MishaBook a demo

Aug 14, 2026

Weekly Cohorts Review Template

A cohort is a group of customers acquired during the same calendar week (or month), tracked together across repeat purchase windows to measure retention, AOV, and LTV decay. Weekly reviews flag acquisition quality shifts before they compound into margin erosion.

Cohort Definition and Grouping

Cohorts must be defined by acquisition date, not signup date or first interaction. For Shopify DTC, acquisition = first purchase completion. Grouping by week (Monday - Sunday UTC) is standard; monthly cohorts obscure weekly seasonality and delay signal detection by 2 - 3 weeks.

Each cohort row tracks the same customers across 12 - 16 weeks post-acquisition. Columns represent weeks 0, 1, 2, 4, 8, 12, 16 post-acquisition. Week 0 = acquisition week; Week 1 = repeat purchase within days 8 - 14; Week 4 = repeat purchase within days 29 - 35.

Exclude test orders, staff purchases, and bulk/wholesale transactions. Flag cohorts with fewer than 50 first purchases as unreliable (confidence threshold). Separate organic, paid, and referral cohorts only if channel mix is >15% of total volume.

Core Metrics and Thresholds

Track three metrics per cohort: repeat purchase rate (RPR) at Week 4, average order value (AOV) of repeat purchases, and gross margin contribution by Week 8.

Repeat Purchase Rate (Week 4): Percentage of Week 0 customers who place a second order by end of Week 4. Threshold: RPR should not drop >15% week - over - week. A cohort with 28% RPR vs. prior week's 32% triggers investigation.

AOV of Repeat Purchases: Average value of second and third orders. Threshold: AOV should not decline >10% from the prior cohort's baseline. Declines signal either lower - intent customers or discounting creep.

Gross Margin Contribution (Week 8): (Repeat revenue × gross margin %) - (acquisition cost per customer). Threshold: Must be positive by Week 8. If negative, the cohort will not achieve payback within 16 weeks.

  • RPR Week 4 baseline: 25 - 35% for most DTC food, beauty, supplement brands
  • AOV repeat baseline: 105 - 125% of first purchase AOV
  • Margin contribution baseline: $8 - $25 per customer by Week 8 (varies by CAC)

Weekly Review Checklist

Run cohort analysis every Monday morning using the prior week's completed transactions. Do not include partial weeks.

Step 1: Pull the most recent 8 cohorts (8 weeks of data). Verify transaction counts match order reports (flag discrepancies >2%).

Step 2: Calculate RPR Week 4 for the oldest cohort in the set. Compare to the prior week's same cohort (should stabilize within ±3%). If it shifts >5%, investigate data quality or refund patterns.

Step 3: Plot AOV repeat for the last 4 cohorts on a simple line chart. Identify direction: flat, rising, or declining. Declining >10% week - over - week requires root cause analysis within 24 hours.

Step 4: Calculate margin contribution for the cohort that is now 8 weeks old. If negative or <50% of target, flag for acquisition channel audit.

Step 5: Document one insight per cohort in a shared spreadsheet. Example: 'Week of Jan 15 cohort: RPR 26%, down 4% from Jan 8. AOV repeat stable at $118. Likely due to higher % of first - time buyers from new TikTok campaign.'

Failure Modes and Red Flags

Cohort quality decay is the most common failure mode. It manifests as declining RPR across 3 consecutive cohorts (e.g., 32% → 30% → 28% → 26%) without corresponding CAC reduction. Root causes: audience fatigue, creative staleness, or channel saturation.

Margin inversion occurs when repeat AOV drops while CAC stays flat. Example: Week 1 cohort had $120 repeat AOV and $25 CAC; Week 5 cohort has $105 repeat AOV and $25 CAC. This signals product - market fit erosion or competitive pressure.

Cohort cliff: RPR collapses at Week 8 - 12 instead of decaying smoothly. Example: Week 4 RPR is 28%, but Week 12 RPR is only 8% (vs. expected 12 - 15%). Indicates a product issue (quality, shipping, or customer service) that emerges after repeat purchase 2 - 3.

Data quality drift: Transaction counts in cohort rows do not match order export counts. Causes include refund timing misalignment, duplicate order IDs, or timezone errors. Audit data pipeline weekly.

Decision Rules

If RPR Week 4 declines >15% week - over - week: Pause new channel spend within 48 hours. Audit acquisition source mix and creative rotation. Do not wait for Week 8 data.

If margin contribution is negative for two consecutive cohorts: Reduce CAC target by 20% or increase first - purchase AOV by 15% within 7 days. Cohort is not viable at current unit economics.

If AOV repeat declines >10% and RPR is stable: Investigate product quality, shipping speed, and refund rate. Do not assume it is a pricing issue. Run a cohort - level NPS survey within 3 days.

If a single cohort is an outlier (RPR >40% or <18%): Investigate acquisition source, promo code usage, and customer geography. Do not assume it is a trend. Outliers often reflect data anomalies or one - off campaigns.

Reporting and Cadence

Maintain a rolling 12 - week cohort table in a shared Google Sheet or BI tool. Update every Monday by 10 a.m. Include columns: cohort week, first purchases, RPR Week 4, AOV repeat, margin contribution Week 8, and one - line notes.

Share a 5 - minute summary with the leadership team every Monday. Format: 'This week's cohort (Week of [date]): [RPR]% RPR, [AOV] repeat AOV, [margin] margin contribution. Trend: [stable / improving / declining]. Action: [none / investigate / pause spend].'

Escalate to executive review if any cohort misses margin target by >30% or if three consecutive cohorts show declining RPR. Escalation should include root cause hypothesis and proposed fix.

Common Pitfalls

Mixing cohort definitions (acquisition date vs. signup date vs. email list join) destroys comparability. Lock the definition on day one and do not change it.

Excluding refunds from cohort calculations inflates RPR and AOV. Include refunded orders in transaction counts but exclude refund value from AOV. This reflects true customer behavior.

Waiting for 16 - week data before acting on a declining cohort costs 8 - 10 weeks of margin leakage. Use Week 4 RPR and Week 8 margin as early signals. Week 16 data is for post - mortem analysis only.

Comparing cohorts across different seasons (e.g., January vs. July) without seasonal adjustment masks real trends. Use year - over - year cohort comparison for seasonal brands.

Questions

FAQ

Should we use weekly or monthly cohorts?

Weekly cohorts are standard for DTC operators. They surface acquisition quality shifts 2 - 3 weeks earlier than monthly cohorts and align with weekly marketing cycles. Monthly cohorts are acceptable only for brands with <500 first purchases per week (confidence issue) or highly seasonal businesses (e.g., holiday - only). Even then, run weekly internally and roll up to monthly for external reporting.

What if a cohort has fewer than 50 first purchases?

Flag it as low - confidence and do not act on its metrics. Combine it with adjacent weeks for analysis, or exclude it from trend calculations. Statistical noise dominates at <50 customers. For very small brands (<20 purchases/week), use 4 - week rolling cohorts instead.

How do we handle refunds in cohort calculations?

Include refunded orders in the transaction count (RPR denominator and numerator) but exclude refund value from AOV calculations. This reflects true repeat purchase behavior. If refund rate is >15%, investigate product quality separately; do not adjust cohort metrics to hide the problem.

When should we act on a declining cohort?

Act on Week 4 RPR data (available by day 35 post - acquisition). If RPR declines >15% week - over - week, pause new spend within 48 hours and investigate. Do not wait for Week 8 or Week 12 data. Early action prevents margin leakage across 4 - 6 subsequent cohorts.

Want this on your account?

Thirty minutes. Bring the number that keeps you up.

Book a demo

More from the blog