# Earliest Delivery Date on Product Cards A/B Test

Page: https://alidemirbas.com.tr/en/lab/ab-test-library/tests/earliest-delivery-date-on-product-cards-ab-test

## Hypothesis

If shoppers see the earliest estimated delivery date on the card, they may be able to tell whether a product will arrive in time without opening its page. Purchase conversion rate could rise as a result, most of all for orders that must arrive by a certain date.

## When to run this test

This test makes sense when delivery time seems to stay unclear at the listing stage, such as shoppers in session recordings who open product pages just to check delivery, support tickets asking how long delivery takes, or cancellations caused by late delivery. If the estimate helps, shoppers pick a product that will arrive in time straight from the listing and make the purchase decision faster. If the estimate logic does not match actual delivery, or operations cannot meet the dates shown, fix that first, because a date the listing cannot keep can turn conversions into cancellations later.

## Primary metric

**Purchase conversion rate**: Compare purchases among eligible visitors who see a delivery estimate in the listing.

## Other metrics

- **Listing-to-product click-through rate**: Check whether visits to product pages change.
- **Delivery estimate accuracy**: Compare actual delivery with the estimate shown to the visitor.
- **Cancellation or return rate**: Track cancellations or returns linked to delivery expectations.
- **Sales of slower-delivery products**: Make sure sales of slower-delivery products do not fall materially.

## What to check before and during the test

- **Estimate**: Show the earliest estimated delivery date for eligible products.
- **Keep constant**: Keep delivery operations, estimate logic, shipping fees, sorting and card content unchanged.
- **Choice**: Check whether visitors shift toward products with faster delivery.
- **Accuracy**: Compare the displayed estimate with actual delivery.
- **Separate test**: Showing a date versus a number of days is a separate copy-format test.

## What to watch when reading the result

- Even if purchase conversion rate rises, do not roll out the variant before you have compared the displayed estimates with actual delivery. Check the cancellation or return rate again once the test orders have been delivered, because the cost of missed dates appears after the conversion is recorded.
- A shift toward faster-delivery products is not a failure on its own; compare any drop in sales of slower-delivery products with the loss limit you set before launch. The effect you measure covers only the eligible products that showed an estimate.
- If the overall gap looks small, the effect may sit in time-sensitive purchases and get diluted in the overall average. Work out up front how many visitors each variant needs with the [sample size calculator](https://alidemirbas.com.tr/en/calculators/sample-size-calculator), divide that by the daily visitors each arm receives, and round the run up to full weeks so that dates shifting by day of the week do not skew the comparison.
- Check that the card clearly labels the date as an estimate and does not present a location-dependent date as certain before the visitor's location is known. Otherwise you are measuring the effect of a promise, not of delivery information.

## Frequently Asked Questions

### How should the delivery estimate appear on cards?

Use the same estimation logic and operations in both arms. Clearly identify an estimate rather than a guarantee. Do not show location-dependent certainty when the location is unknown.

### Can the variant ship immediately if conversion improves?

First assess estimate accuracy against actual delivery. An unrealistic date can turn an initial purchase gain into later cancellations, returns or support contacts.

### How can delivery dates affect slower-delivery products?

Use purchase conversion as the primary outcome. Monitor slow-delivery product sales against the predefined guardrail. A shift toward faster delivery is not automatically a failure.