How do I know if discounts are hurting repeat purchases?
You know discounts are hurting repeat purchases when discounted first-time customers perform worse than comparable full-price first-time customers after the first order. Do not judge the offer by first-order conversion rate alone. Compare cohorts by second-order rate, days to second order, second-order AOV, refund rate, product mix, net revenue per customer, and gross margin per customer over the same time window.
A discount can look successful on launch day and still damage retention if it attracts buyers who only wanted the deal, choose low-margin products, refund more often, delay their second order, or only return when another discount is available.
Direct diagnostic: build two groups of first-time buyers from the same period: customers whose first order used the discount and customers whose first order did not. Then compare what happened 30, 60, 90, or 180 days after that first purchase.
The discount is probably healthy if discounted buyers come back at a similar or better rate, place second orders in a normal window, buy products with acceptable margin, and generate enough net revenue to repay the first-order margin tradeoff. It is risky if second-order rate, second-order AOV, refund-adjusted revenue, and margin all fall at the same time.
Terms to define before you start
| Term | Operator definition | Fields you may need |
|---|---|---|
| Discounted first-order cohort | Customers whose first order used a discount code, automatic discount, sale price, or promotion. | Customer ID, order number, first order date, discount code, discount amount |
| Non-discounted cohort | Customers whose first order did not receive a tracked discount or promotion. | Customer ID, order number, first order date, discount amount |
| Second-order rate | The share of first-time buyers who placed a second order within your measurement window. | Customer ID, order count, order dates |
| Days to second order | The number of days between first purchase and second purchase. | First order date, second order date |
| Net revenue | Revenue after removing discounts, refunds, returns, and other reductions you track. | Gross sales, discounts, refunds, returns, taxes, shipping if applicable |
| Gross margin | Revenue left after product cost, and ideally fulfillment or shipping subsidies where available. | Net revenue, cost of goods, shipping subsidy, fulfillment cost |
| Refund-adjusted LTV | Customer value after refunds and returns are removed from the revenue stream. | Customer ID, all orders, refunds, return amounts |
| Product mix | The SKUs, categories, bundles, or subscriptions customers buy on first and later orders. | SKU, product title, category, quantity, order number |
Build a discount cohort map before judging the offer
Start by mapping the discount to customer cohorts, not just orders. A promotion should be judged by the customers it brings in and what those customers do next.
At minimum, group customers by their first order. The first order matters because it tells you whether the discount was used for acquisition, not just as a later retention incentive.
Useful cohort cuts
- First-order discount code or promotion name.
- Discount level, such as 10%, 15%, 20%, free shipping, bundle discount, or gift with purchase.
- First-order date range, such as campaign week, holiday period, launch month, or evergreen period.
- Acquisition channel, such as paid social, paid search, email, affiliate, organic, influencer, or direct.
- First product, SKU, bundle, subscription plan, or product category purchased.
- Customer type, limited to first-time customers for this analysis.
Use the same maturity window when comparing cohorts. A cohort from last week should not be compared to a cohort from six months ago if you are measuring repeat purchase behavior. If your normal reorder cycle is short, a 30- or 60-day window may be useful. If your products are seasonal, durable, or higher consideration, use 90 or 180 days.
Operator rule: compare customers who had the same amount of time to repeat. If one cohort has had 180 days to place a second order and another has had 21 days, the comparison will make the newer cohort look worse even if it is behaving normally.
Example cohort map
| Cohort | First-order condition | Measurement window | Comparison question |
|---|---|---|---|
| WELCOME15 buyers | First order used WELCOME15 | 90 days after first purchase | Do these customers repeat like full-price first-time buyers? |
| Free shipping buyers | First order used free shipping offer | 90 days after first purchase | Does shipping relief attract better buyers than percent-off? |
| Holiday sale buyers | First order during holiday sale period | 180 days after first purchase | Are seasonal buyers returning after the event? |
| Full-price buyers | First order had no discount | Same window as discount cohort | What is the baseline retention and value? |
Check whether discounted buyers come back differently
Once the cohort map is built, look at repeat behavior before you look at campaign-level revenue. The key question is not only whether discounted customers bought once. It is whether they behaved like future customers after the first purchase.
Metrics that show whether discounted customers come back
| Metric | What to compare | What it may mean |
|---|---|---|
| Second-order rate | Discounted first-order customers vs non-discounted first-order customers | Lower repeat rate can indicate deal-seeking, poor fit, weak onboarding, or a product experience issue. |
| Median days to second order | How long each cohort takes to place order two | Slower repeat behavior can delay cash recovery and weaken lifecycle timing. |
| Second-order AOV | Average order value on order two, excluding first-order discount noise | Lower second-order AOV may mean the first discount anchored customers to a lower spend level. |
| Second-order product category | What customers buy when they return | A healthy cohort often moves into replenishment, add-ons, refills, bundles, or higher-fit products. |
| Repeat order discount dependency | Share of second orders that also used a discount | If most repeat orders require another promotion, the first offer may be attracting offer-dependent buyers. |
Interpret these metrics together. A discounted cohort with a slightly lower second-order rate may still be acceptable if customer acquisition cost is meaningfully lower and gross margin holds. A discounted cohort is more concerning when it has a lower second-order rate, slower second purchase timing, lower second-order AOV, higher refund rate, and lower net revenue per customer.
