Why "conversion lag" can make or break your BFCM

Labor Day weekend sales driving crazy shifts across “All Other Channels” with big red arrows across everything else. This is the Q4 before Q4. Now the games begin.

It is somehow almost mid-September already, so I assume most of you are building Black Friday Cyber Monday / Q4 plans right now, if they aren’t already in full swing.

Unfortunately, I bet most of you are building benchmarks for ROAS performance that are gonna be totally wrong.

We have a tendency as growth marketers to look at our ads like a spreadsheet, avoiding basic human psychology and well-established human buying behaviors.

Here's the thing we don't talk about enough: your ads are driving revenue over time: there is a natural delay between when someone sees your ad and when they purchase.

Not all your revenue shows up in one day. It shows up over the next week, the next month, sometimes the next quarter. Yet still many of us are only making strategic choices on one-day numbers.

Human psychology is unavoidable because we are selling to humans, and just because somebody doesn't click an ad today and purchase today doesn't mean they don't end up clicking tomorrow, or the next day, or the day after that.

This delay is called conversion lag, and the closer you get to Black Friday, the longer it gets. Everyone's got a cart full of stuff. They know the Black Friday sale is coming. They're just sitting around waiting for that discount code.

This is one of the most important and expensive concepts in Q4 media buying, and I don't see enough people calculate how it works for their own businesses.

Let's fix that, shall we?

Conversion lag is the amount of time that passes from a user seeing an ad to making a purchase.

That's the whole idea. The real magic is understanding the distribution of your conversions over time.

What percent of all revenue driven by an ad has arrived in your account after one day? After 7 days, 14 days, 30 days, 60 days? Do you know this number off the top of your head?

If you think about it, "average conversion lag is 2.8 days" and "79% of purchases happen within one day of the click" are two different analyses with two different strategic actions to take.

So conversion lag is the reality of what happened. It’s that gap between awareness and conversion, manifested. It's the actual measure of how well your ads coerce your customers into purchasing, and on what length of time.

Upper-funnel channels like Meta or CTV will have a longer conversion lag, where lower-funnel channels like Google will be shorter. This is because buying intent is higher for people touching Google ads: they are searching for stuff to buy.

The length of your conversion lag is influenced by innumerable variables. It’s a function of your offer, your price point, your ad creative, the need state of your customers, how many times they've seen your ads, how funny or compelling or convincing those ads are.

Conversion lag has a material impact on your reporting.

For example, we have an apparel business whose 1.0 same-day Facebook ROAS closes at 1.35x after 30 days. That's a 35% lift.

If you turn off those ads based on their 1-day window performance, without considering how they’ll drive impact over the next 30 days, you won’t be capturing their actual performance. 

You don’t want to make judgements on ads too quickly, before they’ve had a chance to drive their full potential revenue. 

You don’t want to judge ads only based on 1-day windows if you have no idea how their performance will net out over a longer window, especially during the holidays. You could miss out on nascent winners or leave a bunch of money on the table.

For this business in particular, if they were judging Meta on day-one data only, they were looking at roughly three quarters of the revenue it actually drove. Therein lies the problem.

Let’s put this in a Black Friday context. For another apparel business we work with, Facebook's day-one to day-14 difference in lag was 25% in October, 29% in early November, and 36% to 37% during Black Friday Cyber Monday week itself.

The space between first touch and conversion widened massively as the holidays approached — yet they were still evaluating all those ads against the same benchmark.

It’s not that we want to just leave ads on longer: it’s about understanding that your one-day benchmark is going to change with seasonality throughout the year.

Conversion lag gets longer the closer you get to a sale, because customers know discounts are coming and they're waiting to purchase. You need to analyze your media buying through this lens of human behavior.

You need to evaluate your ads on their predicted total return across the conversion lag, not the one-day ROAS, because that one-day ROAS is highly influenced by promotional periods, holidays, and other natural phenomena that could confound your understand of your own ad performance.

So, how do we pull this off?

Your same-day ROAS tolerance should loosen as you approach the week of BFCM. Most media buyers run a fixed number all quarter, but this ignores the fact that their customers are human.

User behavior is changing in Q4. Why would we not change our measurement techniques?

