Cash-Flow Timing: When CAC Savings Actually Hit the P&L After a Sprint
A 90-day CRO sprint doesn't yield 90 days of CAC savings. Here's the realistic ramp from week 4 to month 6 — and how to model it without overstating quarterly cash.
Quick answer
A 90-day CRO sprint typically delivers 15-25% of its full-year CAC savings inside the sprint window, ~55-65% by month 6, and ~100% only by month 9-12. First measurable savings land in week 4-6 when the first winning test reaches significance and ships to 100% of traffic — not on day one.
Cash-Flow Timing of CAC Savings After a CRO Sprint
The realistic month-by-month curve describing when conversion-rate lifts from a CRO sprint actually reduce blended CAC on the P&L.
Cash-flow timing of CAC savings is the gap between when a CRO sprint runs and when its conversion lifts show up as lower customer acquisition cost on the P&L. The gap exists because tests need traffic to reach significance, only some variants win, and winning variants only compound once they ship to full traffic and stack with later wins. A finance team modeling a sprint as a step-function saving from day 1 will overstate Q1 cash by roughly 4-6x. The realistic shape is an S-curve: near-zero in weeks 1-3, accelerating from week 4 to month 4, then flattening as the testing pipeline matures.
If you're building the business case for a CRO sprint, the timing of savings matters more than the size. A 12% lift booked in month 7 doesn't help a CFO who pre-committed paid-media spend in Q1 against projected efficiency gains.
Why the lag exists (the mechanism)
Three structural delays stack. First, a test needs enough sessions to reach statistical significance — on a Shopify store doing 80k monthly sessions and a 2.4% baseline conversion rate, detecting a 10% relative lift takes roughly 3-4 weeks per variant.
Second, only about 1 in 5 variants wins cleanly. The other four either lose, come in flat, or sit in the inconclusive band. So a sprint that runs 8 tests typically ships 2 winners, not 8.
The realistic ramp, week by week
Weeks 1-3: 0% of savings (tests building toward significance). Week 4-6: first winner ships; ~20% of monthly run-rate savings active. Weeks 7-12: second and third winners stack; ~50-60% of run-rate by end of sprint. Months 4-6: compounding across funnel stages reaches ~85%. Month 9-12: full run-rate, assuming the testing pipeline keeps producing winners post-sprint.
How to detect where you actually are on the curve
Track three metrics weekly during and after the sprint: percentage of traffic exposed to winning variants, blended conversion rate vs. pre-sprint baseline, and blended CAC by channel. The first two move before the third — CAC lags conversion rate by roughly 2-4 weeks because paid-channel optimization algorithms need to recalibrate to the new conversion volume.
If your blended conversion rate is up 8% but CAC hasn't moved, you're in the lag window, not in a failed sprint. Wait two media-buying cycles before drawing conclusions on the CAC line.
How to model it in the forecast
Use a phased savings schedule instead of a flat annualized number. A working template: 0% of projected annual savings in month 1, 5% in month 2, 12% in month 3, then 18%, 22%, 25%, 28%, 30% through month 8, levelling at the run-rate from month 9. Sum the monthly figures — you'll see roughly 60-65% of the annualized number lands in year one.
This is the number that should feed your business case for the sprint, not the steady-state annual figure. Pair it with a sensitivity band: model what happens if only 1 in 8 tests wins instead of 1 in 5, which pushes the ramp out by roughly 6 weeks.
The two failure modes finance teams hit
(1) Booking the full annualized saving against Q1 spend — this overstates Q1 cash by 4-6x and forces awkward re-forecasting in May. (2) Refusing to credit the sprint at all because savings haven't fully landed by month 3 — this kills the next sprint's funding before the curve has even inflected. Both come from treating CRO as a step-function instead of a ramp.
Experiments to de-risk the ramp
Start the sprint with one quick-win test — a checkout-form simplification or a PDP trust-badge change — that historically wins on apparel and beauty stores. Shipping a winner in week 3 instead of week 5 pulls the entire ramp forward by ~10 days and gives finance a real data point before the quarter closes.
Run two tests in parallel on independent funnel stages (e.g. PDP and cart) rather than sequentially on the same stage. Parallel testing doesn't cut significance time per test, but it doubles the number of winners shipped per calendar month — which is what actually drives the slope of the ramp. This is also why funding a CRO sprint from projected CAC savings only works when you've modeled the curve, not the annual average.
Cash-flow timing FAQ
Typically week 4-6, when the first winning variant reaches significance and ships to 100% of traffic. The blended conversion rate moves first; blended CAC follows 2-4 weeks later as paid-media algorithms recalibrate.
Roughly 15-25%. The rest compounds over months 4-9 as additional winners ship and stack across funnel stages. Booking 100% of annual savings against the sprint quarter overstates Q1 cash by 4-6x.
Paid-media bidding algorithms (Meta, Google) need 2-4 weeks of new conversion data to recalibrate target CPA and bid strategies. Until they do, you're paying the old CAC to acquire customers who now convert at a higher rate — the gain shows up as more orders, not lower cost-per-acquisition.
That's common — about half of sprints don't. It pushes the entire ramp out by 2-4 weeks but doesn't change the shape of the curve. Re-forecast Q1 conservatively and front-load proven quick-win patterns (form simplification, trust badges, shipping-cost transparency) in the next sprint.
Below ~40k monthly sessions, significance windows stretch to 5-7 weeks per test, pushing the first winner to month 2. The full ramp can extend to month 12-15. Low-traffic stores should either run fewer, larger-effect-size tests or accept a longer payback model.
Monthly. Quarterly buckets hide the inflection point between month 2 and month 4, which is where the curve steepens. CFOs modeling quarterly cash need to see month-by-month or they'll either over-credit Q1 or under-credit Q2.
Stalled means no winning test in 8 weeks of running and >3 tests completed. Slow means winners are landing but their effect sizes are smaller than projected. Stalled requires a hypothesis-pipeline review; slow just requires patience and a sensitivity-band update to the forecast.
Roughly, but the lag is shorter. AOV-focused wins (upsells, bundle pricing) show up in the next order — there's no paid-media recalibration delay. Expect AOV impact to land 1-2 weeks after a winning variant ships, vs. 4-6 weeks for full CAC impact.
They multiply, not add. A 6% PDP-to-cart lift combined with a 4% cart-to-checkout lift produces roughly a 10.2% overall conversion improvement (1.06 × 1.04 = 1.1024). This is why the ramp accelerates in months 3-5 — winners from different funnel stages start compounding rather than competing.
Show three things: the month-by-month ramp table, a sensitivity band (1-in-5 win rate vs. 1-in-8), and a rolling re-forecast checkpoint at week 6. Framing it as a phased model with checkpoints — not a binary 'savings or no savings' bet — is what gets the sprint funded.
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.