Can an AI agent plan your eCommerce marketing calendar for BFCM?

From May to September 2026, Mully sent zero marketing email campaigns. There was no plan and no calendar, just a lot of good intentions. So in late October we built an agent whose only job is deciding what we promote, when, to whom, on which channels, with what offer and one message.

This guide covers what that agent read, what it found, the calendar it produced and the problems it caught before they became launch-day surprises.

Can an AI agent really plan a holiday marketing calendar?

Yes, as long as it plans from your data and not from generic BFCM checklists. In one run, our agent read six months of Mully data from Klaviyo, Shopify, Meta Ads, Google Ads and PostHog. It then studied five brands we admire in our space: Huckberry, Uncrate, Bespoke Post, Malbon and Bad Birdie.

The output was a calendar of 17 campaigns from late October to early February. Each row has dates, an offer, an audience, channels, one key message, the assets needed, who builds it, a brief due date and a KPI.

Campaign type	Count
Editorial (weekly letter, destinations, year-end list, PGA Show)	6
Edits and gift guides	4
Sales (members early access, Black Friday, Cyber Monday)	3
Holiday deadlines (last shipping day, last-minute gift)	2
Test slots	2

Only 3 of the 17 are sales. That was a deliberate choice, and the next section explains why.

What did the agent learn from our own data?

It found three things we would have argued about for weeks without numbers:

1. A gift beat a discount. Our June campaign that gave away a rangefinder was our best Meta campaign of the year: $4,941 spent, 95 purchases, ROAS 3.78. We checked that against the Meta Ads account while writing this.
2. Past buyers respond. On our October Mystery Closeout, past buyers converted about 5 times better than people who had never bought.
3. Our BFCM is small and specific. Cyber Monday 2025 brought 21 new customers and Black Friday 12. Planning for a giant spike would have been fiction.

The benchmark brands ran offers from 20% to 60% off. We cannot win on price against that, so the plan wins on curation: picks and stories in every send, gift guides before discounts.

That lines up with the market. Klaviyo's BFCM 2025 recap found consumer spending rose 11% year over year while discounts fell 10%, and revenue from repeat customers grew 13.5%, faster than revenue from new buyers (Klaviyo). Your category may differ. In Klaviyo's sporting goods benchmark, first-time buyers drove 64.7% of BFCM GMV (Klaviyo benchmarks), which is why the agent reads your own buyer split before choosing an audience.

What rules should the agent plan under?

Write the rules down before the agent plans anything, or it will plan the sale you cannot afford. Ours:

• Drafts only. It never sends, schedules, creates discount codes, changes prices or spends money.
• Offer cap. 10% off at most, or a free product worth 10% or less of the order.
• Brief 14 days out. Every campaign gets a brief two weeks before launch.
• One approval. The owner approves each launch once, 5 days before it goes out.

Under those rules the plan landed on 10% off one-time orders for Black Friday, applied automatically with no code and excluding subscriptions, and a free gift with orders of $150 or more on Cyber Monday, with the gift's unit cost capped at $15. Send load is about 2 emails a week, rising to 5 in Black Friday week, with text messages only on four deadline moments and only to people who opted in.

How far ahead should you start for Black Friday?

Earlier than the week of. Adobe reported that Black Friday 2025 online spending grew 9.1% to $11.8 billion, faster than Cyber Monday's 7.1% (Adobe), and its season recap credited consumers embracing earlier deals (Adobe holiday recap). Shopify merchants did a record $14.6 billion over the BFCM weekend (Shopify). Everyone is in the inbox at once.

With 14-day briefs, our dates work backward like this:

Campaign	Launch	Brief due
Gift Guide	Nov 17	Nov 3
Members early access	Nov 23	Nov 9
Black Friday Edit	Nov 27	Nov 13
Cyber Monday gift	Nov 30	Nov 16
Last call for Christmas delivery	Dec 14	Nov 30

If you are reading this in November and have nothing briefed, cut to fewer campaigns rather than compressing the lead time on all of them.

What did the agent flag that a person might miss?

This was the most useful part. Before anything was built, the plan listed five risks:

1. Stock. Every shop item showed zero or less on hand in Shopify, and no unit costs were loaded. That blocked the Cyber Monday gift until stock and cost were confirmed.
2. Old offers still live. Last year's 50% BFCM code, a 30% VIP code and several 15% to 25% codes were still active. Any of them would beat the new 10% Black Friday offer.
3. A campaign depending on something that does not exist. The last-minute gift campaign needed a gift card and a prepaid gift option we had not built.
4. Billing collision. Subscription renewals might land in early December, so big sends should avoid that date.
5. Stale drafts. Two old campaign drafts were sitting on hold and should be cut or folded in.

None of these are clever. All of them are the kind of thing that breaks a launch at 9pm the night before.

How do you measure a marketing calendar agent?

Measure whether campaigns go out on time, not whether the plan looks good. Our KPI is the on-time launch rate, with a target of 90%, tracked weekly against the calendar. Every Monday the agent checks which briefs are due, which are late and which launches are at risk. Revenue per campaign is measured too, but the first job is simply shipping the plan.