Email Marketing Campaigns: From Brief and QA to Follow-Up

An email marketing campaign is a goal-bound email send or series for a defined, eligible audience over a set period. A reliable campaign starts with a brief that fixes the objective, audience, message, call to action, timing, and measurement; passes the exact production build through an independent QA gate; and uses delivery, click, conversion, unsubscribe, complaint, and reply evidence to decide who should receive a follow-up—or no further email.

Campaign Monitor’s working definition describes an email marketing campaign as a coordinated set of messages deployed for a specific purpose over a period. That set can be one send or a series. The number of emails is therefore not what makes the work a campaign. A shared goal, a bounded audience, and a controlled path from production to follow-up do.

The word campaign is not used consistently by email platforms. EmailOctopus, for example, calls a scheduled one-off message a campaign and subscriber-triggered steps an automation. Campaign Monitor allows a campaign to mean one message or a series. Use platform labels in the tool, but write operational definitions in the brief:

• Campaign record: the goal, audience, messages, decisions, and evidence that belong together.
• Broadcast: a message released to a segment at a planned time.
• Automation: messages or actions triggered for an individual when defined conditions become true.
• Transactional or relationship message: a classification based on the message’s actual primary purpose, not the template name or the sending feature.

That last boundary matters. The US Federal Trade Commission’s CAN-SPAM guide distinguishes commercial from transactional or relationship email by primary purpose. Other jurisdictions use different rules. Record the classification and the responsible reviewer rather than letting a folder named “transactional” decide it.

An email campaign has no single success formula because it is a workflow, not a metric. Its sub-metrics do have formulas, and their denominators must be stated. Two useful versions are:

Click rate = unique recipients with at least one tracked click ÷ successfully delivered emails × 100

Delivered-audience goal conversion rate = recipients who completed the defined goal ÷ successfully delivered emails × 100

Campaign Monitor documents the delivered-audience conversion formula. Mailchimp’s metric definitions use successful deliveries for open and click rates, while warning that bot activity can distort engagement. Another platform may count total clicks rather than unique clickers or use sends rather than deliveries. Freeze the definition with the brief so the denominator does not change after results arrive.

Here is an illustrative example, not real company data. A campaign attempts 20,000 sends, delivers 19,600 messages, records 490 unique recipients with a tracked click, and attributes 98 goal completions under its documented rule. Click rate is 490 ÷ 19,600 × 100 = 2.5%. Delivered-audience goal conversion rate is 98 ÷ 19,600 × 100 = 0.5%. Those figures describe the funnel; whether they are good depends on the declared goal, audience, offer, and a comparable baseline.

There is no universal campaign-success benchmark. Mailchimp notes that open and click averages vary by audience, industry, company size, and other factors. Its Apple Mail Privacy Protection guidance also explains that preloaded tracking pixels make open status unreliable for affected contacts. Use like-for-like internal history and the campaign’s goal event. Treat Gmail’s provider-specific spam guidance as a delivery boundary, not a business KPI.

Write the decision before writing the email#

A brief is useful when it prevents downstream interpretation. Email on Acid’s two-page campaign brief includes objective, audience, assets, sender, call to action, QA, timing, and KPIs. Extend those fields into a send contract: every important choice needs an owner, a value, and evidence that can be checked before release.

Do not write “customers interested in the feature” as the audience. Write a query another operator can reproduce, the timestamp at which it is evaluated, and the exclusions applied after targeting. A useful set expression is:

sendable audience = target segment ∩ eligible recipients − suppressions − campaign exclusions

Eligibility and relevance are separate decisions. An address can be technically present in the database yet outside the permission, preference, geography, product, or frequency boundary for this message. If the governing rule is unclear, stop and route it to the responsible privacy or legal owner; the brief is not legal advice.

The objective also needs a real event. “Drive engagement” is too vague to choose a CTA or evaluate a follow-up. “Qualified recipients submit the product briefing form within seven days” names an action and a window. The goal does not have to be a purchase, but it must be observable and connected to a decision the team will make.

Design measurement and follow-up before production#

Measurement added after the creative is approved usually produces a report, not an answer. Before building, assign one campaign ID across the email platform, destination analytics, CRM or product event, and closeout record. Decide which system owns the goal event and how long after delivery it remains attributable under the campaign’s stated rule.

For web destinations measured in Google Analytics, Google’s campaign URL guidance identifies utm_source, utm_medium, and utm_campaign as the core manual tags; utm_content can distinguish creative or link variants. Values are case-sensitive. Define the vocabulary once, keep utm_medium=email consistent, and make the campaign value match the stable campaign ID or approved name. Do not put recipient data in campaign tags.

Test the full destination, including redirects, before freezing the message. Confirm that the landing page carries the promise made in the email and that the goal event fires only when the intended action is complete.

Keep these steps separate so the follow-up can address the actual break in the path.

If the campaign includes an experiment, write the hypothesis and assignment unit now. Change one decision-relevant variable in an A/B test unless the design and sample genuinely support a multivariate test. State whether the winner is determined by click, conversion, or another goal. An open-based winner may be convenient, but it is a weak foundation when privacy features and security systems generate opens or clicks that did not come from a person’s action.

Build one production version with known edge cases#

The production record should point to the final audience query, message version, destination version, and tracking map. A copied template is only a starting asset. Freeze the values that can change the recipient experience:

• visible From name, From address, and Reply-To path;
• subject and preheader as one inbox presentation;
• message body, plain-text alternative, images, and alt text;
• personalization fields, dynamic-content rules, and every fallback;
• primary and secondary destinations, redirects, and campaign tags;
• footer, sender identification, preference controls, and unsubscribe behavior; and
• send time, time zone, throttling or batch plan, and recipient query version.

