Campaigns

Campaigns Overview

A campaign sends one message to an audience. The platform could already send one message to one person; almost all the distance between those two things is the machinery that stops messages going out.

Where to find it

Architect Panel → Commercial:

• Campaign Manager — the console — build, schedule and monitor
• Campaigns — the underlying campaign records

Architect Panel → Automation:

• Tasks — E-mail Campaign Sender, every 5 minutes

Architect Panel → Activity:

• E-mail Log — individual sends
• SMS Log — text sends

Three channels

E-mail, SMS and WhatsApp. The channel is chosen per campaign and determines which account and template are used.

The lifecycle

A campaign moves through six states: Draft, Scheduled, Sending, Paused, Sent and Cancelled.

Paused is the one worth noticing. A campaign that has started can be stopped mid-flight — which is the feature you will be extremely glad of on the day somebody spots an error after the first few hundred have gone.

What a campaign holds

• A name and subject.
• The channel, and the account and template for it.
• A segment — who it goes to.
• A topic — what people are subscribing to or opting out of.
• A schedule, and a throttle in messages per minute.
• Open and click tracking switches.

Sending is a background task

The E-mail Campaign Sender task runs every five minutes and sends the next batch. It ships disabled — enable it under Automation → Tasks, or a scheduled campaign will sit at Scheduled forever looking correct.

Batched sending is also what makes pausing possible and what keeps a large campaign from overwhelming your sending reputation.

The stopping machinery

Before any individual message goes, the platform checks the suppression list and the recipient's topic preference. Those checks are the reason bulk sending is safe to use, and they are covered in their own articles.

It is worth understanding that they exist and are applied per recipient at send time — not when the segment is built. Somebody who opts out after a campaign is scheduled is still excluded.

Every recipient is tracked individually

Each one carries its own status, send time, a link back to the record it came from, and any error. So "did this person get it" is answerable, and a partial failure is visible rather than hidden inside an overall total.

Start small

Send to a segment of yourself and a few colleagues first, on the real channel with the real template. Every campaign problem worth catching — a broken merge field, an unsubscribe link that does not work, a subject line truncated on a phone — is visible in that test and invisible in a preview.

Worked example

A team builds a campaign, sends it to a three-person test segment, finds the merge field for first name is empty for two of them, fixes the segment mapping, tests again, then schedules to the full audience with a throttle. The campaign sends over an hour rather than a minute, and nothing needed pausing.

Recommendations

• Enable the sender task before scheduling anything.
• Always send a real test to a small segment first.
• Know where the pause control is before you need it.
• Set a topic on every campaign so opt-outs mean something.

Building a Segment

A segment is the audience. It names a datastore, the criteria for inclusion, and which fields hold the contact details.

Where to find it

Architect Panel → Commercial:

• Campaign Manager — where segments are built

Architect Panel → Data:

• Datastores — the datastore a segment selects from

What a segment defines

• A name.
• The datastore to select from.
• Criteria — which records qualify.
• The e-mail field and name field.
• A phone field, for SMS and WhatsApp.
• An address source, where postal details are held elsewhere.

Any datastore, not just contacts

A segment can select from whatever you have modelled — customers, members, applicants, tenants, alumni. The only requirement is a field holding an address for the channel you are using.

Map the right field

The commonest segment error is pointing the e-mail field at the wrong column — a secondary address, an account contact rather than the person, an internal owner's address. It produces a campaign that sends successfully to entirely the wrong people.

Check the mapping by looking at a handful of resolved recipients before sending, not by re-reading the configuration.

Criteria are a data quality test

Building a segment usually reveals how good your data is. Records with no e-mail address, obvious test entries, people who left years ago — all of it surfaces the first time you count an audience.

Fix the data rather than working around it in the criteria. A segment excluding "records with 'test' in the name" is a note that your data has test records in it.

Exclude what should not receive anything

Suppression and topic preferences are applied automatically at send time, so you do not need to filter for them. What you do need to think about is business exclusions — customers in dispute, accounts in arrears, people already contacted this week.

Check the count

Look at how many records a segment resolves to before scheduling. A number wildly different from what you expected means the criteria are wrong, and it is far better to find that out now.

An audience of 40,000 when you expected 4,000 is the kind of mistake that cannot be taken back.

Keep segments named clearly

They get reused, often months later by somebody else. "Active customers, UK, opted in to product news" is a segment somebody can use safely; "Segment 3" is one they will guess at.

Re-check reused segments

Criteria that were right six months ago may not be now — a status value renamed, a field repurposed. Before reusing a segment for a significant campaign, look at the count and a few resolved recipients again.

Worked example

A team builds a segment of customers who bought in the last year, expecting about 900. It resolves to 3,400, because the date criterion matched the record's creation date rather than the order date. Caught before sending, fixed in a minute — and it would have been a substantial embarrassment.

Recommendations

• Check the count before every send.
• Inspect a few resolved recipients, not just the configuration.
• Fix data quality rather than filtering around it.
• Name segments so somebody else can use them.

Sending and Throttling

Sending is deliberately gradual. A campaign is claimed in batches by a background task rather than dispatched at once.