For example, most businesses can tolerate a lower one-day ROAS on Meta in late October and early November, compared to what they usually see on BFCM.

Here’s real numbers as an example: for that second apparel business I mentioned, their ROAS on ads running in early November ultimately grew 23% once those ads hit the 60-day mark.

That meant we could tell them in the middle of November that the $2.14 million in one-day revenue they were looking at was on track to become roughly $2.59 million.

That's half a million dollars in revenue that didn't exist on screen. But it was coming, and the team wasn't planning around it.

That's what we're getting at here. You have future money coming in and you don't want to disrupt it. Patience, young grasshopper.

Let’s talk about how to actually do this.

Before we calculate, let's lay out some frameworks. One idea: break out the holiday season into phases, with different benchmarks for each phase.

Here’s an example of some phases you could use:

Each phase gets its own lag multiplier and its own targets, because lag behaves differently in each one. These are loose guidelines for our measurement. Purchasing behavior can be hard to predict, especially with something like a midterm election thrown into the mix.

For most businesses, November 1 to 17 is about building awareness. Getting in front of prospects so they know you have a Black Friday sale, know about your product, and hear your storytelling.

This period will look bad on a one-day view. It is supposed to look bad, though.

If you cut your prospecting on November 8th because it looks soft, you're turning off the demand you were planning to harvest on the 28th.

Remember, we're talking about human psychology here. It takes time for people to consider buying your stuff, especially when they know a Black Friday sale is coming.

We worked with one business where half of the revenue that came in on Black Friday on Meta last year came from touchpoints that happened before Black Friday even began. That's conversion lag!

So when you're bidding aggressively at 11:00 p.m. on Black Friday because the numbers look great, understand that a meaningful chunk of that performance probably came from your spend three weeks prior.

You are not necessarily buying new revenue in that moment. People have seen a lot of your ads, they know a sale is coming, and you're just reminding them to get across the finish line.

This is the cheapest early warning system you can have in Q4. If you know what share of your revenue is coming from same-day touchpoints, you can benchmark that against the same day last year. Remember: every year is different, different customers, different macro situation. It won’t be a perfect comparison, but it’s a start. 

For one beauty business we work with: that benchmark was around 81% of new revenue coming from same-day touchpoints for November overall and 76% on Black Friday.

That meant ad spend through the majority of November was actually better at bringing in new revenue than the ad spend on Black Friday itself.

This is how you make sure you're not overspending. If day two of your BFCM sale delivers another $500k in new customer revenue but only $250k of that is coming from the one-day window, take a closer look.

That split means you're getting paid for revenue generated earlier, not for how you're managing your buying right now. If you scale spend into that, you could be overspending on customers who don't need to see your ads anymore and are just waiting for the right time to buy.

Across our customer base, for the five days of Black Friday Cyber Monday in 2025, Black Friday took the largest spend increase but was the only day where efficiency moved backwards. Saturday and Sunday took the smallest increases and delivered the best efficiency gains.

Now, Black Friday is still the biggest day of the year. You need to be spending. But the name of the game is know exactly how much.

There’s a tendency for some brands to overspend on Black Friday itself, as it can be tough to pull back spend to an efficient number, especially with a lean team or smaller spend levels. Keep an eye on your shoulder days. Maybe you need to plan your cuts earlier so you don't end up overspending in the afternoon. It is a holiday, and even if you have people running your accounts around the clock, mistakes happen.

Alright folks, enough theory. Here’s how you calculate this.

This is just the difference between your two windows expressed as a lift. It's how you understand exactly how much revenue you should expect from a certain set of ads over a certain number of days.

Always do this by channel. Don't blend it. Meta and Google have completely different lag shapes because they have completely different user behaviors.

Then, we want to flip this around so we can use this calculation as a reference:

Your day-one target = your long-window target ÷ your multiplier

Here's a working example. Say your 30-day new customer ROAS target on Meta is 2.0 and your cyber week Meta multiplier is 1.36. Your day-one target during cyber week is 2.0 ÷ 1.36, which is about 1.47. So a 1.5 on Black Friday morning is not a problem. Without that multiplier, you would have cut that campaign, right?

Remember that BFCM is a little different and a little crazy every year. These are ranges, not lines. There are a lot of variables involved, from your ads to macroeconomic conditions to your price points, and they're just different year over year. Don't get religious about your decimal points.

This is something you probably want to do once a quarter.

The multiplier tells you how much revenue to expect. The distribution tells you when that revenue actually arrives. That’s what we need so we can measure ads properly across a timeframe.

What you're building is a table: for each number of days after a touchpoint on an ad, what percent of the total revenue attributable to that ad has landed. Here's the methodology our team uses.

Two things worth calling out separately here, because people mix them up:

Here's a real example. For one subscription food business we work with, average click-based lag was 2.8 days. 79% of purchases landed within one day, and another 10.5% landed between day two and day seven. So 90% of conversions landed roughly inside a week.

That totally justifies a 7-day attribution window. A high-AOV purchase would produce a completely different curve and a completely different answer, because it's a high ticket item with a totally different consideration phase.

For that same business, new customer revenue was $759k at 7-day, $776k at 14-day, and $785k at 30-day. Going from 7 to 30 days credits only about 3.4% more revenue, and two thirds of that lift lands in the 8 to 14 day range, not the 15 to 30 day range.

That's real, actionable understanding of how your customers behave. Stretching to 30 days doesn't buy them anything. 14 days captures all the genuine conversion behavior.

In a spreadsheet, build a grid. Rows are your Q4 phases. Columns are your channels. The cells are your 7-day, 14-day, 30-day, 90-day, whatever ranges you need for each channel.

Calculate your conversion multiplier for each channel on each range of time and put those in the table. It should probably look something like this:

Then, as you're going through Q4, take whatever your one-day number is on a given day and multiply it by the figure for that phase and that channel. That's your realistic expectation of performance, based on last year's actual conversion behavior.

This helps you know if your performance is headed in the right direction so you don't cut something just because it looks soft on a random day.

The conversion lag chart lives on your overview page. Go there, add tiles, and follow our documented setup for it here.

Here are the common mistakes we see with conversion lag. With great power comes great responsibility.

Do not change your measurement framework mid-flight. Stick to the plan. If your team has a routine, if they have a model they like, if you understand your conversion lag and your lookback windows and your accounting mode, don't go changing those close to Black Friday out of panic or a desire to push extra performance. You're more likely to just make mistakes and confuse yourself. Also, don't build your conversion lag benchmarks in the middle of November. Build them now, in September, while you have clean data and time to think clearly.

Don't panic. There's a natural pre-BFCM softening that happens with real ROAS performance. Like I said, people are sitting around waiting for discount codes before they buy, which means all of your ads are going to perform a little worse over this stretch. Don't be overly reactive. Everyone's holding out for the sale.

Don't run your postmortem too early. If you judge your BFCM performance on December 5th, you'll probably end up cutting the wrong channels through the end of the month. Wait until BFCM week has a full 30 days of attribution built up, realistically the new year, then look at the 30-day view. Set a reminder and pull again at 60 days. That will give you the full view.

You've got about 2 weeks to do the following:

Do these five things and you'll spend Black Friday making tons of money instead of making tons of wild guesses.

There's no real trick that's going to 10x your Q4 performance. There's only knowing which numbers actually matter for your business.

If you want to make this analysis part of your stack, it's time to get on a demo with us.

We're giving away free Blackstone grills to qualified businesses who book demos right now. So what are you waiting for?

🍝 Join us for dinner in NYC. We’re hanging out with QRY and you’re invited.

🤖 Meta launched Muse, a free personal AI agent that also helps people buy things. Like a personal assistant meets that one friend you always pressures you into buying that expensive thing you only sort-of want.

💀 Is Meta going to die in 2026? I went on the What’s Working podcast to dish about everything I’ve learned in four years of writing 200+ issues of this newsletter.

💰 Roman Khan is here to teach you how to get rich. The legend of ecommerce speaks on the OpenResidency podcast.

📦 We went hard on this BFCM Operators Guide, by 1800DTC. This isn’t entry-level stuff - serious strategic value inside.

Love this newsletter? Please forward it to your friends, coworkers, or that one juror in Boston. Subscribe here to get this data in your inbox every week.