First, check clicks and provider splits before trusting the open rate
The open rate is the least reliable figure in an email report. An open is recorded when a tracking image loads, and some mail clients load images for the reader whether or not anyone looks, while others block them, as the email metrics and testing guide explains. So start with the figure that cannot be faked by a machine: unique clicks per delivered message. If clicks held while opens fell, treat the drop as a measurement question first. If both fell, something real happened.
Next, split opens and clicks by mailbox provider for the sends either side of the drop. A fall concentrated at one provider while the others hold steady is placement at that provider. A fall spread evenly across providers is either audience or message. Then date the drop against the send log and the list's growth: a new sign-up source, an import, a platform migration, a template rebuild or a change of sender name in the same week is usually the explanation. Work through the table with those three readings in hand.
Diagnosis table
Diagnosis
| Symptom | Check | Likely cause | Fix |
|---|---|---|---|
| Opens shift sharply while clicks stay level | Whether the platform began filtering machine opens; open rate by mail client | Privacy features pre-fetching images, so opens no longer mean reads | Report on clicks; label opens approximate; move open-based triggers to clicks |
| Opens and clicks fall at one provider only | Engagement by provider; that provider's postmaster view | Placement moved to spam at one mailbox provider | Diagnose placement and fix its cause before changing content |
| Rate falls as the list grows quickly, with new subscribers rarely opening | Opens and clicks by sign-up source and join date | List growth from a low-intent source diluting the rate | Fix the source's offer and wording; segment its subscribers until they engage |
| Gradual slide that steepens, every send to the whole list | Share of recipients with no click in a long window | Inactive subscribers left on the send list | Introduce engagement tiers; re-permission and suppress the inactive tier |
| Drop begins with a new subject style, sender name or send day | The change log for the sends either side of the drop | Subject line, sender name or send time changed | Revert or test the change properly, judged on clicks |
| Opens fall to almost nothing, or vanish for one template or flow | Tracking settings; whether the pixel is in the sent message | Open tracking removed or broken by a template or platform change | Restore tracking; annotate the gap in reports |
Fix, verify and prevent each cause
The counting changed, not the readers
Pre-fetching inflated opens for years, so the visible change is often a platform that begins excluding machine opens, a report that switches to a filtered figure by default, or a shift in the client mix of the list. Any of these produces a step change in the open rate with no change in behaviour.
Fix: confirm what the platform now counts and from when, then make unique clicks per delivered message (CTRShare of impressions that result in a click.) the headline engagement figure and label opens as approximate, as the metrics and testing guide recommends; move any re-engagement trigger or segment that depends on opens to clicks or site visits.
Verify: the click trend across the date of the change is continuous, and the moved triggers still catch genuinely dormant subscribers.
Prevent: annotate every reporting change on the dashboard so a step in the line is never read as behaviour.
Mail moved to spam at one provider
When a provider starts filing mail as junk, opens and clicks from its users fall together, while the platform still reports the mail as delivered. The other providers carry on as before, which is the giveaway.
Fix: open that provider's postmaster view, where Google's sender guidelines point senders for spam rate and reputation, and work through the placement causes in emails going to spam; the mechanisms behind them are in the deliverability guide. Do not change subject lines while placement is the cause.
Verify: engagement at the affected provider returns to its level relative to the others.
Prevent: keep a provider split in the weekly report, so the first send affected is noticed rather than the fifth.
A new source of subscribers who do not open
Fast growth from one channel can lower the rate without anything going wrong for existing readers. Competition entries, discount pop-ups and co-registration tend to bring people who wanted the prize or the code, not the newsletter.
Fix: compare opens and clicks by sign-up source and join month; if one source is far below the others, rewrite its offer and wording so it promises the newsletter, and keep its subscribers in their own segment until they engage, as the list building and consent guide advises for each placement.
Verify: older cohorts show their usual engagement once the new source is reported separately, and new sign-ups from the reworked offer engage closer to the rest.
