- Grow 📧 Monetize
- Posts
- 📩 Half your opens are machines
I am finishing this issue from an airport lounge, waiting on a flight out.
Tomorrow I am speaking at a publishing accelerator, a room full of the sharpest financial newsletter operators in the business. So this is getting written and pushed out somewhere between the first boarding call and the last.
Fitting, though, that the topic I am putting in front of that room tomorrow is the same one I am putting in front of you right now.
The open rate you have been reporting is not what you think it is.
Limited space remaining: Accredited investors can earn an 11.5% fixed annual return, paid monthly, over a two-year term. At the end, convert your investment into Groundfloor equity at a discount, roll your capital over into a new high-yield Convertible Note, or withdraw funds. Non-accredited investors can access a 12-month 8.5% fixed rate Note, or shorter term 1-and 3-month Notes.
Groundfloor has offered Notes since 2017 and has paid every Note investor 100% of principal and interest owed, in full and on time. This offering has limited space and is funding quickly. Use NOTES100 for a $100 bonus on your first investment over $1,000. Lock in your spot and start earning now.
*Ad
The confession nobody printed
Here is the thing. Quietly, without a press release, the major email platforms conceded the argument.
They shipped new metrics. Machine opens. Human opens. Verified opens. Non-MPP opens.
Every one of those is a confession. If your open rate were real, none of them would need to exist.
The number sitting in your dashboard is a blend. And once you see what is blended into it, you cannot unsee it.
What actually lands in your dashboard
Picture your newsletter hitting an inbox.
Apple Mail Privacy Protection preloads your tracking pixel on Apple's servers, whether the subscriber ever looks at the email or not. Security scanners at corporate domains open and probe every message before a human sees it. Image proxies fetch your pixel as routine processing.
Every one of those events shows up as an "open."
Apple Mail alone now accounts for more than half of all tracked opens across the industry. Roughly half of what your dashboard calls engagement is machines talking to machines.
And it cuts the other way too. Images-off readers and plain-text loyalists read every word you write and register nothing.
Two distortions running at once. The dead get counted. The living get missed. That is opener bloat, and it is structural, not an edge case.
Stop typing what you could say in 10 seconds.
Wispr Flow turns your voice into clean, professional text inside any app. Emails, Slack, client updates — speak once, send without editing. 4x faster than typing.
*Ad
The metric switch nobody announced
The platforms know all of this. So they started splitting the open into tiers. Here is where the major ones stand, and what the metric is actually called in your dashboard.
Braze draws the line out loud. Reports can include or exclude machine opens, and its "Estimated Real Open Rate" goes further, statistically modeling human opens from click data. Notice what Braze trusts to model reality. Clicks. Even the open-rate fix runs on click signal.
Klaviyo tags every open event with an Apple Privacy Open flag, true or false. You can filter it in segments, exclude it from engaged-audience definitions, strip it from conversion tracking. It takes configuration, and the default view is still the blended number. Read Klaviyo's own docs closely, though, and you find the most honest sentence any ESP has published. When MPP is enabled, there is no way to distinguish a true human open from an automated one. Sit with that.
Beehiiv filters most auto-generated opens, the MPP preloads and image prefetching, out of your reported rate by default. No configuration required. It runs the same logic on the metric that actually matters, too. Verified Clicks strips bot and scanner clicks using IP, user-agent, and click-pattern signals, and shows you the raw and filtered numbers side by side.
Customer.io has the cleanest split in the business. "Human Opened" and "Machine Opened" sit next to each other as separate metrics. MPP, Gmail prefetch, bots, and scanners firing within seconds of delivery all land in the machine column.
Brevo bakes MPP and bot opens into your open rate by default. There is a settings toggle that pulls them out of reports and segmentation. Most senders have never flipped it.
Iterable filters bot clicks out of campaign analytics by identified user agents. Opens, though? Still raw. Half the job.
Optipub reports human opens as the headline number, with MPP opens and bot opens each broken out as their own visible metric. One of the clearest breakdowns of opens on the market.
SendGrid leaves machine opens in the dashboard. But the event webhook flags every suspected MPP open with a boolean. The honest data exists. You just have to parse it yourself.
The 2-minute rule
There is one filter no ESP will hand you, and it is the sharpest one there is.
Timestamp math.
An open that fires within two minutes of delivery is not a human. Nobody sees your email, opens it, and renders your pixel inside the first 120 seconds at scale. That is a security scanner. A proxy prefetch. MPP doing its thing.
Your ESP dashboard will not show you this. But if you send through an API or SMTP layer, SendGrid, Mailgun, SparkPost, the raw event data is already yours. Delivered-at timestamp. Opened-at timestamp. One subtraction.
Anything under two minutes goes in the machine column.
It is the closest thing to a universal human-open filter that exists, and it works on any platform that exposes event-level data. The ESPs building "human open" metrics are running variations of this exact logic. You can run it yourself.
Stop being the middleman between your own finance tools.
Most finance teams work on five different tools and a prayer. Ramp replaces all of it: corporate cards, bill pay, expense management, travel, and procurement. AI-powered with real-time visibility and real control.
*Ad
"Human open" is still a euphemism
Here is the part the feature announcements skip.
A "human open" does not mean a human opened your email. It means the ESP could not prove a machine did.
An MPP-flagged open might still be a real reader on an iPhone. A non-MPP open might be a corporate scanner the filter never caught. Filtered opens are less wrong. They are not right.
Which is why the decision framework has not changed. It has just been vindicated.
Use filtered opens as a deliverability canary. A sudden ten-point drop in human opens means something broke, inbox placement, authentication, reputation. That is what the metric is for.
Never use opens, filtered or not, to decide who stays on your list.
The final call belongs to clicks
That last decision belongs to clicks. A click cannot be preloaded.
Yes, the experts will tell you bot clicks run rampant. Here is the truth. If your list leans on top-level domains like Gmail or Yahoo, you see far fewer click bots, and mostly real clicks get reported. So for all intents and purposes, a click is a hand raised.
It is the reason the Base Sending Segment requires a click on record for anyone past the new-subscriber grace window. Not an open. Not a "human open." A click.
Your engaged audience density is measured in clicks. Your sponsors are buying clicks whether they know it yet or not. Your inbox placement is earned with clicks.
The platforms just spent four years building metrics that prove the point.
So this week, do two things. Check what your platform calls its filtered open, and switch your reporting over to it. Then stop making final decisions with it. Reply and tell me what your machine-open share turned out to be. I have a feeling it is higher than you want it to be.