Simple diagnostic sequence
- Filter to first-time customers in the campaign period.
- Split them into discounted and non-discounted first-order cohorts.
- Remove customers who have not had enough time to reach the selected repeat window.
- Calculate second-order rate for each cohort.
- Calculate median days from first order to second order.
- Compare second-order AOV and second-order product category.
- Check how many repeat orders used another discount.
- Review refunds, returns, and net revenue before making a decision.
Test AOV, margin, refunds, and product mix after the first order
A first-order discount is not automatically bad because it lowers first-order margin. Many acquisition offers intentionally trade margin for a new customer. The problem starts when later orders do not repay that tradeoff.
Move from gross sales to contribution impact. Gross revenue can hide a weak discount because it does not show the cost of the offer, refunds, product cost, or shipping subsidies.
Financial quality view
| Layer | What to subtract or inspect | Why it matters |
|---|---|---|
| Gross revenue | Total item revenue before deductions | Useful starting point, but too optimistic on its own. |
| Discounts | Code discounts, automatic discounts, sale markdowns | Shows the immediate cost of the promotion. |
| Refunds and returns | Refund amount, returned units, partial refunds | Reveals whether discounted customers are more likely to reverse revenue. |
| Shipping subsidies | Free shipping cost or subsidized shipping cost | Free shipping offers can reduce contribution even when AOV rises. |
| Payment and fulfillment costs | Transaction fees, pick-pack costs, fulfillment fees | Helps estimate order-level contribution more realistically. |
| Product cost | COGS by SKU, bundle, or category | Exposes low-margin product mix created by the offer. |
| Contribution per customer | Net revenue minus available variable costs | Shows whether the cohort is creating profitable customer value. |
Product mix often explains why a discount appears to hurt repeat purchases. The offer may not be attracting bad customers; it may be pushing customers into the wrong first product. If the discounted cohort over-indexes on low-retention SKUs, one-time gifts, poor-fit starter kits, or products with high return rates, the issue may be offer design rather than discounting itself.
Do not ask only, “Did the discount increase sales?” Ask, “Which customers did it acquire, what did they buy first, what did they buy next, and what revenue remained after refunds, returns, discounts, and product cost?”
Decision tree: healthy discount, weak offer, or retention leak?
After the repeat and financial checks, classify the result before changing the promotion. The same symptom can point to different actions.
Outcome 1: healthy acquisition discount
The discount is likely healthy when discounted customers have a similar or better second-order rate, normal days to second order, acceptable second-order AOV, stable refund behavior, and enough margin over the measurement window.
- Keep the offer active if acquisition economics work.
- Monitor margin by product, not just revenue by campaign.
- Watch for discount fatigue if the same audience sees the offer repeatedly.
- Test smaller incentives to see whether performance holds.
Outcome 2: offer-dependent buyer problem
This happens when customers buy the first order because of the deal and only return when another discount is available. Warning signs include low full-price second orders, heavy repeat discount usage, lower second-order AOV, and weak net revenue per customer.
- Reduce the first-order discount depth for low-quality channels.
- Replace blanket percent-off offers with thresholds, bundles, gifts, loyalty perks, or product education.
- Segment discounted buyers separately in winback flows.
- Avoid training every new customer to wait for the next promotion.
Outcome 3: product or experience retention leak
If both discounted and non-discounted cohorts fail to repeat, the discount may not be the root cause. The issue may be product fit, replenishment timing, onboarding, shipping experience, subscription friction, unclear expectations, or weak post-purchase education.
- Review first-product repeat rates by SKU and category.
- Check refund reasons, support tickets, delivery delays, and product reviews.
- Improve post-purchase instructions, usage education, replenishment reminders, or subscription onboarding.
- Fix the experience before blaming the offer.
Analyze your order export without stitching every tab manually
SignalOps helps ecommerce operators connect order, product, customer, discount, refund, and repeat-purchase data so you can see whether promotions are creating valuable customers or just discounted first orders.
