Email marketing metrics that explain store sales

Choose email marketing metrics with clear denominators. Compare delivery, clicks, orders, revenue and unsubscribes before changing a retail campaign.

Email marketing metrics show whether a campaign reaches customers, earns clicks and leads to orders that leave a contribution after costs. Read them in that order, using a named denominator for every rate and the same audience and observation window throughout.

A high open rate cannot establish that customers read the message. Attributed revenue cannot establish how many sales the email caused. Use the report to locate the next question, then connect it to order records and the customer’s experience after clicking.

Follow the campaign from send to order

Follow a campaign through delivery, engagement, conversion and commercial outcome to decide which part needs attention. Each stage answers a different question.

StageMeasures to readDecision it supports
DeliverySuccessful deliveries, bounces and delivery rateInvestigate address quality or delivery failures before rewriting the offer
EngagementUnique clickers, click-through rate and tracked opensAssess whether the audience reaches the linked content
ConversionPurchasing recipients, attributed orders and conversion rateCheck the offer, product page and buying process
Commercial outcomeRevenue per recipient, average order value and contributionDecide whether the orders justify the campaign’s costs

Delivery means the receiving server accepted the email. It does not establish inbox placement or readership, as Bloomreach’s metrics guide explains. A delivered email can still reach a spam folder.

Keep the reporting unit fixed. One campaign sent to 10,000 addresses differs from a month containing repeated sends to the same people. A campaign-level unique clicker count deduplicates within that campaign. Adding those counts across campaigns can count the same person again.

For orders, reconcile the platform’s attribution record with the store’s paid order IDs. Check cancellations and refunds separately. Confirmed payment establishes that an order happened. The attribution model supplies the rule that credits it to email.

When comparing campaigns, use the same audience definition and elapsed observation period. A campaign with seven days of order data has had more time to accumulate purchases than one sent two hours ago.

Define rates before comparing them

Define each rate’s numerator and denominator before comparing dashboard values. The following definitions use one campaign and one observation window. Multiply each ratio by 100 to express it as a percentage.

Rate definitions for one campaign

MetricFormulaWhat the definition counts
Delivery rateSuccessfully delivered messages / attempted sendsFinal delivery outcomes for this send
Bounce rateBounced messages / attempted sendsFinal hard and soft bounce outcomes, counted once per attempted message
Click-through rate (CTR)Unique recipients with a tracked click / delivered messagesEach recipient once, even if they click repeatedly
Click-to-open rate (CTOR)Unique tracked openers who clicked / unique tracked openersClickers within the recorded opener group
Recipient conversion rateUnique purchasing recipients attributed to the campaign / delivered messagesBuyers, with each person counted once
Unsubscribe rateUnique recipients who unsubscribe through this email / delivered messagesThe campaign’s recorded opt-outs

‘Attempted sends’ excludes people suppressed before the send. If a report still has deferred or pending messages, delivery and bounce rates need not total 100%. Compare final outcomes after the same waiting period.

These are explicit working definitions, not a promise that every provider uses the same labels. Mailchimp’s email report definitions use successful deliveries for click rate and distinguish unique clickers from total clicks. Its clicks-per-unique-opens measure counts recorded openers who also clicked.

Keep the click-to-open numerator inside the opener group. A tracked click can exist without a recorded open. If you divide all unique clickers by tracked openers instead, label that separate definition and use it consistently.

For conversion, the desired event also needs a name. HubSpot’s metrics guide defines conversion around people completing the chosen action relative to delivered emails. A retailer can choose a purchase as that action. A lead form submission measures a different outcome.

Why opens and clicks need care

Tracked opens measure a technical signal, not a reliable count of readers. Apple says Mail Privacy Protection downloads remote email content in the background regardless of engagement. That download can trigger open tracking without the person reading the message. Blocked images can also leave a genuine read unrecorded.

As a result, an open-based denominator can change without the email becoming more or less persuasive. Use tracked opens as a qualified diagnostic. Avoid declaring a subject line successful from opens alone, or treating click-to-open rate as a pure measure of content quality.

Repeated clicks create another mismatch. One recipient clicking five times is one unique clicker and five click events. Automated link checks can also create recorded activity. Record the provider’s bot-filter setting and keep it consistent across comparisons.

If two reports disagree, first match the campaign and window. Then check unique versus total counts, bot filtering and the chosen denominator. A rate difference caused by definitions requires a reporting correction before a marketing change.

Connect email activity to revenue and margin

Connect attributed orders to their revenue and costs before deciding whether a campaign pays for itself. Order count, customer count and revenue answer different questions.

Revenue per recipient divides the campaign’s attributed revenue by its defined recipient count. State whether that count means attempted sends or successful deliveries. Average order value (AOV) divides the same revenue by attributed orders. It measures order size, while revenue per recipient also reflects how many recipients purchase.

Keep the money basis explicit. Discounts reduce what the customer pays for merchandise. Returns and refunds reduce retained revenue. Shipping charges and tax treatment can differ between reports, so compare amounts on the same basis.

For a campaign contribution calculation, start with merchandise revenue after discounts and refunds, excluding tax. Subtract the goods cost retained against those sales and variable fulfilment or payment costs. Then subtract the relevant campaign costs. Label the result ‘contribution after campaign costs’ because rent and other overheads can remain unpaid.

Attribution assigns credit

An attribution window is the period after an eligible interaction during which a platform can credit an order to a message. An open-based window and a click-based window can credit different orders.