Prevent: judge every new sign-up placement on the behaviour of the subscribers it produces, not only on how many it collects.
Inactive subscribers still on every send
A list that is never pruned carries more non-readers each month, and the open rate drifts down until a seasonal whole-list send makes the slide visible all at once. Those inactive addresses also weaken the engagement providers read, so the problem can tip into placement.
Fix: define engagement tiers by recent clicks, send the inactive tier a short re-permission sequence and suppress those who do not respond, following the segmentation guide; M3AAWG's sender practices treat list hygiene as part of responsible sending.
Verify: the open and click rates on regular sends recover as the delivered volume falls, while total clicks stay roughly level.
Prevent: refresh tiers automatically and make the send list a decision for each campaign.
A change in subject, sender or timing
Sometimes the drop is the message. A new subject line style, a sender name changed from a person to a brand, a preheader left at the template default, or a move to a different day can each reduce how many people open. Because opens are unreliable, this is the cause to consider last.
Fix: find the change in the log, revert it, or test it properly with a split large enough to read, judged on clicks; the newsletter design and copy guide covers subject lines and preheaders, and its advice against tricks applies here.
Verify: a fair A/B TestControlled comparison of two variants. or a return to the previous style restores clicks on the following sends.
Prevent: change one element at a time and record each change with its date.
Tracking that stopped working
A template rebuild, a move to a new platform, or a setting switched off during a privacy review can remove the tracking image or disable open tracking for some messages. Opens then fall to almost nothing for the affected template or flow, an abruptness no real audience change produces.
Fix: send a test, inspect the received message for the tracking image, check the platform's tracking settings at account, template and flow level, and restore them, as the metrics and testing guide describes for the delivery and engagement layers.
Verify: opens reappear on the next send at a level consistent with clicks.
Prevent: add "tracking present" to the pre-send checks for every new template and migrated flow.
Prevention: click-based reporting and provider-level monitoring
Most false alarms about opens disappear once the report is built differently. Lead with clicks per delivered message, conversions and revenue per email, and show opens below them with a note that they are approximate. Report each of those by mailbox provider and by sign-up source, so a placement problem at one provider or a weak new source shows as a single line moving rather than as a mystery in the total. Keep a change log beside the dashboard for template rebuilds, platform settings, sender names, send times and list imports, and mark each on the chart.
Choose the key indicators (KPIA metric tied directly to an objective.) deliberately: click-through on the main call to action, the active share of the list, and complaint rate by provider are harder to distort than opens and more closely tied to whether the programme works. Review the inactive tier monthly, and read the postmaster view for the largest providers weekly. With those habits, a falling open rate becomes a prompt to look at three other numbers that already explain it, rather than a reason to rewrite the next subject line.
A charity's newsletter opens fall by a third
A charity's monthly newsletter goes to 36,000 supporters and has shown an open rate around 42% for a year. In September it reports 28%. Clicks per delivered message are 3.1%, against 3.0% in August. The platform's release notes show it began excluding opens it identifies as machine-generated on 1 September. Split by provider, every provider fell by a similar proportion, and no new sign-up source appeared that month. The team annotates the dashboard at the date of the change, moves its re-engagement flow from "no open in six months" to "no click in six months", and switches the board report to clicks and donations per email. The October issue reports a 29% open rate and a 3.2% click rate, and the re-engagement flow now targets 9,400 supporters rather than 4,100, because it no longer counts machine opens as engagement.
If inactive addresses are diluting the list, the inactive subscriber sunset test measures what suppressing them costs in revenue against what it saves in complaints.
The comparison of open rate and click rate explains which tests and reports each figure still belongs in.
Common mistakes
Rewriting subject lines before checking clicks. Reading a total when one provider moved. Comparing open rates across a reporting change. Triggering re-engagement on opens. Celebrating fast list growth without reporting by source. Testing subject lines on opens alone.
