Digital marketing campaigns: the parts worth your attention

What to take away

A campaign is a bounded effort with a start condition, a named owner, and an end decision. The work that separates a controlled campaign from a busy one happens in three places: before launch, on a fixed cadence, and after the last dollar.

Preflight runs per campaign, not per template

A preflight checklist copied from last quarter checks last quarter's risks. Build it from this campaign's asset map, then sign it off line by line.

Each answer needs a name and a date, not a tick. A preflight signed by the person who built the asset is not a review.

The asset map names owners, reviewers and status

List every asset the campaign will publish, then give each row four fields: owner, reviewer, due date, approval status. Status values should be few and unambiguous, for example draft, in review, approved, retired.

The reviewer must be someone who can say no. If the same person owns and approves a claim-bearing asset, the map is decorative.

The operating cadence is set before launch

Decide the review rhythm while the campaign is still cheap to change. Daily for the first week on a new destination, then weekly once delivery stabilizes, is a common shape. Write the cadence into the brief so nobody negotiates it mid-flight.

Each review looks at absolute counts next to rates. A 4% conversion rate on 25 sessions is not a signal. A 1.2% rate on 4,000 sessions is.

Pause authority belongs to one named person per shift, with a documented backup. The most expensive campaign failures are the ones where everyone could see the problem and nobody could stop the spend.

Closure is a scheduled decision

Set the closure date at launch. On that date, reconcile what the campaign cost against what it produced, using the definitions written in the measurement plan.

Separate four kinds of evidence: what you observed directly, what the platform attributed, what a model estimated, and what an experiment measured. They will disagree, and that disagreement is information.

Write the outcome in one paragraph: what the campaign can support, what stays uncertain, and what to test next. A campaign that ends without that paragraph gets re-run on memory.

Access cleanup after closure

Dormant access is how a retired campaign keeps spending. Within a week of closure, remove platform logins for people who have moved on, pause automations tied to the campaign, and archive the creative so it cannot be republished by accident.

Keep the brief, the claim record, the approvals, the source exports and the final decision. That set is what makes the next campaign's preflight faster, and it is small enough to store.

Worked example: a two-channel launch

A B2B software team ran paid search and a webinar for a mid-market audience. Preflight caught that the webinar confirmation email linked to a staging page. Fixing it cost an afternoon.

The cadence review in week two showed search converting at target while webinar registrations arrived from outside the target firmographic. The team paused the social promotion driving those registrations and kept search running.

At closure, observed pipeline was positive, platform-attributed pipeline was roughly double it, and no experiment had been run. The written outcome said the campaign supported continued search spend and that the webinar needed a qualification question before the next run.

Where outside guidance helps, and where it stops

The GAO evaluation design guide is a useful check on how an evaluation question, an evidence need and a design choice fit together. It is written for federal program evaluation, so treat it as design discipline, not as proof that a marketing result is causal.

The W3C Privacy Principles statement gives teams shared vocabulary for privacy and warns against shifting privacy work onto individuals. Apply that to the data flow behind the campaign. It does not replace the law, your contracts, or a review of the actual configuration.

The GOV.UK technology selection guidance turns into good buying questions: can this choice change later, who controls the data, what is the ownership cost. Those are public-service rules, not private-sector mandates.

W3C's tips for developing accessible web content cover labels, image alternatives, page language, structure and keyboard access. Test the complete journey against the current standard rather than the asset in isolation.

Accessibility testing belongs inside a broader digital marketing strategy that connects customer decisions, channel roles and measurement, because an unreachable destination distorts every number downstream of it.

Common questions

What belongs in a campaign preflight?

The checks that this campaign can actually fail: destination completion, claim accuracy against the record, tracking and consent behavior, and service capacity. Sign each line with a name and a date.

How often should a live campaign be reviewed?

Daily for the first week on a new destination, then weekly once delivery stabilizes. Show absolute counts beside rates, and never act on a rate built from a small denominator.

Who owns the decision to stop a campaign?

One named person per shift, with a documented backup and the authority to pause spend without waiting for approval. Write that into the brief before launch.

What should be deleted after a campaign ends?

Platform logins for people who have moved on, automations tied to the campaign, and republishable creative. Keep the brief, claim record, approvals, source exports and final decision.