Marketing strategy

How Security Scanners Skew Email Open Metrics and How Agencies Can Track Real Humans

Security scanners open and click client emails before any person does. Here's how agencies filter the machines out and report the human signals clients trust.

A man writing business terms like Market Share on a transparent glass wall
Photo by Campaign_Creators

A 58% open rate looks great in a client report. Then the client asks why only eleven people clicked, nobody replied and the booking calendar stayed empty. On a list of business addresses, a real share of those opens came from software that read the email before any person did. Security scanner opens distorted email metrics on B2B lists long before anyone called it a reporting problem, and now that clients look at their own numbers, it's yours.

Here's the argument of this piece. You don't need to give up on opens. You need to change what you promise. Filter machine activity out before any number reaches the client. Show human opens next to the signals bots fake badly, like bookings and replies. Then reset the client's expectations once, on your terms, before a sharp-eyed client resets them for you.

Below you'll find why agencies feel this harder than in-house teams, how to spot scanner activity in your own data, where the filter belongs, which signals deserve the top line of a report, and a short script for the conversation when filtered open rates drop.

Your client's open rate is partly written by machines, and that is now an agency problem

An open is a tiny image loading. That's all it ever was. When the email arrives, anything that fetches that image counts as an open, whether it's a person scrolling on a train or a security filter checking the message for threats.

Two kinds of software do this at scale. Corporate security filters sit in front of business inboxes, fetch the tracking pixel and follow every link to check it's safe, often within seconds of delivery. Mail-app privacy features preload images through a proxy when the message downloads, so the pixel fires whether the person reads the email or not. We've covered the mechanics of how security scanners distort open tracking separately, so this piece stays on what it means for the way you report.

What it means is credibility. Your monthly report is the one document your client reads closely. If its headline figure is partly machine-made, the client eventually sees the gap between opens and everything else. When they do, they don't think "security software." They think "my agency was dressing up the numbers."

Inflated opens are worse for agencies than for in-house marketers

An in-house marketer can shrug at a fuzzy metric. Their boss knows the sales figures anyway, and nobody's contract hangs on the open rate. You sell against that number. You quoted it in the pitch, you put it on page one of the report, and you set targets around it. A noisy number becomes a broken promise. It breaks in three places.

The renewal review where opens look great and revenue doesn't

Twelve months of 50%-plus open rates sound like a winning year. Then the client's finance director asks what email brought in, and the answer is four enquiries. The open rate didn't help you there. It made the gap look bigger, because the client now wonders what else in the report was flattering.

A/B subject line tests won by the wrong variant

Split a list in two and send each half a different subject line. If one half happens to hold more addresses behind aggressive corporate filters, that half wins on opens, whatever the subject line said. You then roll the "winner" out to the full list and write it into the client's playbook. On small B2B lists of a few thousand addresses, a handful of filtered domains is enough to flip a result.

Automations that follow up with people who never opened

"Opened but didn't click" is a popular trigger for a second email. When a scanner did the opening, your client's prospect gets a follow-up that says "saw you had a look at our offer," for an email they never saw. That reads as careless at best and creepy at worst, and it goes out under your client's name.

B2B lists get hit hardest because corporate filters open everything

Why does a client's campaign show opens seconds after it was sent? Because on a business list, the first reader is often the company's security gateway. Email security tools scan inbound mail before it lands. Many of them fetch images and visit every link in a sandbox to check for malware or phishing pages. They do it the moment the message arrives, from data-centre servers, for every recipient on that domain.

A list of accountancy firms, law practices or manufacturers is mostly addresses on company domains, and those companies buy this protection. So a B2B send shows a burst of opens and clicks in the first minute that no group of humans would produce. Consumer lists look different. Personal inboxes rarely sit behind link-scanning gateways, so the distortion there comes mainly from mail-app privacy features that preload images. You'll see opens arriving in batches, sometimes hours after sending, from proxy servers, with no clicks attached. Same inflated open rate, different fingerprint.

