubSpot capabilities, including fallback handling.
Read the worked answer
The proposed personalization is uneven. First name is missing for 4% of contacts, so the greeting will break unless a fallback is set. City in the subject line can feel intrusive and may be wrong after a move; it is a weak relevance signal for a meal-kit menu. Last ordered meal may be out of season or discontinued, and inserting it into a hero image is hard to maintain. Better: use HubSpot active lists based on dietary preference and order recency, then serve dynamic content blocks by list-for example, a vegetarian block, a high-protein block, and a 'new to you' block. Use personalization tokens only where data is reliable, with a default value such as 'Hi there.' Keep the subject line focused on the menu value, not location. This raises relevance without exposing sensitive data or creating broken merge fields. It also avoids the prior incident of inserting a full address, which was a privacy and trust failure, not a personalization win.
2. How would you test whether the new personalization improves performance without confounding results?
Read the worked answer
Run a controlled test. Randomly split the eligible list into a treatment group that receives the personalized version and a holdout that receives the current generic version; keep send time, From name, and offer constant. Change one variable at a time-first test dynamic content blocks, not subject line, city, and hero all at once. Primary metric should be click-through rate to the menu or conversion, not open rate, because Apple Mail Privacy Protection inflates opens. Run for at least four weekly sends to reduce week-to-week noise, and check whether the sample is large enough to detect a meaningful lift. Segment results by dietary-preference availability and by region. If the treatment wins on clicks but not conversions, do not roll it out as a revenue win; investigate the landing experience. Keep a permanent holdout for long-term measurement. This design isolates personalization from seasonal menu changes and prevents a false positive from a single strong week.