Where to find it

Architect Panel → Commercial:

• Campaign Manager — schedule, monitor, pause

Architect Panel → Automation:

• Tasks — E-mail Campaign Sender, every 5 minutes

Architect Panel → Activity:

• E-mail Log — each message and its result

Throttling

The messages per minute setting caps the rate. It is not a performance limitation — it is a deliverability control.

Sending ten thousand e-mails in a minute from a domain that normally sends fifty a day is the single clearest signal of a compromised account, and receiving mail systems treat it accordingly. The result is your messages in spam folders, and a sending reputation that takes weeks to repair.

Spreading the same campaign over an hour or two looks like normal traffic and lands in inboxes.

Choosing a rate

Be conservative on a domain with no bulk-sending history. Start low, watch the results, and increase over subsequent campaigns as your reputation establishes.

If a campaign genuinely must arrive within a narrow window, that is a reason to plan it earlier, not to remove the throttle.

Scheduling

A campaign set to a future time waits at Scheduled until the sender task picks it up. Since the task runs every five minutes, sending starts within that window rather than exactly on the second — fine for marketing, worth knowing if timing matters.

Pausing

A campaign in flight can be paused. Nothing further is claimed, and what has already gone has gone.

Know how to do this before you need it. The moment you need it — a wrong link, a wrong name, a wrong audience — is not the moment to be learning where the control is. Every minute of hesitation is more messages sent.

What to do after pausing

Establish how many were sent and to whom before doing anything else; the recipient records tell you exactly. Then decide whether to fix and resume, cancel, or follow up with a correction — and remember that a correction e-mail to people who received the original is itself a send that needs care.

Reading the outcome

Each recipient carries its own status, send time, a link to the message in the e-mail or SMS log, and any error. So a campaign that reports partial success can be broken down into exactly who did not receive it and why.

Look at the failures on every campaign, not just problematic ones. A rising bounce rate is the earliest sign your data is decaying.

Do not resend to everybody

When something goes wrong, the instinct is to resend the campaign. Send to the people who did not receive it, not to the whole segment — receiving the same message twice is the most reliable way to generate unsubscribes.

Worked example

A charity schedules an appeal to 12,000 supporters at 40 per minute, so it sends over five hours starting at 07:00. Twenty minutes in, someone spots the donation link points at the test site. The campaign is paused with about 800 sent; the link is fixed, those 800 are sent a brief correction, and the remainder go out on the corrected campaign.

Recommendations

• Always throttle — start conservative and build up.
• Find the pause control before your first real campaign.
• Review failures after every send.
• Never resend to the whole segment.

Tracking Opens and Clicks

Campaigns can record who opened a message and which links they followed. Both are switches per campaign, and both have caveats worth understanding.

Where to find it

Architect Panel → Commercial:

• Campaign Manager — the tracking switches and reporting

Architect Panel → Security:

• Cookie Consent — your wider consent position

How click tracking works

Links in the message body are rewritten to pass through the platform, which records the click and forwards the visitor. Each click records the campaign, the link, the recipient, the time, the address and the browser.

Clicks are reliable; opens are not

A click is a real action by a real person. Open tracking depends on an image loading, and a great deal now interferes with that: mail clients that block remote images by default, privacy features that pre-load every image whether or not anybody looked, and corporate scanners that fetch everything.

So treat opens as a rough directional signal at best. A campaign with 60% opens and 0.5% clicks did not engage anybody. Judge campaigns on clicks.

What to do with the data

The useful questions are comparative rather than absolute: did this subject line do better than that one, which link in the message drew interest, which segment responds.

Absolute rates are hard to interpret and vary enormously by sector and list quality. Your own trend over time is worth more than any benchmark.

Link rewriting changes the URL

Recipients see a platform address rather than the destination, and some people check before clicking. It is worth being aware of, particularly for audiences trained to be cautious about links — and it is a reason to send from a recognisable domain.

Tracking is personal data

A record that a named individual opened a message and followed a link, from an address, at a time, is information about them. It is not anonymous analytics.

That means it belongs in your privacy notice, needs a retention period, and should be restricted like other personal data. Enabling tracking because it is available, without deciding any of that, is the common failure.

Consider whether you need it

If nobody will act on the data, do not collect it. Tracking that nobody reads is pure liability — you hold behavioural data about your contacts and get nothing for it.

Where you do use it, being able to say what it is for is both good practice and a good discipline.

Be careful using it as a signal about individuals

Treating an open as interest and acting on it — a sales call because somebody opened an e-mail — is a step beyond measuring a campaign, and people find it uncomfortable when they realise. Given how unreliable opens are, it is also frequently wrong.

Worked example

A team enables click tracking and leaves open tracking off, on the grounds that opens would not change any decision they make. They compare click rates across subject lines over six campaigns, find one style consistently better, and set a twelve-month retention on click data. Their privacy notice says what is recorded and why.

Recommendations

• Judge campaigns on clicks, not opens.
• Do not enable tracking you will not act on.
• Set retention and cover it in your privacy notice.
• Compare against your own history, not external benchmarks.