Are your open rates fake, and by how much? Nobody outside your platform can give you one honest percentage, because it depends on the list. A list heavy with large firms behind strict gateways shows far more machine activity than a list of sole traders using personal mailboxes. Your own data tells you. Look for these signs in the activity log of any recent campaign.

  • Opens landing within one or two seconds of delivery, often before the send has even finished.
  • Every link in the email clicked at once by the same contact, including the unsubscribe link and the footer social icons.
  • Clusters of opens and clicks from the same data-centre IP ranges across many different recipients.
  • Contacts who opened and clicked but never reached the landing page, filled a form, replied or booked.

What a scanner open looks like next to a human one

How to tell machine activity from a real reader in a campaign's activity log
SignalSecurity scannerHuman reader
TimingWithin seconds of delivery, all at onceSpread over hours and days, peaks in working hours
Links touchedEvery link, often in under a secondOne or two links, the ones that matter to them
Repeat behaviourSame pattern on every campaign, every contact on the domainIrregular, some campaigns opened, some ignored
Follow-on actionsNone, no time on page, no form, no replyPage views, form fills, replies, bookings
Where it comes fromData-centre or proxy addressesOffice or home connections, varied devices

Filtering machine activity has to happen before the report, not in a footnote

You could clean the numbers yourself. Export the activity log, sort by timestamp, strike out the instant opens and the all-links clickers, recalculate. It works once. By the third client and the third month, it's half a day you don't bill for, it's done differently by whoever on your team does it, and the client still logs into the platform and sees the raw figure. Now there are two numbers, and you're the one explaining why yours is lower.

A footnote saying "opens include some automated activity" is worse. It admits the number is wrong without telling the client what the right one is. The filter has to sit in the platform, so every view of the data shows the same cleaned figure.

That's how we built reporting in SendTo. Reports exclude security scanner opens, so the client portal and your partner view both show real human opens, clicks, bookings and the results of each variant. The client sees what you see. There's no export, no manual adjustment and no second version of the truth. If you want the step-by-step on setting this up per campaign, our guide on how to track real human open rates covers it.

That limit is the reason the next section matters. Filtered opens are a better number. They still shouldn't be the number your report leads with.

Clicks get faked too, so pair opens with signals that cost the reader effort

Can security filters fake clicks as well as opens? Yes, and on B2B lists they do it constantly. The same gateway that fetches the pixel follows every link to check where it goes. To your tracking, that's a click. So moving the headline metric from opens to clicks doesn't fix anything by itself. You've swapped one inflated number for another.

The fix is to rank signals by how much effort they cost the reader. A machine fetches an image for free. It doesn't sit on a pricing page for ninety seconds, type its name into a form, pick a Thursday slot in a calendar or write back asking about fees. The more effort a signal takes, the harder it is to fake, and the more a client believes it. Here's the order, from easiest to fake to hardest.

  1. Open. Faked by scanners and privacy preloads on almost every list.
  2. Click. Faked by link-scanning filters, which click everything.
  3. Landing page visit with time on page. Scanners hit the page but leave instantly, so real time on page is a decent filter.
  4. Form submission. Needs a person typing, and a bot-protected form blocks most automated attempts.
  5. Booking. A calendar slot chosen by a named person.
  6. Reply. Written by a human, about something they care about.
  7. Purchase. Money changed hands.

Should you stop reporting opens altogether? No. Filtered opens still tell you whether subject lines and send times are working, and a client who's used to seeing them will ask where they went. Keep them in the report as a diagnostic. Just move them down the page.

Which metric should lead a client email report now? At least one signal from the bottom half of that list. For a firm that sells consultations, lead with bookings per campaign. For an online shop, lead with orders. For a professional services client with a long sales cycle, lead with replies and form submissions. Then show human opens and clicks underneath, as the explanation for why the top number moved.

Reset the client's expectations once, before they find the gap themselves

Switch on filtered reporting and the open rate drops, sometimes by half. If the client finds that drop on their own, in the portal on a Monday morning, it looks like you got worse at your job. If you tell them first, it looks like you caught something nobody else was checking. The number is the same in both cases. What changes the story is who mentions it first.

