Email marketing software key metrics: definitions

Email marketing software key metrics are useful only when another analyst can reproduce them. Define the event and eligible population first, then write the numerator, denominator, unit, period, timezone, source, query version, inclusions, exclusions, owner and uncertainty.

This England-focused guide draws on the Information Commissioner's Office (ICO) storage and access technology guidance, the Office for National Statistics (ONS) quality in official statistics material and the Committee of Advertising Practice (CAP) substantiation advice. It provides reproducible working definitions, not target rates or benchmarks. No platform export or campaign result was examined, so these definitions are not an ESP's proprietary reporting rules.

What to take away

Standard email and deliverability metrics

Use these as explicit working definitions, not claims that every email service provider uses the same events or denominators. Keep the provider's status mapping, eligible population and reporting window with each result; these definitions do not set performance benchmarks.

For inbox placement rate and spam-folder rate, define the monitored messages and valid placement results: messages observed in the inbox or spam folder, respectively, ÷ monitored messages with a valid result. State the measurement panel and time; a test panel is not a census of all campaign recipients.

Domain and IP reputation and blocklist status are source observations, not universal campaign rates. Record the sending domain or IP, the dashboard or list, the observed status and the observation time. Sources can include Google Postmaster Tools and Microsoft's Smart Network Data Services (SNDS); keep their scope distinct from campaign outcomes.

Attempted messages

Event: the authorised release process submits a send instruction for one message and one recipient record. Population: records in the signed release snapshot after documented exclusions. Numerator: submitted instructions meeting the event definition. Unit: message-recipient attempts.

Record whether test sends, retries, duplicates, transactional messages and cancelled jobs are included. Use the scheduler log and release identifier, name the timezone and preserve the extraction query.

Provider-accepted observation

Metric: provider acceptance rate. Numerator: messages the supplier records in its defined acceptance state. Denominator: attempted messages eligible under the same job and period. State whether the count covers unique or total accepted events, and report the count beside the proportion.

For an illustrative release of 10,000 attempts, if the provider accepts 9,600, the accepted proportion is 9,600 / 10,000 = 96%. Label this as an example, not a target or a benchmark.

Keep the provider definition, product edition, export time and status mapping. This is supplier-reported operational evidence. It is not independent proof of inbox receipt, human attention or commercial effect.

Tracked opens and open rate

For the open rate, count eligible recipients with at least one specified tracked-open event and divide by eligible recipients with a provider-reported delivered message in the same job and reporting window. Define the event and missing statuses; a tracking request is not proof of reading.

Image blocking, automated activity and privacy controls can affect observation. The Information Commissioner's Office's storage and access technology guidance calls for analysis of the actual mechanism and purpose. The Privacy and Electronic Communications Regulations (PECR), the UK General Data Protection Regulation (UK GDPR) and the Data Protection Act may apply to tracking and associated personal-data processing. Get a privacy review of the actual collection; metric design alone is not approval.

Recorded clicks and click-through rate

For the click-through rate, count eligible recipients with at least one qualified in-scope redirect or destination event, then divide by eligible recipients with a provider-reported delivered message in the same job and reporting window. State how repeated clicks, automated checks, unsubscribe links and navigation links are treated.

Do not compare this with a provider's differently defined click rate. Retain redirect and destination evidence separately, along with unmatched identifiers and late events.

Verified internal outcome and conversion rate

For a campaign conversion rate, count eligible recipients with a validated completed action and divide by eligible recipients with a provider-reported delivered message in the stated period. Set the eligible population from the decision question, not from people who happened to click. Record identity-matching rules, cancellations, returns, duplicate actions and the cut-off for late outcomes; state any different denominator explicitly.

Association after an email contact is not causal incrementality. If the report needs a causal claim, specify a credible comparison and evaluation design before launch.

Suppression integrity

Event: an objection recorded in the authoritative source reaches every named destination. Measure count of changes, failed destinations or elapsed time, but define clock start, stop, pauses and timezone. This operational metric should trigger investigation rather than become a marketing score.

Quality and claim checks

The Office for National Statistics (ONS) quality in official statistics concerns official statistics, yet its attention to relevance, accuracy, coherence and accessibility is a useful review prompt. The Committee of Advertising Practice (CAP) substantiation advice means external performance claims need documentary support before publication.

Assign each metric a warning, review trigger and stop condition. If a definition, source or query changes, start a new version and mark the break. Never repair a disappointing value by moving exclusions after seeing the result.

Retain the prior calculation and explain any restatement. A corrected series should show which periods changed, who approved the correction and whether the original decision must be revisited.

Source each metric from its own system: attempted messages from the scheduler log and release identifier, accepted and open events from the provider export, and qualified clicks from the redirect or destination log. Keep the export, query version and timestamp with the metric.

Before you act

Common questions

What must be defined before writing a numerator or denominator?

The article says define the event and eligible population first. Then write the numerator, denominator, unit, period, timezone, source and query version. Also record inclusions, exclusions, owner and uncertainty. Each stage needs its own definition.

Why is provider acceptance not proof that a message was seen?

The article describes provider acceptance as supplier-reported operational evidence. It is not independent proof of inbox receipt, human attention or commercial effect. The provider definition, product edition, export time and status mapping must be kept, and the count reported beside the proportion.

How should a change to a metric definition be handled?

If a definition, source or query changes, start a new version and mark the break. Retain the prior calculation and explain any restatement. A corrected series should show which periods changed, who approved the correction and whether the original decision must be revisited.