Analyze your order exportWhat to do when discount cohorts underperform
The right action depends on why the cohort underperformed. Do not immediately remove the discount if the real issue is timing, product mix, or lifecycle follow-up.
| Diagnosis | Likely issue | Operator action |
|---|---|---|
| Discounted buyers repeat, but slowly | The post-purchase flow may be too early, too late, or not matched to the product’s reorder window. | Adjust replenishment reminders, education emails, SMS timing, and winback windows based on median days to second order. |
| Discounted buyers have low second-order AOV | The offer may be anchoring customers to low spend or attracting small-basket buyers. | Test bundles, free-shipping thresholds, tiered offers, or product recommendations that raise order quality. |
| Discounted buyers buy low-margin SKUs | The promotion may be attached to products that do not create strong downstream value. | Limit eligibility, exclude weak SKUs, promote higher-retention starter products, or rebuild bundles. |
| Discounted buyers refund more often | The landing page, ad promise, sizing, product education, or expectation setting may be misaligned. | Review refund reasons, PDP claims, size guides, shipping promises, reviews, and support themes. |
| Discounted buyers only return with another discount | The cohort may be deal-dependent. | Test loyalty credits, early access, replenishment perks, gifts, thresholds, or segmented winbacks instead of deeper discounts. |
| Both discounted and full-price buyers fail to repeat | The problem is probably broader than the discount. | Investigate product-market fit, onboarding, customer experience, shipping, subscription friction, and merchandising. |
How long should you wait before judging the discount?
Use your product’s natural buying cycle. A consumable product may show meaningful second-order behavior within 30 or 60 days. Apparel, home goods, gifts, and durable products may need 90 or 180 days. For seasonal campaigns, compare against a similar seasonal period rather than an evergreen month.
If you do not know the natural buying cycle, calculate median days to second order for full-price buyers over the last few mature cohorts. Use that as the baseline window for judging discounted buyers.
Discount cohort diagnostic matrix
Use this matrix to compare offers without relying on one metric. Fill one row per discount cohort, campaign, channel, or product group.
| Column | What to enter | How to use it |
|---|---|---|
| Cohort name | Example: WELCOME15 January first-time buyers | Keeps analysis tied to a specific customer group. |
| Discount code | Code, automatic discount, sale name, or no-discount baseline | Separates offer performance from overall customer performance. |
| First-order count | Number of new customers in the cohort | Shows whether the sample is large enough to trust. |
| First-order AOV | Average first-order value after or before discounts, as long as you label it consistently | Shows initial basket quality. |
| Second-order rate | Share of customers with a second order inside the maturity window | Primary repeat purchase signal. |
| Median days to second order | Median days between order one and order two | Shows whether customers are returning on time. |
| Second-order AOV | Average value of the second order | Shows whether value recovers after the first discount. |
| Refund rate | Share of orders or revenue refunded | Flags expectation, quality, sizing, or offer-fit problems. |
| Net revenue per customer | Revenue per customer after discounts and refunds | Better than gross sales for judging customer value. |
| Gross margin per customer | Net revenue minus product cost and other available variable costs | Shows whether the offer creates contribution, not just revenue. |
| Top first-order SKUs | Most common products bought first | Reveals whether the offer is acquiring through the right entry products. |
| Top second-order SKUs | Most common products bought on the second order | Shows whether customers move into replenishment, complements, or higher-value items. |
| Recommended action | Keep, reduce, restrict, retime, bundle, segment, or stop | Turns the analysis into an operating decision. |
Example interpretation
If WELCOME15 has a high first-order count but low second-order rate, low second-order AOV, and high repeat discount usage, it may be acquiring offer-dependent buyers. If FREE-SHIP has a lower first-order count but better second-order rate and stronger margin, it may be a better acquisition offer even if it produces fewer first orders.
Common mistakes when measuring discount impact
Discount analysis goes wrong when the team measures the easiest number instead of the number that explains customer quality. Watch for these mistakes before changing your offer strategy.
- Judging the discount on conversion rate alone. A higher first-order conversion rate does not prove the offer creates valuable repeat customers.
- Mixing immature and mature cohorts. Customers acquired yesterday have not had the same chance to repeat as customers acquired 120 days ago.
- Ignoring refunds and returns. Gross sales can make a discount look better than it is.
- Comparing holiday buyers to evergreen buyers without labeling the season. Holiday cohorts may behave differently because of gifting, urgency, inventory, or sale expectations.
- Using gross revenue instead of net revenue or margin. A discount can increase revenue while reducing contribution per customer.
- Ignoring product mix. The offer may be pushing customers into SKUs that rarely lead to a second order.
- Failing to separate acquisition channel from offer impact. A discount used in one channel may look worse because the traffic source is lower quality.
- Changing the offer before checking lifecycle follow-up. Slow repeat purchase may be caused by poor post-purchase timing, weak education, or missing replenishment reminders.
- Combining first-time and returning customers. Acquisition discounts and retention discounts should be measured separately.
- Not checking whether second orders also used discounts. Repeat purchase is less valuable if customers only come back when margin is cut again.
The practical answer: compare discounted and non-discounted first-time customers over the same maturity window, then judge the offer by repeat behavior, net revenue, refunds, product mix, and margin. If the discount creates customers who return profitably, keep improving it. If it creates one-time or offer-dependent buyers, change the offer, the audience, the product path, or the lifecycle follow-up.