BFCM produces more data in five days than most brands see in a month, and most of it gets used for one slide in December. That is a waste. Read properly, the weekend tells you which channels can scale, which offers your customers actually respond to, and how many of the people you paid to acquire will come back. Read badly, it tells you to plan next year around a discount-driven spike you cannot repeat.
Here is how I turn the weekend into a plan, in the order I do it.
1. Separate new demand from pulled-forward demand
A big BFCM is partly sales you created and partly sales you moved. Returning customers who would have bought in December bought on Friday instead, at a discount. To see the difference, compare the eight weeks before and after BFCM to the same weeks last year. If November rose and December fell by a similar amount, a good share of the weekend was timing. If both held up, you grew.
Then split weekend revenue into new and returning customers. Returning revenue at a heavy discount is margin you gave away. New customer revenue is the part that can pay off later.
2. Judge channels on new customers, not ROAS
Platform ROAS during BFCM is the least reliable number of the year. Everyone is buying, so every channel looks efficient. The question for planning is which channels brought in new customers at an acceptable CAC while CPMs were at their peak. A channel that held its new customer CAC under that pressure can usually take more budget in a quieter month. A channel whose ROAS was mostly returning buyers probably cannot.
This needs deduplicated attribution with a new versus returning split. ThoughtMetric reports that split by channel, campaign, and product. (Disclosure: ThoughtMetric sponsors this site.) Any attribution tool that reconciles to real orders will do. Ads Manager will not.
3. Wait for the cohort before you decide on LTV
Holiday customers behave differently. Many are gift buyers who will never come back for themselves, which I covered in why gift shoppers distort your Q4 new-customer numbers. So do not set next year’s CAC targets from BFCM until you have 60 to 90 days of repeat data on that cohort. Compare their repeat rate to customers acquired in September and October. If the BFCM cohort repeats at half the rate, your break-even CAC for discount-season acquisition is roughly half too.
4. Read the offers, not just the totals
BFCM is the one week you run several offers at real volume. Look at which discount depths, bundles, and free-shipping thresholds moved conversion and which only moved margin. Product-level data matters here. If one hero SKU drove most new customers, that is your acquisition product for Q1 ads. If deep discounts on slow movers brought in buyers who never returned, do not repeat that next year.
5. Write down what broke
Tracking gaps, a checkout bug, a creative that got approved late, an email that went out at the wrong hour. Write them down the week after, while people remember. Most of next year’s BFCM improvement comes from this list, not from strategy.
6. Turn it into a plan
By late January you should have enough to set the year. Budget mix: weight toward channels that held new customer CAC under peak pressure. CAC targets: set from the cohort’s actual repeat behavior, not the weekend’s ROAS. Offers: keep the ones that drove new customers at decent margin, drop the ones that only rewarded existing buyers. Calendar: plan smaller promotional moments through the year using what you learned about which offers work. And testing: pick the one channel that looked promising but unproven over BFCM and run a proper test in Q1 when costs are lower. I described a cheap way to do that in running a cheap geo holdout.
What not to do
Do not annualize the weekend. Do not cut a channel because its BFCM ROAS lagged, since upper-funnel channels often do their work in October and get no credit in November. And do not let the post-mortem be a single number. A deck that says “revenue up 22%” teaches you nothing about what to do next. For the measurement side of the post-mortem, see tools to measure whether BFCM actually worked.
The checklist
- Compare eight weeks before and after to last year to spot pulled-forward demand.
- Rank channels on new customer CAC during the peak, from deduplicated data.
- Wait 60 to 90 days for cohort repeat data before setting CAC targets.
- Keep the offers and products that acquired customers at decent margin.
- Write the what-broke list the week after, and fix it before next October.
Leave a Reply