One in Three Consumers Unsubscribe Over Holiday Email Volume — and AI Makes It Worse
One-third of consumers say holiday email volume alone triggers unsubscribes, per Validity's 2026 survey. Generative AI cheapens messaging but raises the cost of bad send decisions.
By Amara Osei
4 min read

What's News
- One-third of consumers say holiday email volume alone is enough to make them unsubscribe, per Validity's 2026 holiday consumer survey.
- Unite.AI argues AI systems optimized for clicks with few frequency constraints will send more messages even when restraint protects the relationship.
- A well-written promotion sent while a refund is unresolved is a decision failure that better content cannot fix.
- Human review should focus on ambiguous context and high-consequence actions rather than routine policy cases.
One-third of consumers say holiday email volume alone is enough to make them unsubscribe, according to Validity's 2026 holiday consumer survey. That figure lands at a moment when generative AI has made personalized messages dramatically cheaper to produce — without making it any easier to decide whether a message should be sent at all.
That distinction sits at the center of a new analysis from Unite.AI arguing that the smartest AI decision is often sending nothing. A system can generate a relevant offer, correctly predict a click, and still make the wrong call. The customer may be waiting on a refund, have an unresolved complaint, or have already bought the product under a different email address. Better writing doesn't fix any of that.
Why do clicks make a poor objective?
Sends, opens and clicks are easy to count, which makes them tempting substitutes for what a business actually wants: a customer who sees value and keeps coming back. An AI system optimized for total clicks, with few constraints on contact frequency, will find that sending more messages improves its score. Each additional send can look reasonable in isolation.
The cost shows up outside the metric being optimized. An unsubscribe is visible; the frustration that preceded it often isn't. As Unite.AI puts it: "If the system gets credit for a click today but no penalty for making a customer less likely to return next month, it has little reason to account for that tradeoff."
When does restraint count as a decision?
Consider a customer who returned an order last week and is still waiting on a refund. A marketing system sees a recent purchaser with a high likelihood of responding and queues a promotional email. From the customer's side, "the brand is asking for more money before returning what it owes."
The data is often fragmented: the return lives in the order system, the refund status in a service tool, the purchase in a marketing profile. Those records may not resolve to the same person. Connecting them changes what the system can know — but not what it will do. The decision process must also treat holding back a promotion as a valid outcome.
The analysis draws a sharp line between failure types. "A poorly written email is a content problem. A well-written email sent while the customer is waiting for their money is a decision problem. You can't solve the second by getting better at the first."
What does the loyalty trap look like?
Loyalty programs repeat the pattern. If success means more enrollments, a system learns to recruit people with a discount. Enrollment rises — but whether those customers return without another discount is a different question. Useful measurement requires connecting loyalty activity with purchase and service history over time, plus a comparison that separates the program's effect from behavior already likely. A larger membership count can't answer that on its own.
Where should humans stay in the loop?
People must own the tradeoffs between retention, margin and customer effort. A model can help evaluate them; it can't take responsibility for choosing them. A practical gating question: how hard would an action be to undo? Holding a promo for review differs from changing a price, issuing a credit or canceling a reservation.
Routine cases don't need manual review. If the policy is clear, a rule can hold the promotion while service communication continues. Human review belongs where context is ambiguous, rules conflict, or consequences exceed the system's authority. Reviewers also need more than a confidence score — they need to see which records informed the recommendation and how current those records are. Repeated overrides should become evidence, revealing missing data or an objective that rewards the wrong behavior.
How do teams test it?
Start with one decision: take a promotional send, define the conditions under which it should be held, and keep necessary service messages outside the suppression rule. Then compare the existing approach against the holdback policy, tracking repeat purchases, retention, complaints and margin over a period that fits the buying cycle. Either result is useful — a restriction may prove too broad.
The engineering burden remains. Records must resolve to the right person, recent events must reach the decision process in time, and permissions must constrain the action. As the analysis concludes, a better objective can't compensate for missing context, just as better context can't rescue the wrong objective. Sometimes the smartest decision an AI system can make is to send nothing at all.
Source: unite.ai
More from Amara Osei
Show full bio
Senior reporter covering companies, deals and strategy in entrepreneurship, startups and venture funding at Business Bearings.
2,157 articles