To measure WhatsApp results in e-commerce, separate three questions: did the message arrive, was there interaction, and was the order paid? Each answer depends on a different piece of evidence. A useful report states what was counted, the calculation basis, and the observation cutoff. Without these definitions, two apparently identical rates can describe very different operations.
Sending, delivery, read and sale are not the same outcome
The first precaution is to understand the meaning of the status in the tool used. The official Meta collection for the Cloud API returns send, delivery and failure events associated with the message identifier. A technical confirmation of send does not prove a purchase.
| Information | What to check | What it does not prove |
|---|---|---|
| Request accepted | The system received the request and provided a reference for tracking. | That the recipient received the message. |
| Delivery confirmed | There is confirmation of delivery for that message. | That the person read, clicked or bought. |
| Read recorded | A read confirmation is available in the report. | That the person understood the offer or intends to buy. |
| Reply or click | An interaction is recorded and linked to the send, according to available tracking. | That the order was paid. |
| Payment confirmed | The shop or responsible system confirmed payment for the correct order. | That the message was the cause of the sale. |
Read should be treated as observed information. The WhatsApp allows disabling read receipts; therefore, the absence of that confirmation does not authorize concluding the message was ignored. Also check whether the system receives and processes the necessary returns.
In custom integrations, message status webhooks help track these events. Delivery of a webhook to your server and delivery of the message to the customer are distinct events: an HTTP response from the server does not substitute for channel confirmation.
Define the base before calculating any rate
Choose a set of messages and keep that set during analysis. For example: messages from a campaign launched on a given date, observed until a cutoff time. If the report mixes new sends with deliveries from old campaigns, the result loses comparability.
Record the campaign or automation, the audience, the sending period, the time zone and the observation window. Then define the unit for each indicator: message, contact or order. A customer may receive two messages about one order and make another purchase in the same period.
- Messages: count each identifier once per stage reached. Repeated returns of the same event are not new sends.
- Contacts: count distinct people or registrations according to a documented rule, without adding each reply as a new customer.
- Orders: use the purchase identification. Two messages about the same order do not generate two sales.
- Accumulated states: a read message also integrates into the count of delivered messages. If the tool shows only the latest status, check how to rebuild totals without excluding reads from the delivery base.
When the necessary data is not available, report “not measured”. Zero means measurement occurred and no result was recorded; it is not a substitute for missing information.
Formulas for a report that can be verified
The formulas below are proposed conventions to organize the analysis. Your tool’s dashboard may use a different basis. Before comparing results, check the meaning of each field and keep numerator and denominator within the same set of sends.
| Indicator | Proposed calculation | Main caveat |
|---|---|---|
| Delivery rate | Messages with confirmed delivery ÷ messages sent × 100. | Define what the system calls sent. Failures prior to sending require separate tracking. |
| Observed read rate | Messages with read recorded ÷ messages with confirmed delivery × 100. | Does not represent all actual reads when confirmations are unavailable. |
| Response rate per contact | Contacts with a response linked to the campaign ÷ contacts reached by the campaign × 100. | Count each contact once and define how to attribute the response to the send. |
| Click rate per contact | Contacts with an identified click ÷ contacts reached by the campaign × 100. | Requires tracking that identifies and deduplicates contacts. Total clicks are not unique people. |
| Payment rate after contact | Eligible orders paid after contact, within the defined window ÷ eligible orders with at least one message delivered × 100. | Link each message to the order and exclude payments made before the contact. |
If the denominator is zero, the rate is not calculated. If you only know the total number of clicks, present that total with its limitation. Do not convert it to a rate per contact without knowing how many distinct people clicked.
For a campaign sent to an audience, it can make sense to track unique buyers among the contacts reached. For a order payment reminder, the primary unit can be the order. These metrics answer different questions and need different names.
Calculated example: messages are not orders
This example is entirely fictitious and intended only to demonstrate the calculations. It does not represent a customer result, a WPP Marketing test, or an expected market rate.
A store selected 100 unpaid orders eligible for a reminder. It sent one message per order, always before payment. By the report cutoff, 90 messages had confirmed delivery and 54 had recorded reads. Among the 90 orders reached, 18 were paid within 48 hours after delivery of their respective reminder. The store waited for the window to close for all orders before finalizing the numbers.