For example, Mailchimp’s attribution documentation describes last-touch attribution within eligible windows. Its default email windows are five days after an open or 30 days after a click, with settings that can be changed. Use the account’s actual settings when interpreting its revenue.

A customer can see an advertisement, receive an email and later buy after searching for the store. Different reports can credit that same order under different rules. Adding each channel’s attributed revenue can therefore count the order more than once.

Existing customers also buy without every email they receive. Attributed revenue describes credited purchases. To estimate additional sales caused by sending, use a comparable group that does not receive the campaign and compare outcomes over the same period. A random assignment makes that comparison more credible than comparing recipients with people who chose not to subscribe.

Work through a campaign diagnosis

Use a fictional campaign to connect the rates to specific checks. This example is invented, including every count, cost and customer question. It is not a store result or an industry benchmark.

The retailer attempts 10,000 sends. Final outcomes show 500 bounces and 9,500 successful deliveries. During seven days, the report records 1,900 unique openers and 380 unique clickers. All 380 clickers are in the tracked opener group in this example.

Those recipients generate 570 click events. A click-only, seven-day attribution rule credits 40 paid orders from 38 different recipients. Each purchasing recipient clicked the campaign. The email also records 19 unsubscribes and two complaints.

Fictional campaign calculations

MeasureCalculationResult
Delivery rate9,500 / 10,000 times 10095%
Bounce rate500 / 10,000 times 1005%
Unique click-through rate380 / 9,500 times 1004%
Click-to-open rate380 / 1,900 times 10020%
Recipient conversion rate38 / 9,500 times 1000.4%
Purchasing recipients per unique clicker38 / 380 times 10010%
Unsubscribe rate19 / 9,500 times 1000.2%
Complaint rate using deliveries2 / 9,500 times 100About 0.02%

Dividing 570 total clicks by deliveries gives 6%. Calling that the unique click-through rate would overstate how many recipients clicked. Using 40 orders instead of 38 buyers also changes the conversion measure. That order rate would be about 0.42%, with repeat orders counted.

The orders contain $4,000 of merchandise revenue after discounts, excluding tax and shipping charges. That gives an initial AOV of $100 and revenue per delivered recipient of about $0.42.

After 30 days, refunds against those same orders total $400. Retained merchandise revenue is $3,600. The retailer records $1,800 of goods costs after returned-stock adjustments, $240 of fulfilment costs and $100 of payment and return-processing costs. Campaign costs are $300.

Contribution after campaign costs is $3,600 - $1,800 - $240 - $100 - $300 = $1,160. Retained revenue per delivery is about $0.38. Contribution per delivery is about $0.12. The initial $4,000 attributed revenue is therefore a poor substitute for the amount left after these costs.

This reconciliation follows the same seven-day order cohort as refunds arrive. It does not add orders from a longer attribution window. Other overheads are outside the fictional contribution calculation.

Choose the next check from the evidence

The fictional 5% bounce rate means 500 attempted messages failed. Review the bounce reasons and their list sources first. A recently imported segment full of invalid addresses needs a different response from temporary receiving-server failures.

If delivery is comparable with earlier campaigns but unique clicks fall, check the audience mix and the email’s links. Inspect the offer and destination promise before assuming the subject line caused the change. Opens alone cannot settle that diagnosis.

If clicks hold up while purchasing recipients per clicker fall, inspect the linked product page and checkout. Check availability, mobile errors and the full delivered price. Segment by product or device only where enough observations support a useful comparison.

Customer conversations can identify what the numbers cannot explain. Suppose the fictional campaign’s visitors repeatedly ask whether the advertised lamp includes a bulb and what delivery costs. Those unanswered questions give the retailer a concrete page change to investigate.

Review conversation excerpts against the product and delivery records. Add accurate information where customers need it, and make the email’s promise match that page. Count the unresolved questions alongside the order outcome. A conversation explains a possible barrier, but does not establish that fixing it will produce a particular sales lift.

Use benchmarks and tests with context

Use email marketing metrics benchmarks as a comparison prompt after matching their definitions to yours. A broad retail average can mix different list sources and campaign types. A purchase reminder sent to existing customers differs from a newsletter sent to recently acquired subscribers.

Check the benchmark’s industry, audience source and campaign type. Match the observation window and denominator, then check whether opens and automated clicks receive the same treatment. A benchmark based on total click events cannot judge your unique click-through rate directly.

Your own comparable campaigns can give a more relevant baseline. Keep promotional sends separate from automated messages triggered by a customer’s action. Compare similar products and discount levels so a deeper discount does not disguise a weaker commercial result.

For the fictional lamp campaign, choose one change: put the confirmed bulb-inclusion answer beside the buying controls on the product page. Keep the email and offer unchanged. Randomly assign comparable campaign visitors to the original or revised page, keeping each visitor’s assignment consistent.

Choose completed purchasers per assigned visitor within seven days as the primary outcome. Count every assigned eligible visitor, rather than only those who use the new information. Review contribution after refunds when the same order cohorts have matured.

Keep unsubscribes and complaints visible in the campaign report, alongside refunds and commercial contribution. For a page-only test, opt-outs describe campaign health rather than an expected page effect. For an email-content test, compare them by assigned recipient group.

Decide the sample and observation period before reading the results. Small order counts can move sharply after a few purchases. Record uncertainty instead of stopping when a favourable difference first appears. Adopt the change only when the observed purchase outcome and commercial effect justify it without an unacceptable increase in complaints or refunds.

Give your store its own shopping agent

Customers ask in their own words. Your agent finds, compares and checks out from your catalogue, right on your site.

See it in actionFree trial · Simple setup