Revenue up 55%. Every dashboard green. Most of it has to be re-bought next month.
A cohort decomposition of one growing DTC brand, and the number the blended view never shows.
Revenue up 55 percent year over year. Conversion rate stable. Average order value healthy. Every dashboard green.
That is the blended view of a DTC brand I work with, and it is accurate. It is also close to useless for the one decision that matters most in their planning: what happens to the media budget next year.
Blended metrics answer the question "how did we do." They cannot answer "where did it come from." So I took twenty-four complete months of their order ledger apart, order by order, and sorted every dollar of consumer revenue into two piles: customers buying for the first time, and customers coming back.
Here is what the decomposition showed.
Seventy percent of monthly revenue is customers who showed up that month.
Across the most recent twelve complete months, 71 percent of this brand's consumer revenue came from first-time buyers. In the twelve months before that, it was 78 percent. In no month across the full two years did first-time buyers fall below 65 percent of revenue. In the strongest acquisition months they exceeded 85 percent.
Read that again as an operator. The topline this brand reports, the one growing 55 percent, must be substantially re-acquired every single month. The revenue does not accrue. It resets. A brand like this is not sitting on a revenue base; it is running on a treadmill that happens to be speeding up.
The blended dashboard shows none of this. Revenue is revenue. A dollar from a customer acquired eighteen months ago and a dollar bought last Tuesday look identical in every topline report your platforms produce.
The growth split: both engines are real, and one is bigger.
Decomposing the year-over-year growth itself: roughly 58 percent of the incremental revenue came from new customers, 42 percent from returning ones. Growth here is genuinely two-engined. But the acquisition engine is still doing the majority of the work, on top of already carrying 70 percent of the base.
This is the finding that changes planning conversations. When growth is presented as one number, "keep doing what we are doing" sounds like a strategy. When growth is decomposed, the actual dependency structure is visible: the majority of both the base and the growth rides on the acquisition program continuing to perform.
The good news the blended view also hides.
The same decomposition carries a second finding, and it points the other way. Two years ago this brand's repeat revenue was single digits as a share of the month. In the most recent month, customers acquired during the current era and coming back on their own accounted for 30 percent of revenue. The repeat engine is not inherited from some earlier golden age of the brand; the fade of the legacy customer base to under 5 percent of monthly revenue makes that visible in the same table. The repeat base being built now is coming from the current acquisition program.
That is what a healthy trajectory actually looks like at this stage: acquisition-heavy, with a young retention engine compounding underneath it. But young is the operative word. A base contributing 30 percent of revenue cannot carry a business through an acquisition pullback. It can cushion one.
So: what happens if you stop acquiring customers for ninety days?
For this brand, the ledger answers the question. Roughly seven of every ten revenue dollars would have no mechanism to arrive. The repeat base would keep producing its share, and everything else would simply not occur. No dashboard this brand looks at daily would have predicted the size of that hole, because every dashboard they look at daily reports the blend.
Most owners have never computed this number for their own business. It is not hidden. It is sitting in the order export, waiting for someone to sort the orders into two piles.
What I would take from this if it were your ledger.
The decomposition does not say acquisition spend is too high. If anything it says the opposite: for this brand, easing off paid acquisition to "let the base carry it" would be a planning error visible from orbit, because the base is two years old and carries less than a third of a month. What it says is that the repeat engine deserves to be measured as its own line, every month, with its own trajectory, because it is the thing that eventually changes the answer to the ninety-day question. The brands that get to ease off acquisition someday are the ones watching that line now.
Decomposing revenue by cohort is part of the analytics and reporting work I do for ecommerce clients. If you want to know what your ledger says, start a conversation.