Front-Loaded vs Late-Flattening Retention Curves: Which Channel Wins Your Next €10K
Two channels with identical 365-day retention can demand opposite budget calls. Read the 90-to-180-day slope to see which curve shape your cash flow can actually afford.
Quick answer
If your next €10K needs to earn back inside 90 days, pick the front-loaded channel (typically Meta, TikTok, paid social) even at a lower 365-day LTV. If you can float payback for 6+ months and your product has genuine repurchase depth, the late-flattening channel (typically Google non-brand, organic, referral) compounds harder. The tiebreaker is the slope between day-90 and day-180 retention — not the endpoint.
Front-loaded vs late-flattening retention curves
Two retention curve shapes that can reach the same 365-day endpoint via opposite paths — one pays back fast then decays, one bleeds early then compounds.
A front-loaded retention curve delivers most of its repeat revenue in the first 60-90 days after acquisition, then decays steeply. A late-flattening curve loses more customers early but the survivors keep buying — the slope between day-90 and day-180 is nearly flat.
Both shapes can converge on the same 365-day retention rate, which is why comparing channels on annual LTV alone hides the real budgeting decision. The curve shape tells you when the money comes back, and that determines which channel your cash flow can actually afford to scale.
Most operators compare channels on CAC and 365-day LTV. That collapses a curve into two numbers and hides the shape between them. Two channels with identical LTV:CAC can require completely different working capital to scale.
Why the two shapes exist
Front-loaded curves come from channels that catch high-intent-but-shallow buyers: paid social pushes a product into a feed, the buyer converts on impulse, and repurchase happens quickly or not at all. Meta and TikTok cohorts almost always land here.
Late-flattening curves come from channels that catch buyers already searching for a solution: Google non-brand, organic, and referral. These buyers churn harder in the first 60 days (weaker fit filtering) but the ones who stay have real intent — and they buy for a long time. This is the mechanism behind Meta front-loading and Google late-flattening across most €1M-€15M stores.
The 365-day trap
If you only look at 365-day retention, a Meta cohort at 22% and a Google cohort at 22% look identical. The Meta cohort earned 70% of its LTV by day 90; the Google cohort earned 35% by day 90 and is still climbing. Same endpoint, opposite cash-flow profile.
How to read the 90-to-180-day slope
Pull retention at day 90 and day 180 for each channel cohort. Compute the slope: (retention_180 − retention_90) / retention_90. A slope steeper than −0.35 is a front-loaded curve. A slope flatter than −0.15 is late-flattening. Anything between is a hybrid — usually email-driven or referral-heavy channels.
For a proper read you need at least 400-500 customers per channel cohort with a full 180-day window observed — thinner cohorts produce noisy slopes that flip sign month-to-month. If your Google Ads cohort is 90 customers, don't diagnose curve shape from it; wait or aggregate quarters.
Picking where the €10K goes
Match curve shape to cash-flow tolerance, not to LTV. If your inventory cycle is 60-75 days and you're financing growth from operating cash, cash-constrained operators should pick front-loaded curves even when late-flatteners show higher 365-day LTV. The compounding you can't afford to wait for isn't yours.
If you sell a genuine repurchase product — subscription skincare, coffee, supplements — the late-flattening channel almost always wins your next €10K. The survivors are the entire business model, and Google/organic filters for them harder than paid social does. An apparel store with a 40% one-time-buyer rate should think the opposite way and lean front-loaded.
Sub-€40 AOV beauty is its own case
TikTok cohorts for sub-€40 AOV beauty show a steep-then-cliff shape: huge day-30 revenue, then a repurchase cliff at day-60. Treat it as an extreme front-loader and never fund it from cash you need back after 120 days.
The survivor-cohort illusion
Late-flattening curves flatter you. By day 180 the survivors look like great customers because the weak-fit buyers already left — the curve isn't flat because retention improved, it's flat because there's nothing left to lose. If you extrapolate LTV from the flat tail without weighting for the day-30 bleed, you'll overpay for that channel by 20-40%.
The correction is to always quote channel LTV as (surviving retention × cohort size at day 0), never as a per-survivor number. If your Google cohort loses 55% in the first 90 days, that loss is a real cost of acquisition on that channel — even if the survivors then buy forever.
A concrete €10K split
Weight the split by curve-shape fit to your cash cycle. A rough rule for a €1M-€15M store with 60-day inventory financing: 60-70% into the channel whose curve pays back inside your cycle, 30-40% into the compounding channel as a bet on next year. Reverse the weights if you're cash-rich and margin-rich.
A worked split for a €3M apparel brand with Meta at −0.42 slope and Google non-brand at −0.12 slope: €6.5K into Meta prospecting to fund the next inventory drop, €3.5K into Google non-brand and branded search to build the compounding tail. Revisit after one full cohort matures — usually 120 days — and rebalance based on realised slope, not planned slope.
Frequently asked questions
A curve where the 90-to-180-day slope is steeper than −0.35, meaning the cohort loses more than 35% of its day-90 retention by day 180. Paid social — Meta and TikTok especially — sits here for most €1M-€15M stores.
A curve where the 90-to-180-day slope is flatter than −0.15. The cohort bleeds early but the survivors barely churn. Google non-brand, organic search, and referral typically produce this shape.
Yes, and it's the most common trap. Two channels can both land at 22% retention at day 365 while earning 70% versus 35% of their LTV by day 90. Same endpoint, completely different cash-flow profile.
At minimum 400-500 customers per channel cohort with a full 180-day observation window. Below that, curve slopes are dominated by noise and can flip sign between adjacent months. For thin channels, aggregate two or three quarters before diagnosing shape.
No. LTV:CAC ignores when the money comes back. A cash-constrained operator should pick a front-loaded curve at 2.5:1 over a late-flattener at 3.5:1 if the difference is 90-day versus 240-day payback. Timing of cash matters as much as amount.
Subscription products push almost every channel toward late-flattening because the survivors auto-repurchase. Subscription-heavy skincare almost always rewards late-flattening channels, and paid social becomes a top-of-funnel feeder rather than a standalone LTV play.
Meta catches impulse buyers in a feed — they convert once, quickly, and often don't return. Google non-brand catches active searchers whose intent survives longer; they filter harder at day-30 but the survivors have real product-need alignment. Channel intent quality drives the curve shape.
Not always. Flat tails can be a survivor illusion — the curve is flat because the weak buyers already churned, not because retention improved. If you quote LTV per survivor instead of per acquired customer, you'll overpay for that channel by 20-40%.
Hybrid — slopes between −0.15 and −0.35. Email is a retention amplifier, not an acquisition channel per se, so its cohorts inherit the shape of the underlying acquisition source. Referral tends to sit closer to late-flattening because referred buyers self-select for fit.
Once per fully-matured cohort — for a 180-day window that's every 6 months at the earliest. Rediagnosing monthly on immature cohorts introduces noise and encourages over-trading your budget. Set the allocation, hold it a quarter, then look.
Track CAC, channels, and funnel conversion in one place
Metricuno connects ad spend, funnel events, and revenue so you can see CAC by channel, cohort, and campaign — without stitching together five tools.