Build fixtures for the edge cases instead of hoping the preview finds them. Include a normal eligible recipient, missing optional fields, longest plausible values, each dynamic-content branch, each language or region used, and a record that must be excluded. Use controlled internal test data; do not reactivate a real suppressed address to make a test case.

For a series or automation, test state transitions as well as individual messages. A recipient who completes the goal after message one should not receive message two if message two repeats the same ask. A recipient who unsubscribes, complains, hard bounces, or becomes ineligible must not remain in the promotional path because the workflow cached an earlier segment.

Make QA a release gate, not a proofreading round#

The person who built the message knows what it was intended to say and can unconsciously fill gaps. Salesforce’s email testing guidance recommends that someone else perform final QA, using representative data, domains, devices, and a checklist. Its suggested checks cover subject and preheader, template, copy, images, links, alt text, legal content, unsubscribe, plain text, rendering, tracking, data, and dynamic elements.

Turn those checks into seven gates against the exact production build:

Click the rendered message, not the URL list in the brief. Exercise the unsubscribe path far enough to verify that the request reaches the expected suppression system. Send tests through the production sending path where safe, then inspect received headers and actual destinations. A green check in a builder cannot prove that a redirect, personalization source, domain configuration, or external form works.

Provider and legal requirements belong inside the gate, with their scope named. Google’s current sender guidelines require SPF or DKIM for senders to personal Gmail accounts and add SPF, DKIM, DMARC, alignment, spam-rate, and one-click-unsubscribe requirements above its bulk threshold. The FTC’s US guidance requires accurate routing information and non-deceptive subjects, sender-location information, a clear opt-out method, and timely handling of opt-outs for commercial email.

Record a release receipt when the gates pass. It should contain the campaign ID, message version, audience-query version and frozen count, test cases and results, unresolved exceptions, approver, scheduled time and time zone, monitoring owner, and stop authority. Any change to the message, audience, destination, or send configuration after approval reopens the affected gates.

Watch the send as a controlled release#

Immediately after release, verify that the platform accepted the expected population and that the first delivered samples match the approved build. Watch for a volume mismatch, unexpected bounces or deferrals, authentication failure, broken destinations, missing goal events, unsubscribe errors, complaints, and replies that need a person. If the platform supports staged sending, define the evidence required before the next batch; do not invent a pause capability the sender does not actually provide.

Gmail recommends keeping its Postmaster Tools user-reported spam rate below 0.1% and avoiding 0.3% or higher. Those values apply to Gmail traffic and sender operations, not to the campaign’s conversion goal. Other destinations have their own current requirements and feedback systems. A team still needs internal stop rules tied to its normal traffic, message risk, and ability to intervene.

The first report should separate three questions:

1. Was the release operationally correct? Compare approved and actual audience, version, time, delivery response, links, and goal instrumentation.
2. Did recipients tolerate the send? Inspect opt-outs, complaints, replies, and provider feedback by segment without dismissing small but meaningful failures.
3. Did the campaign advance its declared goal? Measure the goal event over the predeclared window, then use clicks and downstream steps to locate friction.

Do not declare a winner from the first convenient metric. Delivery can stabilize before the business outcome appears. A long buying action can require more observation than a content download. The brief should have set both the early operational review and the final decision window.

Follow up from recipient state, not from a generic cadence#

There is no universal wait time or email count that fits every campaign. The follow-up moment depends on the promised timing, offer window, decision cycle, audience expectations, and observable state. Define the branches before the first send so a weak result does not tempt the team to improvise more volume.

Apple Mail Privacy Protection makes “did not open” an especially weak branch. Mailchimp explains that affected messages can be reported as opened regardless of recipient activity, distorting open-based tests, segments, and automations. Its separate resend guidance offers a product workflow for one non-opener resend but warns that resending can increase unsubscribes and complaints. That is a bounded tool option, not a reason to make non-opener resend the default.

Prefer evidence closer to the intended outcome: clicks, replies, completed forms, bookings, purchases, activations, or another declared event. Before any follow-up, rerun eligibility and suppressions, remove people who already completed the goal, account for other scheduled messages, and QA the follow-up as a new production send. Salesforce explicitly advises repeating QA for a correction message; urgency does not make a second error less damaging.

Close the campaign with a reusable decision record#

A campaign is not finished when the last message leaves the provider. Close it when the result, incidents, and next action are recorded. Keep the closeout short enough to use and specific enough to reproduce:

• the final campaign and message versions;
• approved, attempted, delivered, bounced, and suppressed audience counts;
• the primary goal value, denominator, attribution rule, and measurement window;
• clicks, downstream steps, goal completions, unsubscribes, complaints, and relevant replies by planned segment;
• any QA escape, delivery incident, correction, or exception;
• every follow-up branch launched, stopped, or deliberately omitted; and
• one decision for the next comparable campaign: keep, change, test, or retire.

Compare results with the baseline named in the brief, not an attractive industry average found afterward. If the goal underperformed but click rate held, inspect the destination and offer. If clicks fell only in one segment, revisit audience-message fit. If complaints or opt-outs rose, review permission, expectation, relevance, and frequency before trying a new subject line. If measurement failed, repair instrumentation before making a performance claim.

The minimum campaign packet is easy to remember: an approved brief, one frozen production build, an independent QA receipt, a monitored send record, state-based follow-up decisions, and a closeout.