Consumer behaviour

Retention & journey

Monthly acquisition cohorts and the path a customer takes through them.

AI verdict

M1 retention is improving — recent cohorts hold 3.6pp more than the older ones.

3 actions

Cohort retention

Each row is one acquisition month. Blank cells have not elapsed yet — they are not zeroes.

Cohort

Dec 25 · 6.9k

100%

62%

44%

33%

26%

23%

21%

20%

19%

Jan 26 · 4.4k

100%

57%

32%

24%

19%

18%

17%

16%

Feb 26 · 7.1k

100%

59%

40%

33%

25%

22%

20%

Mar 26 · 7.8k

100%

72%

52%

40%

31%

27%

Apr 26 · 5.8k

100%

52%

29%

20%

14%

May 26 · 6.1k

100%

59%

47%

32%

Jun 26 · 7.9k

100%

58%

39%

Jul 26 · 4.6k

100%

72%

Aug 26 · 6.3k

100%

Months since first order

0%100%

Purchase journey

Step-through at each stage. The lowest step is the binding constraint — improving anything above it only delivers more people to the same wall.

Sessions1.3M

Product viewed55%695.3K

Added to cart▼ 26.3%182.9K

Checkout started59.8%109.4K

Purchased51.9%56.8K

Repeat within 90d37.9%21.5K

Retention analysis

Sightrz Analyst · claude-opus-5 · 2026-08-07 06:00 UTC

M1 retention is improving — recent cohorts hold 3.6pp more than the older ones.

Across 9 cohorts, M1 retention — the first month after acquisition — averages 61.4%. The drop from M0 to M1 is 38.6pp, larger than every subsequent month combined, which puts the entire retention problem inside the first eight weeks. The weakest funnel step is added to cart at 26.3%.

What the data says · 3

WorkingConfidence80%

Cohort quality is rising

Newer cohorts are retaining better at the same age than older ones. Something changed in acquisition mix or onboarding, and it is worth identifying before it is undone.

Recent three cohorts 62.9% at M1 vs 59.4% for the oldest three

CriticalConfidence90%

The first 60 days decide everything

Customers who reach a second purchase retain at roughly 3.8× the rate of those who do not. Every retention lever that fires after M1 operates on a population that has already mostly left.

M1 average 61.4% · M2 onwards flattens within 5pp

WatchConfidence84%

Added to cart is the funnel's binding constraint

At 26.3% it is the lowest step-through in the journey. Improving any step above it simply delivers more people to the same bottleneck.

Sessions 1.3M → Product viewed 695.3k → Added to cart 182.9k → Checkout started 109.4k → Purchased 56.8k → Repeat within 90d 21.5k

Recommended actions · 3

Rebuild onboarding around day 7, 21 and 45

The decay curve has three inflection points and the current sequence fires on none of them. Aligning sends to the curve costs a scheduling change.

+5.8pp M1 retentionMedium effortConfidence77%

Instrument drop-off at added to cart

The step loses more users than any other and there is no event-level detail on why. One session-replay sample answers it in a day.

+1.8pp conversionLow effortConfidence72%

Report acquisition on cohort-indexed value, not first-order revenue

A young cohort always looks worse than a mature one on calendar reporting, so channels get cut on their age rather than their quality. Cohort indexing removes the illusion.

Removes a systematic budget errorHigh effortConfidence88%