Email marketing metrics: what shows real progress?

Which email marketing KPIs belong in your report?

These working definitions relate to a single email send. Only analyse later orders once the selected attribution window has closed. Your export should make clear which people, events and periods each metric covers.

In a flow comprising several messages, the same person may be contacted multiple times. The number of emails delivered does not, therefore, represent the number of different people. Label the report accordingly, rather than mixing metrics per message and per person.

Worked example: more revenue, less revenue per email

Fictitious figures for illustrative purposes only; not actual customer results: Campaign A is sent 10,000 times and delivered 9,800 times. 294 people click a total of 490 times. 49 recipients are each attributed one order worth €100, giving total attributed revenue of €4,900.

• Delivery rate: 9,800 ÷ 10,000 × 100 = 98 %.
• Unique click rate: 294 ÷ 9,800 × 100 = 3 %.
• Order rate per person: 49 ÷ 9,800 × 100 = 0.5%.
• Revenue per email delivered: €4,900 ÷ 9,800 = €0.50.

If you use ‘all 490 clicks’ instead of ‘294 people’, you get 5 per cent. This describes a different metric: multiple clicks do not turn one person into five interested recipients.

Campaign B has 19,600 delivered emails and €7,840 in attributed revenue. Compared with A, total revenue is up 60 per cent. Per delivered email, however, it is €7,840 ÷ 19,600 = €0.40: 20 per cent less than with A.

The next step would be to examine the differences: Was the additional segment less likely to make a purchase? Did the basket value change? Were there different prices or discounts? The figures show a discrepancy, but not yet a cause. Without comparable recipients and offers, A versus B is not a controlled test either.

Why the open rate alone is not enough

Apple Mail Privacy Protection can preload content and register a technical open without the recipient reading the message. Klaviyo explains how to view these opens in your reports. When protection is active, it is not possible to reliably distinguish between human and automated opening.

You should therefore treat the open rate as a limited indicator. A noticeable change may warrant further investigation, but is not definitive proof of a better subject line. Take clicks, orders and negative reactions into account. Click data also requires an examination of the filters available in the tool for automated activity.

Klaviyo and GA4: different perspectives

Attribution is the rule-based assignment of an order to a touchpoint. For example, a customer might click on a link on Monday and make a purchase on Wednesday. Whether the order is attributed to the email depends, amongst other things, on the configured time window and other interactions taken into account. Klaviyo describes this in the Documentation on conversion measurement.

Google Analytics 4 tracks activities recorded on websites or apps. Differences in identification, recorded events and attribution rules may lead to discrepancies in reports. Klaviyo also explains these differences in its Documentation on UTM tracking. You should therefore first compare the measurement rules rather than automatically dismissing a report as incorrect.

Attributed revenue does not tell you how many orders would not have happened without the email. That requires a suitable comparison, such as a carefully designed experiment with a random control group and enough data. Assess total revenue, discounts and contribution margin alongside it.

Finding newsletters in Google Analytics

UTM parameters identify campaign links. Define a consistent scheme for source, medium and campaign, for example utm_source=newsletter, utm_medium=email and utm_campaign=autumn-collection. This allows visits recorded in this way to be attributed to a campaign. Google describes the parameters and relevant acquisition reports in the Guide to campaign URLs.

Use consistent spelling and documented campaign names. Different capitalisation may result in separate values. Also test whether redirects retain the parameters. Do not include personal recipient data in such link labels. A tagged URL alone does not replace either the configured Analytics tracking or the verification of purchase events.

Frequently asked questions about email metrics

What constitutes a good newsletter click-through rate?

A general target figure does not take into account the industry, target audience, delivery method or measurement definition. Start by comparing similar campaigns of your own. External benchmarks serve as a guide provided the timeframe and methodology are transparent.

How often should the team review the results?

After sending, first check for any technical issues. The content evaluation then requires sufficient time to allow for expected reactions and the measurement window used. Supplement individual campaign analyses with a regular review of longer-term trends.

Comparing Klaviyo and GA4: first define what you want to measure

Write the question above the report: are you examining orders Klaviyo attributes to a message, or revenue in website sessions from an email source? Two reports headed ‘Email revenue’ may answer different questions. The GA4 Traffic acquisition report uses session-based dimensions; an attribution report takes a different perspective.

The period and currency need to match, but that does not make different attribution rules identical. Compare values only when their definitions are clear, and label the different perspectives explicitly. Klaviyo’s Conversion documentation explains message attribution; Google describes the Attribution models in Analytics.

Fictitious example: the same three orders, two reports

Own calculation example with fictitious figures, not a client report: Three identified individuals click on a newsletter email. Within the selected example time frame, each places exactly one order for 100 euros. All three orders are recorded once in both systems. The time period, time zone, currency and order value are the same; refunds are not relevant in this example.

For Klaviyo, we explicitly assume that all three clicks are valid for message attribution, fall within the set time window, and that no subsequent Klaviyo message is credited. The second report is deliberately the GA4 session report with the ‘Session – Source/Medium’ dimension; it is not a simulation of the data-driven attribution model.

Under these assumptions, Klaviyo attributes 300 euros for the email; the selected GA4 report shows 100 euros in email sessions and a total of 200 euros in the two search channels. No orders are missing in this example. The same 300 euros are viewed from two perspectives. Adding them together would constitute double counting.

In a real-world account, there may also be data collection gaps or deviating values. Investigate the causes of these individually. A key performance indicator cannot be improved simply by setting the attribution window to a longer duration. To demonstrate additional revenue, a suitable comparison with a clear experimental design is still required.

Turning a deviation into a specific task

A helpful report concludes with a well-founded course of action. If many emails are delivered but hardly any relevant landing pages are visited, check the relevance of the content, links and presentation. If there are good visitor numbers but few orders, examine the product range, availability and landing page. If key figures are only conspicuous in one report, first check the definition of the metric and how it is recorded.

For each subsequent change, note down the hypothesis, the expected observation and the date for the evaluation. This creates verifiable learning steps. The following illustrates how content and measurement questions fit together: annotated newsletter design from our work. For a more comprehensive overview, the Klaviyo audit guide.