So have the conversation once, before the first filtered report goes out. Book a ten-minute call for it instead of sending an email. The client will have a question, and you want to answer it while you're both looking at the same chart.

Three lines that explain the drop without sounding like an excuse

Keep it to three points. Why the number fell, why the new one is the honest one, and what you'll report instead. Here's a version you can adapt to your own voice.

  1. "Your next report will show a lower open rate, and I'd like you to hear why from me first. Your contacts' company email systems check every message for threats before a person sees it, and that check used to count as an open."
  2. "We now filter that machine activity out before any number reaches you. The new figure counts people, so it's lower, and it's the one I'd want in front of me if I were in your seat."
  3. "From this month, your report leads with bookings per campaign. Human opens and clicks sit underneath, so you can see why bookings moved. Your bookings haven't changed. We're just counting your readers properly now."

Then stop and let them ask. The usual question is "so were the old numbers wrong?" Answer it straight. They included machine activity, the whole industry counted opens that way for years, and you've stopped. Don't apologise for the old figure and don't blame whoever reported it before you. Give the fact and move on to the bookings.

Priya's client lost 27 points of open rate and kept all 14 bookings

Priya runs a four-person agency serving accountancy firms. Her clients' lists are almost entirely practice and company domains, so nearly every address sits behind a security gateway. When she moved one client, a twelve-partner practice, onto filtered reporting, its reported open rate fell from 61% to 34% overnight. Bookings per campaign stayed at 14, exactly where they'd been for the previous quarter.

She rang the managing partner two days before the report went out. "I told him the 61 was never his readers," she says. "The 34 is people, and the 14 bookings were always real. He asked one question, which was how we get the bookings up. That's the conversation I want to be having." The practice renewed at its annual review, and its report now opens with bookings.

Clean metrics only matter if the segments and automations use them too

A clean report sitting on top of dirty automations leaves half the problem running. If "opened" triggers a follow-up or puts a contact into an "engaged" segment, machine opens corrupt those as well. On a B2B list, an engaged segment built on raw opens fills up with every address behind a link-scanning gateway. You send your client's best offer to that segment, it underperforms, and everyone decides the offer was weak. The offer was fine. Half the segment was software.

Do machine opens break your A/B tests and automations? Yes, until the data feeding them is filtered. Once it is, three segments start doing real work.

  • Clickers who reached the landing page. These are your warmest contacts, so send them the offer with the booking link at the top.
  • Unclicked opens. People who read and didn't act. Give them a different angle or a case study, not a "saw you had a look" nudge.
  • Non-openers. Try a new subject line once, then let them rest instead of pushing them towards a spam complaint.

In SendTo you build these from clicks, unclicked opens or non-openers on any past campaign, and because the activity behind them is already filtered, a scanner can't land a contact in the warm group. Our guide to segments based on campaign interaction history walks through setting them up, so we won't repeat it here.

Bad addresses add their own noise. Dead mailboxes, abandoned role inboxes and old imports full of past unsubscribes inflate the send count and drag every rate around. SendTo verifies addresses before sending and applies platform-wide suppression when you import a list. If you're still deciding whether that's worth the trouble, our piece on automated list cleaning makes the case for B2B lists in particular. Be clear about what it doesn't do, though. A valid address behind a strict gateway still gets scanned. Verification and scanner filtering fix two different problems, and you need both.

Honest reporting is a reason to keep paying you, so build it into the retainer

Picture two agencies at a renewal meeting. The first defends a 58% open rate and hopes nobody asks about revenue. The second says the client averaged 14 bookings per campaign, 34% of real people opened, and next quarter's test is moving the booking link above the fold. The second agency is asking for next year's budget with evidence. The first is hoping nobody looks too closely at the numbers.

So make the honest report part of what the client pays for. Write the lead signal into the scope of the retainer, something like "a monthly report leading with bookings per campaign, human engagement underneath, and one recommendation for next month." The report stops being admin you do at the end of the month and becomes a deliverable with your name on it. Our guide to retainer strategy for B2B email marketing clients covers how to scope and price that kind of retainer.

