How to Diagnose a Drop in Repurchase Rate

Why this metric is easy to misread

A falling repurchase rate often gets interpreted as "old customers stopped buying." That conclusion may be wrong. The rate can also fall because the denominator grew quickly after a new-user campaign, because customer mix shifted, or because the event definition changed.

A good analyst does not start with a coupon plan. A good analyst first asks: did repeat behavior actually get worse, or did the composition of buyers change?

Define the metric before explaining it

Repurchase rate is commonly defined as repeat buyers divided by total buyers in a period. Some companies define the denominator as eligible existing buyers instead. These two definitions answer different questions, so confirm the version before analysis.

repurchase_rate = repeat_buyers / total_buyers

Before comparing two periods, document the time window, buyer definition, repeat-purchase definition, and exclusion rules for refunds, test orders, internal users, or invalid transactions.

Use a simple business example

Last month, 50 of 100 buyers purchased again, so repurchase rate was 50%. This month, repeat buyers stayed at 50, but total buyers increased to 300 after a new-user acquisition campaign. The new repurchase rate is 16.7%, even though repeat buying did not actually decline.

The headline metric fell sharply. The business story is different: acquisition expanded the denominator faster than repeat purchases grew. In that case, a retention campaign may not be the first action. The better question is whether the new users are lower quality, too early in their lifecycle, or simply not eligible to repurchase yet.

Decompose the change

Compare the previous period and current period in three steps:

1. Calculate repeat buyers, total buyers, and repurchase rate for both periods.
2. Hold total buyers fixed to estimate the repeat-buyer effect.
3. Hold repeat buyers fixed to estimate the total-buyer effect.

This separates numerator pressure from denominator pressure. It is not a perfect causal model, but it is a fast and explainable way to stop the team from guessing.

SQL example: monthly repurchase rate

The query below creates a monthly buyer table and flags whether each buyer also purchased in a later month. Adapt the order table, date functions, and eligibility rules to your warehouse.

WITH buyer_month AS (
  SELECT
    customer_id,
    DATE_TRUNC('month', order_date) AS order_month
  FROM orders
  WHERE order_status = 'completed'
  GROUP BY 1, 2
),
monthly_buyers AS (
  SELECT
    order_month,
    customer_id,
    CASE
      WHEN EXISTS (
        SELECT 1
        FROM buyer_month future_orders
        WHERE future_orders.customer_id = buyer_month.customer_id
          AND future_orders.order_month > buyer_month.order_month
      )
      THEN 1 ELSE 0
    END AS is_repeat_buyer
  FROM buyer_month
)
SELECT
  order_month,
  COUNT(DISTINCT customer_id) AS total_buyers,
  SUM(is_repeat_buyer) AS repeat_buyers,
  1.0 * SUM(is_repeat_buyer) / NULLIF(COUNT(DISTINCT customer_id), 0) AS repurchase_rate
FROM monthly_buyers
GROUP BY 1
ORDER BY 1;

SQL example: segment the drop

After the headline metric is calculated, segment the change. Channel, cohort, product category, geography, and platform are usually more useful than one global average.

WITH monthly_segment AS (
  SELECT
    DATE_TRUNC('month', order_date) AS order_month,
    acquisition_channel,
    customer_id,
    COUNT(*) AS orders
  FROM orders
  WHERE order_status = 'completed'
  GROUP BY 1, 2, 3
),
segment_summary AS (
  SELECT
    order_month,
    acquisition_channel,
    COUNT(DISTINCT customer_id) AS total_buyers,
    COUNT(DISTINCT CASE WHEN orders >= 2 THEN customer_id END) AS repeat_buyers
  FROM monthly_segment
  GROUP BY 1, 2
)
SELECT
  order_month,
  acquisition_channel,
  total_buyers,
  repeat_buyers,
  1.0 * repeat_buyers / NULLIF(total_buyers, 0) AS repurchase_rate
FROM segment_summary
ORDER BY order_month, repurchase_rate;

How to read the chart

The workbook includes a segment chart. Use it to identify which segment has the largest rate decline and which segment contributes the most buyer volume. These are not always the same segment.

If paid traffic has a large rate drop but represents only 5% of buyers, it may not explain the business-wide movement. If new customers from one campaign account for most of the denominator growth, that campaign deserves closer inspection even if its own rate is not the lowest.

Use the Excel template

The downloadable workbook includes a Repurchase Diagnosis sheet with formula-backed cells for previous period, current period, rate change, repeat-buyer effect, and total-buyer effect. It also includes a segment table and chart area for comparing channel or cohort performance.

Download Excel template Download checklist

Common mistakes

• Comparing new users and mature customers in the same denominator.
• Calling a metric drop a retention problem before checking acquisition mix.
• Ignoring tracking, refund, and order-status changes.
• Looking only at rate movement without buyer counts.
• Using calendar months when customer cohorts would answer the question better.

Recommended analyst workflow

• Confirm the exact repurchase-rate definition.
• Calculate repeat buyers and total buyers separately.
• Decompose the numerator and denominator effect.
• Segment by acquisition channel, cohort, category, geography, or platform.
• Check tracking, refund, and eligibility changes.
• Write one recommended action and one follow-up analysis.

FAQ

Is repurchase rate the same as retention rate?

No. Repurchase rate usually focuses on buying again. Retention rate can refer to active usage, subscription renewal, login behavior, or any retained state. The right definition depends on the business model.

Should new customers be included in the denominator?

Only if the business question requires it. If you want to know whether all buyers are repurchasing, include them. If you want to know whether existing customers are becoming less loyal, use eligible existing customers or cohort-based retention.

What is the best next chart?

Use a segment bar chart for contribution diagnosis and a cohort heatmap for lifecycle behavior. The segment chart explains where the current movement comes from. The cohort heatmap shows whether customer quality has changed over time.