- Delivery: 90 ÷ 100 × 100 = 90%.
- Observed reads among deliveries: 54 ÷ 90 × 100 = 60%.
- Payment after contact among orders reached: 18 ÷ 90 × 100 = 20%.
- Payment after contact over all selected orders: 18 ÷ 100 × 100 = 18%.
The last two calculations use the same 18 payments but answer different questions. The report should show the base name, not just “conversion”. The 48-hour window was chosen for the example; it is not a universal recommendation or a product rule.
The ten messages without confirmed delivery should not automatically be labeled failures: available states must be checked. You also cannot claim the 18 payments came from the 54 reads without linking records. A payment can occur without a read confirmation.
If the same order receives a second reminder, it remains one order. In consolidated analysis, count its payment only once. In message-level analysis, define in advance how to allocate credit among contacts and make clear that summing overlapping reports can double-count results.
Payment after the message does not prove the sale was caused by it
A person may receive a reminder and pay because they already intended to complete the purchase, because they saw an email, or because they received help from support. Temporal sequence is evidence of association; by itself, it does not measure how much the message added to the outcome.
To monitor the operation
Use a descriptive metric, such as “orders paid after message delivered, within the defined window”. Document the initial event, the window duration, the link to the order, and the handling of cancellations, refunds, and repeated contacts. When summing values, clarify whether revenue considers shipping, discounts, and returns. Associated revenue is not profit.
To investigate incremental effect
When conditions allow running an experiment, randomly assign eligible orders or customers to a group that receives the optional reminder and to a comparison group. Avoid placing the same person in both groups and keep other conditions comparable. Do not suspend communications necessary for customer service or order fulfillment to run this test.
Compare payments across all eligible individuals in each group according to the initial assignment, taking sample size, duration and uncertainty into account. Comparing only those who received or read the message may select people with different characteristics and bias the conclusion. This design is a proposed evaluation method, not an experimentation feature provided by WPP Marketing.
How to use the numbers to decide the next adjustment
A decrease in a rate is a signal to investigate, not a definitive diagnosis. The same result may have more than one cause.
- Fewer deliveries: check available error reports, contacts, connection, sending conditions and return handling before changing the text.
- Fewer recorded reads: verify measurement coverage and compare similar audiences and time periods. Do not attribute the whole difference to recipient interest.
- Interactions without payment: check whether the conversation was attended to, whether the link works, whether the order is correct and whether there are checkout obstacles. These are verification hypotheses, not proven causes of the rate.
- More payments with more stop requests: reassess audience, frequency and relevance. The WhatsApp Business Messaging Policy requires permission to contact and respect for stop requests, including outside WhatsApp.
Prefer comparing your operation to its own history using the same definition. A rate published by another company is only comparable if you know the audience, channel, period, sample and formula. An email benchmark, for example, does not set a WhatsApp read-rate target.
What to check in WPP Marketing reports
WPP Marketing compiles message reports with filters by period and status, exports, and delivery, read and error information as available from the connection. Commercial tracking links messages to payments and store sales. To interpret that result, check the rule used for the metric and which events the integration provides.
The team can build its analysis from those records and the store information. The formulas, deduplication between campaigns and the experiment described in this article are a proposed working method; they do not mean all those controls are available on a product screen.
If your operation uses its own system, the Events API handles supported commercial events, while status webhooks report message-related occurrences. The integration used by the store determines which information can be related. Before concluding that a campaign failed or succeeded, confirm that the necessary events arrived.
Sources consulted
Query performed on October 20, 2026. Technical definitions and terms of use were checked in the sources below. The formulas, the fictitious example and the verification criteria are an editorial proposal for measurement, without an internal benchmark or claim of testing with customers.
- Meta — official WhatsApp Cloud API collection on Postman: examples of send, delivery and failure responses.
- WhatsApp — privacy settings: read-receipt controls.
- WhatsApp Business — Messaging Policy: permission and stop-request rules.
- WPP Marketing — status webhooks: distinction between message events and returns to the integrated system.
To link communication to your store events and review available records, get to know the WPP Marketing platform and assess the features appropriate for your operation.