Step 3 · Launch & measure
What you'll do
Launch the campaign, monitor the first hour for anything wrong, then read causal lift in the report 48–72 hours later.
Launch checklist (last pre-flight)
Before clicking Launch, run through this:
| Check | What you want |
|---|---|
| Approval status | Approved (Agency tier) / not required (other tiers) |
| Recipient count | Matches segment − holdout − fatigue skips |
| Test send | At least one sent to your own email; it landed in inbox (not spam) |
| Render preview | Looks right on mobile and desktop |
| Tracking | Open + click + conversion tracking enabled (default) |
If everything is green: Launch.
What happens in the first hour
The system processes sends in batches — typical send rate for an email campaign is ~50K/hour per shop, scaled up on higher tiers. WhatsApp + SMS are slower (rate-limited per Meta / Twilio quotas).
You'll see live counters tick up:
- Queued → Sent → Delivered → Opened → Clicked → Converted
Two things to watch in the first 30 minutes:
- Bounce rate spike. If bounces exceed 5%, you have list-hygiene issues — investigate the bounce reasons.
- Unsubscribe rate spike. If > 1% of recipients unsubscribe in the first hour, the message was off-brand or the audience was wrong. Don't panic — pause the campaign, fix, relaunch on a fresh segment.
Read the report — 48 hours later
48 hours is the sweet spot for first-read: most opens + clicks land in the first 24h, and conversions trail by a day. Open the campaign in Sales engine → Campaigns → [your campaign] → Report.
You'll see:
| Section | What it tells you |
|---|---|
| Funnel | Sent → Delivered → Opened → Clicked → Converted. Where customers drop off. |
| Per-channel | If you sent multi-channel, which channel drove the conversion. |
| Holdout comparison | The number you came for. |
Read the holdout (causal lift)
The holdout report has two numbers:
- Treatment group revenue — what the recipients of the campaign spent in the 7-day window after send.
- Holdout group revenue — what the 10% who didn't receive the campaign spent in the same window.
If treatment beats holdout by a margin larger than statistical noise, the campaign caused the lift. The system reports the lift as a percentage with a confidence band: e.g. +18% lift, 95% confidence.
Common findings on a first campaign:
| Pattern | Meaning |
|---|---|
| Big positive lift (e.g. +20%) | Audience + message fit. Run it again. |
| Small positive lift (e.g. +3%) | Audience right, message weak. Iterate the copy. |
| No lift / slight negative | Audience wrong, or recipients would have bought anyway. Pick a more responsive segment. |
| Lift but with high unsub | Reached people, but wrong message. Tighten audience or soften message. |
Decide what's next
Three productive responses to the first campaign report:
- Run a variation — keep the audience, change one variable (subject line, send time, offer). Measure the lift.
- Expand the audience — if the lift was strong, broaden the segment 2–3× and re-send to the new entrants.
- Move up the funnel — convert this campaign into a journey (Journeys & automations) so it fires automatically on new customers entering the segment.
What you have now
A complete first campaign, measured with causal lift — not a guess about whether it worked.
Tutorial complete 🎉
You've covered the full loop: audience → message → launch → measure. Repeat this loop weekly; the team's data-driven muscle compounds fast.