The holiday season is when retail media networks either prove their maturity or expose their gaps. Demand becomes so intense that your normal playbook stops working.
Traffic surges. Your platform's usual monitoring cadence, built for a quiet Tuesday in March, can't keep pace with what's happening hour by hour.
It's both your biggest opportunity and your harshest audit. Get it right and you capture a meaningful share of the year's gross merchandise value (GMV). Get it wrong and you won't just lose revenue for a week, you’ll miss your yearly targets and damage hard-to-build trust with sellers in the process.
I've lived this cycle from inside a retail media network at Cdiscount, where Black Friday was the year's real exam. Everything you postponed, underfunded or half-automated during the year shows up in a single, unforgiving week.
Here's what I'd tell any team preparing for peak season today.
Start peak-season preparation early
Serious work begins about two months before Black Friday. The first month goes to the data analysts, who assess the period's potential and calibrate it against the retailer's own GMV forecast.
Then sales works with the sales house and marketing to plan communications and budget milestones, so sellers have enough budget, at the right time, to capture as much GMV as possible.
The peak paradox: More spend, less relative investment
Ad spend goes up during peak, but the relative investment rate (opens in a new tab) goes down. Why? Organic traffic is enormous at the start of the season, and seller budgets don't scale at the same rate.
Retailers also build their strongest offers around first-party (1P) inventory, so 1P often captures a larger share of GMV.
The uncomfortable realization: enormous traffic, yet little revenue captured. That gap is why advance planning matters.
Holiday readiness priorities for retail media networks
1. Build the calendar backwards from Black Friday
My six checkpoints:
- Retrieve the forecast from the core business teams, by category.
- Assess your existing sellers, using last year and recent months to estimate reachable GMV and budgets.
- Plan communications with the business teams: what sellers need to do, and by when.
- Schedule seller meetings well in advance.
- Verify that budgets and timelines are adjusted in the platform as the weeks go by. (Almost the most important checkpoint.)
- Brief operations on every seller's objectives.
Teams most often skip No. 5. Budgets should ramp up gradually; check only in the final week and the peak has passed.
Pacing that is too slow for the budget is a related risk: you're forced to spend, far less optimally, killing performance as ACoS begins to soar. That's why demand and traffic forecasting matter.
2. Close your commercial commitments early
Have holiday budgets committed by November 1, or November 10 at the latest, depending on the site. After that, it's over: December declines sharply from around December 10, and by December 20 almost nothing is left.
Unless inventory is plentiful, concentrate budget where demand is heating up.
3. Raise floors, but with discipline and transparency
Yes, floors should change going into peak. Conversion rates are extremely good on certain days, because people need gifts under the tree.
If conversion rises while CPC stays flat for an equivalent basket, targeting has improved significantly and, when offers outnumber demand, raising floors makes sense. With that in mind, here are some rules to keep in mind:
- Be reasonable and pre-calculate. Damage seller costs and it's a deal you only do once.
- Inform everyone, including your own teams. I've seen floors raised without sellers or operations being told. ACoS (opens in a new tab) drifts, which is very bad for the relationship.
- Warn manual CPC sellers well in advance. If the floor moves and the seller hasn't seen it, their bid sits below the floor and you miss revenue.
4. Protect the seller relationship
The most expensive mistake I see is believing you can keep the same investment rate on Black Friday as the rest of the year, or that sellers will accept CPCs disconnected from their conversion rate or ACoS.
It can work temporarily, but once trust is lost, growing the relationship and seller budgets becomes almost impossible.
For many sellers, this is also the period where they make their entire year, with no bandwidth to listen. Do the heavy lifting before the peak and take as much off your sellers’ plates as possible.
5. Optimize the SKU mix, not just the peak list
You don't need a special holiday SKU playbook. Recommending SKUs is a daily job: spot those gaining momentum and make sure you're not short.
During peak, sellers are tempted to say, "I want three SKUs in my campaign." That constrains spend and ignores halo effects: advertising one SKU doesn't stop the shopper moving to another from the same brand.
6. Get close to the merchandising calendar
Ideally I'd want the deal calendar 15 days ahead, but deals are often decided the same day. So I believe in automated systems that show how a SKU is promoted upstream and right now, letting you act when it exceeds its ceiling.
Expect the organic mix to shift too: attractive 1P pricing changes available keywords, and a major promotion can leave very specific searches empty.
What not to touch during peak
Freeze the code two to three weeks before. This is the "last safe change" we don't talk about enough, on the ads side and the site. You may deploy to stabilize the site, but new code can disrupt the business.
Don't increase ad surface. Reporting behaves differently and a new format creates risk. Adding one now usually means something was mismanaged beforehand.
Tighten relevance, don't relax it. Issues reported daily are reported ten times more often during peak. Be rigorous with every change.
How to run peak week
Monitor hour by hour, starting two or three days before Black Friday. When I was at Cdiscount, we tracked forecast revenue against actual spend, and asked: Is CTR at the expected level? Is ACoS increasing sharply? Are we losing impressions?
The goal: Keep commercial performance on track, while eliminating technical problems.
Set an operating cadence. For larger accounts, expect two or three interventions a day when things are intense, especially when the forecast is off. Triggers are deviations in spend, CTR and performance.
Turn on peak-only alerts, including:
- Hourly pacing monitoring.
- Technical indicators like impression volume, unserved queries and budgets stopped mid-day.
- Convert weekly macro indicators into daily macro indicators.
Keep floor decisions simple. Some systems can't show a floor increase's impact on paid CPCs for 24 hours, and even with real-time data, long-term effects are hard to read. Adjust floors once a day at most, until you have powerful AI-driven algorithms. If my ACoS target is 10% and I push floors until campaigns run far above it, CPCs fall by month end. Did I really win?
Judge the peak on trajectory, not revenue. The numbers almost don't matter. The trajectory does. Are we doing better than last year? Can sellers hold last year's ACoS? Are last year's gaps covered? A record day followed by nothing, or ads that make no sense to shoppers, means you have lost.
Your holiday readiness checklist
- Eight to six weeks out: Get the forecast, size sellers, plan communications, book meetings.
- Five to three weeks out: Commit budgets, announce floor changes, check ramp-up, brief operations, freeze code.
- Peak week: Monitor hourly, check in two or three times daily, keep floors simple, add no formats.
- January: List friction points, measure forecast accuracy, map demand gaps, audit tooling.
A word on AI and agentic commerce
2026 is a key year for agentic surfaces, and retailers can already use AI for file and opportunity analysis.
But today, LLMs from Google and OpenAI aren't great at comparing prices and surfacing good deals, which is what Black Friday shoppers want.
They help with discovery, a broader behavior shift rather than a peak issue. We'll see in three months; I may be wrong.
Where to go from here
Holiday readiness comes down to a few habits: prepare early, communicate clearly, protect seller trust, resist change when stability matters most and judge yourself on trajectory.
If you want to go further on building a program that can handle peak and everything else, explore my blog series:
