Denominator Drift: Why Revenue per Recipient Falls When Your Customers Haven't Changed
You've started to wonder whether your lifecycle team is coasting. You haven't said it in a meeting yet.
The slide seems to back you up. Revenue per recipient, the number your lifecycle lead uses to show what each subscriber is worth to email and SMS, has been sliding for months.
Before you act on it, there's a question almost nobody asks about that number. What is it counting?
What Is Denominator Drift, and Why Should a CMO Care?
Revenue per recipient has two halves. Everyone argues about the top one: the creative, the offers, the timing. The bottom one, the count of recipients, gets treated as a fact.
It often isn't one. I call this denominator drift: revenue per recipient falls because what the bottom number counts has changed, while your customers behave exactly as before.
Three things cause it. At the brand I work with, all three showed up at once:
Colder people joined the programs.
The same people got counted twice.
Some of those people got mailed twice, and nothing flagged it.
Only the first is about your audience. The other two are about your plumbing, and they're the ones nobody checks.
The Problem Underneath: The Count Isn't Counting People
Cause 1: Colder people joined
You already know this one. Add people who are less likely to buy and the average falls.
We added a second tracking source that recognizes more site visitors and sends them into the same programs. The new people browse more and buy less. Your version might be lead-gen ads, a giveaway, or the peak-season discount wave.
This cause is real, and it's the only one most teams think to look for.
Cause 2: The same people, counted twice
How big was the overlap? That depends on what you divide by, which is the point of this whole post.
When we measured it, the answer swung from small to large depending on the base. Round numbers show why. Say 1,000 people are in both programs. Against the 15,000 people mailed by either one, that's under 7%. Against the 3,000 in the newer program, it's a third.
Same people, same week. "A small overlap" and "a third of the new program" are both true, and only one of them would make you look twice.
That overlap distorts the number in two ways, depending on how your slide is built. If it adds recipients across programs, those customers are counted twice. If it shows each program separately, only one program gets credit when they buy. The other keeps a recipient who looks like they didn't buy. Because the overlap is a bigger share of the smaller program's audience, that program takes most of the hit.
Neither view is wrong. Summing overstates your reach, and the per-program view understates whichever program lost the credit.
The orders are real. Only the count is off. But on the slide, some of your highest-intent customers look less valuable than they are, and the program that found them looks worse than it is.
This is the cause growing fastest right now. Identity tools that promise to recover more abandoned checkouts work by recognizing visitors your platform missed. Some of those visitors your platform already knew. If you added one of these tools, assume you have overlap until someone shows you the number.
Cause 3: The same people, mailed twice
There's supposed to be a rule that stops one person from getting the same email from two programs. Ours had quietly stopped working. It didn't throw an error, and nothing on any dashboard changed. It just stopped doing its job.
So some people who abandoned a checkout got both versions of the reminder. These are high-intent customers, close to buying. Getting the same nudge twice from the same brand is the kind of thing that ends in an unsubscribe.
This cause is the expensive one. The first two make a number look worse. This one makes your best prospects' experience worse, and the dashboard shows it as a gentle decline in a ratio.
It also compounds. Every week the rule stays broken, another batch of high-intent customers gets the same reminder twice.
What acting on the wrong number costs
The rebuild. Teams tear down programs that still work for the people they were designed for. They redo creative, cut sends, retest offers.
Lost learning. Months of test history go with each rebuild, and next quarter gets compared against a program that no longer exists.
Deliverability. Colder recipients and duplicate sends both raise unsubscribes and spam complaints, and mailbox providers like Gmail judge your brand on how everyone you mail responds. Push it far enough and emails start landing in spam for your best customers too.
You've probably seen the symptoms without connecting them:
Revenue per recipient is falling while total lifecycle revenue is flat or up.
A new list-growth source or identity tool went live in the last two quarters.
Someone has already proposed rebuilding a program that used to work.
If two of those sound familiar, check the denominator before anything gets rebuilt.
Why the Obvious Check Lies Too
Once you suspect the new people, the natural move is to look at your older subscribers alone and see whether they held. That check misleads in its own way. Subscribers engage less as they age, so any group you follow over time drifts down on its own.
The fair comparison holds subscriber age constant, meaning how long since someone joined. Take people in their first 90 days now and compare them with people in their first 90 days in the same months last year, split by entry source, the channel that brought them onto the list. If the same source at the same age earns about what it used to, the programs are fine.
A source that's new this year has no last year to compare with. Compare its first 90 days with your established sources' first 90 days instead.
If nobody tracked where subscribers came from a year ago, you can still hold age constant. Compare everyone in their first 90 days now with everyone in their first 90 days last year, and start tagging source today so next year's comparison is clean.
This is the step most lifecycle reporting skips.
What Goes Wrong When Leadership Reads the Blended Number
Scenario 1, the rebuild. The line falls, the team gets asked to fix it, and a working program gets torn down. The new version launches against a baseline that was never broken, so whatever it does next quarter reads as progress or failure for the wrong reasons.
Scenario 2, the early cut. Someone sees the new tool dragged the average down and cancels it. Part of that drag was double counting, which made the tool look worse than it was, and it may have been adding contribution all along.
Scenario 3, the duplicate nobody sees. Nobody rebuilds anything and nobody cuts anything. The broken rule keeps mailing high-intent customers twice, and the only trace is a ratio easing down a little each month.
The common thread is a decision made on a number nobody audited. Lifecycle reporting sits apart from the tools and acquisition choices that change what it counts. It's the same root I wrote about in inventory blindness and the personalization gap.
The Denominator Drift Framework: How to Read the Line
Layer 1: Count each person once
Ask for unique people mailed, meaning each person counted once however many programs reached them, alongside total recipients. The gap between the two is double counting. Report revenue per unique person next to revenue per recipient, and look at each program both ways before judging any of them.
Layer 2: Test the safeguards
Every rule that stops duplicate sends points at something. Ask when each one was last checked against a live program. These rules can stop working without any error, so checking is the only way to know.
The same goes for reports. When the platform's report and the raw send counts disagree by more than 5%, the slide shows a range instead of one number.
Layer 3: Separate mix from decay
Split by entry source and hold age constant, as above. If the blended line fell further than the age-matched line, the gap is mix.
What's left is real decay. It doesn't tell you why, since pricing, merchandising, and demand shifts all show up there too. Treat it as the start of a diagnosis. Blame can wait until you know the cause.
Layer 4: Judge each source on what it adds
Whether a colder source or a new tool stays is a leadership call. The final answer is 1-Year LTV:NCAC for that source, built on the total contribution it adds. The early read is 60-Day LTV %, compared with your other sources at the same age. Count each source's people once, after removing the overlap, or the tool looks worse than it is. And check its unsubscribe and spam complaint rates against your list average.
If You Lead the Team: 5 Questions Before Your Budget Review
1. How many unique people did we mail last quarter, against total recipients?
Timeline: 1 week
Owner: you, with whoever runs lifecycle reporting
Output: two numbers and the gap between them
2. When was every duplicate-send rule last checked?
Timeline: 1 week
Owner: lifecycle lead
Output: each rule, what it points at, and the date it was last checked
3. What does revenue per recipient look like by source, at the same subscriber age?
Timeline: 2 weeks
Owner: analytics
Output: first-90-day revenue per recipient by source, against the same months last year
4. Is any rebuild on the table justified only by the blended number?
Timeline: now
Owner: you
Expected impact: those rebuilds wait for the answers above. Fixes to programs that are actually broken go ahead.
5. Are peak-season subscribers tagged as their own source?
Timeline: before they arrive
Owner: lifecycle lead and analytics
Expected impact: in the budget review, discount-wave subscribers show up as their own line instead of dragging down the blended one
If your team can't answer questions 1 to 3 within two weeks, that's the gap to close first. It's the kind of work I do with brands at this stage. Here's how that works.
What Success Looks Like in the 90 Days Before Your Budget Review
This year, that's October through December.
Month 1: the real count
You know unique people against total recipients. Every duplicate-send rule has been checked against a live program. The split by source exists.
Month 2: the peak-season cohort gets its own line
Peak-season subscribers are tagged as a source before they arrive. Their per-recipient number will be low. That's expected, and now it's visibly theirs.
Month 3: the slide that goes into budgeting
The slide shows revenue per unique person, the blended line, the age-matched line, and the gap between them. The budget conversation is about which sources and tools earn their keep, and nobody rebuilds a program on a ratio.
What success does not look like
The blended line probably keeps falling through the budget review. Peak season adds the coldest cohort of the year. Watch revenue per unique person and the age-matched line instead.
The CFO version
You stop paying three times: once to acquire a colder audience, once for a tool to find customers you already knew, and once to rebuild programs that weren't broken.
The Bottom Line
Before anyone argues about the top of the ratio, ask what the bottom is counting.
Two of the three causes of a falling revenue per recipient have nothing to do with your customers.
Metrics & Analytics is where I break down the numbers ecommerce leaders make decisions on, and what those numbers leave out. Subscribe if you're the one making the calls.
Frequently Asked Questions
What is denominator drift?
Denominator drift is when revenue per recipient falls because what the recipient count measures has changed, while customer behavior hasn't. It has three causes: colder people joining lifecycle programs, the same people being counted in more than one program, and the same people being mailed twice because a duplicate-send rule failed silently.
Why does revenue per recipient drop when my email list grows?
Revenue per recipient divides revenue by the number of people mailed. When list growth comes from colder sources, the count grows faster than revenue, so the ratio falls even if total revenue rises. In a simple example, adding 15,000 low-intent subscribers to 50,000 existing ones cuts revenue per recipient by 17% while total revenue grows 7.5%.
Can identity resolution tools inflate email recipient counts?
Yes. Tools that recognize more site visitors often recognize people the email platform already knew. Those customers end up in two programs. On a summed view they're counted twice. On a per-program view, the program that didn't get credit for the order keeps a recipient with no revenue. The overlap can look small against everyone mailed and large against the newer program's own audience, so the base you measure against changes the story.
How can I tell whether my email programs are getting worse or my audience changed?
Hold subscriber age constant. Compare subscribers in their first 90 days now with subscribers in their first 90 days in the same months last year, split by entry source. If the blended line fell further than the age-matched line, the gap is audience mix. Anything left over is real decay, which still needs a diagnosis, since pricing and merchandising changes show up there too.
Can a colder audience or duplicate sends hurt deliverability?
Yes. Colder subscribers and people who get the same email twice unsubscribe and mark emails as spam more often. Mailbox providers like Gmail judge a sender on how everyone it mails responds, so enough of either can push emails toward spam for engaged customers too.
What is the first thing a CMO should do if revenue per recipient is falling?
Ask for unique people mailed against total recipients for the last quarter, and when every duplicate-send rule was last checked. Hold any program rebuild until those answers and a split by source exist.
Comments