SendTo lets you run all of this under your own brand. The client portal carries your name, you set what your clients pay, and the sending, warmup on new domains and address checks are handled for you. Your team spends its hours on the campaigns and the report, not on running mail servers. For what SendTo costs you as a partner, see the pricing page.

One trade-off comes with this. Honest reporting doesn't make a weak campaign look good. If a send produces two bookings, the report shows two, with no inflated open rate to soften it. That's the point, but it means the work behind the report has to hold up. Agencies that are confident in their campaigns gain the most from this. Agencies relying on a flattering open rate lose the most.

What to do this week

It's early March, and Q1 reviews are coming up for a lot of retainers. That makes this week a good time to get ahead of the numbers. Four jobs, roughly in this order.

  1. Pull one client's last three campaigns and look for the scanner signs. Opens within a second or two of delivery, every link clicked at once, clusters from data-centre addresses, and contacts with no follow-on action at all.
  2. Decide which hard signal leads each client's report. Bookings, replies, form submissions or orders. Pick one per client and keep it there every month.
  3. Brief each client before their next report goes out, using the three-line script above. A call is better than an email.
  4. Check every automation and segment that triggers on an open. Pause the "saw you had a look" follow-ups until they run on filtered data, then rebuild the engaged segments.

None of this takes more than an afternoon per client, and it means the next time someone asks why the opens don't match the bookings, you've already explained it.

If you're a consultant, creator or affiliate who recommends tools to agencies, you can refer them to SendTo and earn 20% of the platform fee (the partner subscription) each agency you refer pays SendTo, every month that agency pays it. You never take a share of what the agency's own customers pay the agency. Attribution lasts 90 days from the first click, the first touch wins, and commission is paid monthly by PayPal.

Join the SendTo affiliate programme

Frequently asked questions

Why does my client's campaign show opens seconds after it was sent?

On a business list, the first reader is often the company's security gateway. It fetches the tracking pixel and follows every link from data-centre servers the moment the email lands, for every recipient on that domain. No group of humans opens that fast.

Are my open rates actually fake, and by how much?

Partly. Nobody can give you one honest percentage, because it depends on the list. Check your activity log: opens a second or two after delivery, every link clicked at once, shared data-centre IPs and no follow-on actions all point to machines.

Can security filters fake clicks as well as opens?

Yes. The gateway that loads the pixel also follows every link to check it's safe, unsubscribe and social icons included, and your tracking counts each one as a click. If you switch your headline metric from opens to clicks, you've just swapped one inflated number for another.

Which metric should lead a client email report now?

Lead with at least one signal that takes effort from the reader: a form submission, a booking, a reply or a purchase. Machines don't type names into forms or pick a Thursday calendar slot, so clients trust those numbers.

If opens are unreliable, should I stop reporting them?

No. Filtered opens still tell you whether subject lines and send times are working, and a client who's used to seeing them will ask where they went. Keep them in the report as a diagnostic, just further down the page.

Do machine opens break A/B tests and automations?

Yes. If one half of a split holds more filtered domains, it wins on opens whatever the subject line said. And 'opened but didn't click' triggers send follow-ups to people who never saw the email. Audit every automation that fires on an open before your next report goes out.

Can a platform filter bot opens automatically, and how accurate is it?

Yes. SendTo's reports exclude security scanner opens, so the client portal and your partner view show the same human figures. It isn't perfect, though. A person who opens the instant an email lands can get filtered out, and privacy preloads are hard to separate. Treat the figure as a close estimate, not a headcount.

How do I explain a sudden drop in open rate to a client?

Explain it once, before the client spots the gap. Tell them the old figure included security software reading their emails and the new one counts people, then point to the bookings and replies that held steady. Priya's client lost 27 points of open rate and kept all 14 bookings.

Everything this blog teaches, on a platform with your name on it.

Email campaigns, websites, landing pages and print for your clients, made from a short brief and sent under your brand. Create your account and look around with a sample workspace loaded.

No contract · Cancel any time · Explore everything before you pay