by Albright Labs
Most campaign problems are handled by finishing the send and doing better next time.
That works for a weak subject line or a mediocre click rate, where the cost is confined to the campaign. It does not work for the two failures that damage the sending domain, because those costs are carried by every campaign you run afterward.
Knowing which kind of problem you have is the whole skill.
Stop Immediately: Bounce Rate Above Threshold
Pick a threshold, write it down, and treat it as a stop rather than a signal to slow down. We use three percent.
A bounce tells the receiving provider you do not know who you are mailing. Sustained high bounce rates are read as evidence of a purchased or stale list, and the reputation damage attaches to the domain. Your own sending provider watches the same number: Amazon SES, for example, puts an account under review at a 5% bounce rate and may pause its sending at 10%.
The instinct at this point is to reduce sending volume and continue. That is the wrong move, because the problem is the list rather than the pace. Sending the same bad addresses more slowly produces the same bounces over a longer period.
Pause, fix the list, resume. Not throttle.
Stop Immediately: Complaint Rate Rising
Complaints are less forgiving than bounces and the thresholds are much smaller. Google asks senders to keep the spam rate in Postmaster Tools below 0.1% and never reach 0.3%. Amazon SES reviews an account at a 0.1% complaint rate. A rate that looks like a rounding error on a dashboard can be enough for a provider to act.
The difference from bounce rate is diagnostic. Bounces mean bad addresses. Complaints mean real addresses and unwelcome mail, because complaint rate tracks how irrelevant the message was to the person who received it.
That distinction matters because the fixes are different. A bounce problem is solved by verification. A complaint problem is solved by targeting, meaning who gets the mail and whether they still want it, and no rewrite of the email will address it.
Do Not Stop For: Low Engagement
A disappointing open or click rate is information rather than an emergency.
It does not damage anything, it is measured over days rather than hours, and stopping early means you never learn how the campaign would have done. Engagement arrives slowly and unevenly, and judging a campaign on its first morning measures impatience.
Let it finish, then diagnose.
Do Not Stop For: A Rough First Day
The first hours of a send are unrepresentative.
Automatic replies arrive immediately and skew the picture. Bounces from the worst addresses cluster at the start. Most real engagement has not had time to arrive, because most of the people who will read it have not read it yet.
The exception is bounce rate, which is meaningful immediately and is exactly why you watch it on day one.
The Diagnostic Order
When something is wrong, check in this order. Each step is cheaper than the next.
1. Did it actually arrive. Send to your own inbox on the same domain and look. Check it on a phone. If the message is landing in spam, every downstream number is measuring the wrong thing, and no copy change will help.
2. Check the destination of every link. A landing, signup or donation page that is down turns the strongest email you send into a broken promise. Certificates expire quietly. Check before each batch rather than once at setup.
3. Look at the bounce composition. Hard bounces, where the address does not exist, are a list problem. Soft bounces and deferrals, where the receiving server is temporarily refusing or slowing mail, point to a reputation or volume problem, and they mean something different.
4. Read twenty of the replies, starting with the unhappy ones. This is the fastest diagnosis available and almost nobody does it. People tell you why the email missed, and the reasons cluster. Too often, not what they signed up for, not for them any more, or the message did not make sense.
5. Check the calendar. Was this the right month for this audience. A contractor in peak season, a tax practice in filing season, or a firm in the first week of January will not read anything, and the campaign was doomed before the copy was written.
6. Only then look at the copy. It is the most edited and least causal part of a campaign.
The Failure That Looks Like a Copy Problem
A campaign sent in the wrong season produces the exact symptoms of a bad email: low engagement, higher complaints, more unsubscribes.
Rewriting the email in response produces a second failed campaign and a false conclusion about the copy. Before touching anything, check whether the send window was right, and hold the campaign rather than sending anyway if it was not.
What to Do During a Pause
A pause is not a wait. It is the point of the exercise.
Re-verify the list and remove everything undeliverable. Remove role addresses such as info@ if the complaint rate was the trigger. Check for duplicate records sending the same campaign to one person twice. Confirm every link resolves. Then resume with a smaller batch of your most engaged subscribers and watch the same two numbers before opening the volume back up.
Resuming a paused campaign at full volume onto the same list is how a recoverable problem becomes a permanent one.
Write the Thresholds Down Before You Need Them
The decision to pause is made badly in the moment, because in the moment there is a campaign running and a number that someone wants to hit.
Put the thresholds in the campaign document ahead of time, along with the instruction to pause rather than throttle. A rule written in advance survives the pressure of a live send. A judgment call made during one usually does not.
A Simple Rule of Thumb
Stop for anything that damages the domain. Finish anything that only disappoints you.
Bounce and complaint rates are the first kind. Everything else is the second.
Final Thoughts
The cost of pausing a campaign is a delay. The cost of not pausing one is a sending domain that every future campaign has to share.
Watch two numbers daily, write the thresholds down before the send, and diagnose in order of cost rather than starting with the copy.
The list side is in The List Work That Protects Your Sender Reputation.
In Sendly, each campaign’s page shows its bounced count next to sent and delivered, and when your server rejects a recipient as unknown, that email is recorded as failed rather than retried on another server. The bounce number is there to check on day one.