Your open rates collapsed after a platform change

Separate a measurement change from a real delivery problem after migrating email platforms so you do not fix a number that was never broken.

The number halved and nothing else changed

You moved email platforms last month. The first campaign on the new system reported an open rate well under half of what the old one used to show. Nobody changed the list and nobody changed the writing and somebody senior now wants to know what went wrong with the migration.

Two very different things produce this symptom. Either fewer people are seeing your mail or the same people are being counted differently. Those demand opposite responses and telling them apart takes a day. Acting before you have is how businesses undo a migration that was working.

What the investigation costs and what you cannot recover

You cannot recover a comparable history. Open counts from two platforms were produced by different rules and no adjustment makes last year's figures sit honestly beside this quarter's. Accept the break in the series and mark the date in your reporting rather than pretending the trend line continued.

The investigation itself costs a day of somebody's attention and no cash. Checking tracking configuration and comparing the two platforms' counting rules and running a placement test are all within reach of whoever administers your email. Where you bring in outside help the fee attracts GST at 18% as a service.

What may cost money is the real cause if it turns out to be one. A new sending domain has to earn trust and that is weeks of restrained volume rather than a setting you can switch.

Who should stop reading here

If opens were never an input to a decision in your business you can close this question in ten minutes. Confirm clicks and orders held up. If they did then your email is working and you have lost a vanity number that was unreliable in both platforms. Some businesses spend a fortnight investigating a metric they have never once acted on.

You cannot skip it if open behaviour drives logic. Automations that branch on whether somebody opened and re-engagement windows and lead scoring models all consume that signal. When the counting rule changes underneath them those sequences start treating engaged people as dormant and that does real damage quietly.

You also cannot skip it if the migration moved you to a new sending domain or a new IP range. In that case a genuine placement problem is plausible and it is the one explanation that gets worse while you wait.

How to tell measurement from delivery

Look at clicks first. An open is recorded when an invisible image loads and a click is recorded when a person acts. If click rate and orders held steady while opens fell then your mail is arriving and being read and only the counting changed. A fall in both points to delivery.

Then compare the counting rules. Platforms differ on whether they report unique or total opens and on the window used to deduplicate and above all on whether they discard machine opens. Mail privacy features on some clients pre-fetch images which registers an open nobody performed. A platform that filters those will always report a lower figure than one that does not and the lower figure is the more truthful one.

Then check the mechanics of tracking itself. Open tracking runs through a tracking domain and a migration usually changes it. Confirm the new tracking domain resolves and is served over a valid certificate and is not blocked. Finally confirm your suppression list came across. An unsubscribed contact who starts receiving mail again complains and complaints damage placement.

What to measure instead

Open rate is the flattering metric and this episode is a reasonable moment to demote it. It was inflated before by automated image loading and it is only approximately a measure of human attention now.

Click to order rate and revenue per thousand delivered are the numbers worth tracking because neither can be triggered by a machine pre-fetching an image. Keep one delivery indicator beside them which is the spam complaint rate your mailbox providers report. Those three survive a platform change with their meaning intact which is exactly why they are better instruments than the one you lost.

The common mistake

The expensive error is reacting to the drop by sending more. Reasoning that fewer people are seeing the mail so volume should rise is understandable and it is the worst available response if the cause is a new sending domain still establishing trust. Increased volume from an unproven domain is the pattern providers filter hardest and a measurement artefact becomes a genuine delivery failure within a fortnight.

The second error is quieter. Automations that branch on opens are left running through the change. A re-engagement sequence calibrated to the old counting rule now classifies a large part of an active list as dormant and begins suppressing or win-back messaging your best customers. Nobody notices because the campaign reports look normal. Freeze open-based branching until you know which explanation you are dealing with.

This guide explains how something works. It is not legal, tax or financial advice, and rules change. Confirm anything regulatory with a qualified professional before you act on it.