15 Surprising Email A/B Test Results and What We Learned From Them

I ran an A/B test for a B2B SaaS client in the HR space where we expected the "clean" version to win. Version A had a short, polished intro and a clear button. Version B looked more like a plain text note from the founder, with one link in a normal sentence and a small P.S.

Version B won by a bigger gap than I expected. Opens were close (about 34% vs 33%), but clicks were about 2.1% vs 1.2%, and demo bookings from that send were 18 vs 9 off a list of roughly 12,000. The only other change was we swapped "Book a demo" for "See if this fits your team", which seemed to lower the pressure.

I learnt that "more professional" isn't the same as "more trusted", even in B2B. Since then, I test one "human" element at a time (sender name, plain text format, softer CTA, P.S.) and keep everything else locked, so I know what moved the needle. Others can copy that approach and stop guessing what their audience reads as salesy.

We ran an A/B test on our monthly newsletter at Doggie Park Near Me that completely changed how I think about email subject lines.

Version A used a promotional subject line: "Spring Special: 20% Off Monthly Memberships This Week Only." Version B used a personal, curiosity-driven subject line: "You won't believe what Max the golden retriever did at the park yesterday."

Version B outperformed Version A by 47 percent in open rates and generated three times the click-throughs to our website. The surprising part was that Version B didn't even mention a promotion. It was pure storytelling. Once people opened the email to read about Max, they scrolled past the story right into our membership offer, which was placed naturally at the bottom.

The lesson was that our audience doesn't open emails for discounts. They open emails for their emotional connection to dogs and the community we've built. Promotional language actually triggered their spam instincts and caused them to scroll right past it.

What I'd recommend to anyone running email campaigns: test a subject line that tells a micro-story or sparks curiosity about a real person or event. People are drowning in promotional emails. The ones that get opened are the ones that feel like a message from a friend, not a billboard.

We conducted a tightly optimised A/B test on a reactivation cohort of inactive car finance claims leads. The control was our standard journey - high frequency, personalised emails with heavily optimised creative content. The variant was cut by over half in terms of send frequency, with predominantly plain-text, compliance-first messaging. Strict suppression rules were used to limit overlap between cohorts, and both were attributed for intent signals and age of data. We assumed more touchpoints and better user experience = better conversion.

In fact, the reduced-frequency, plain-text variant drove better engagement, more claim starts completed, and had a materially lower unsubscribe rate. We found that with claims, perceived intent was more important than creative optimisation. Customers were more trusting of our brand and less sceptical about the legitimacy of the claim thanks to a compliance-led tone we usually think of as a creative limitation. Claims are a sensitive topic and customers tend to question why a brand is contacting them about potential claims. This also fed through to frequency. With less messages going out, those that were landed caused less noise and prompted a response when the customers were ready. This had operational benefits too - we've since aligned our send strategy with our ops capacity to prevent creating spikes in demand we cannot service within compliance, and have adapted plain-text, trust-first messaging for mid- and low-intent journeys.

The danger here is taking this lesson as "always less is more". The test results only applied because we correctly isolated both cohorts to remove cross channel contamination, and also because car claims emails sit within a trust-sensitive, regulation-heavy category. Removing richness from your email programmes blindly will only reduce your volume. Email in both claims and auto finance acts as a credibility filter - too much optimisation and you risk filtering yourself out.

We were 3 days into a subject line test when I almost killed the variant that ended up winning. The test pitted our standard benefit driven subject line against a version that read like a calendar reminder. Something like Tuesday check in on your investor search. No urgency, no hook.

The reminder style beat the original by 41% on open rate. Click through was up 28%. We ran it again the following week with a different segment and got nearly identical numbers.

What I think happened is that founders get so many pitch style emails that anything resembling a personal follow up cuts through. That means the emails we spent the most time crafting were actively working against us. I have not fully resolved what that implies for our overall email strategy.

When creating an A/B test for regional segmentation with dedicated offers and everything, I still found that activity rates mainly drove sales. To make this clearer, we created a test in which we split the newsletter audience into North and South Germany for a travel company. In the different newsletter versions, we had offers that were in the north and south, respectively. We expected higher open and especially click rates with this approach. Overall, the newsletters performed similarly to a non-segmented one. But what we could clearly see was that subscribers who generally had higher click rates also clicked more here. We have seen this in multiple cases that over-segmentation isn't always helping. From our experience, activity level and the last product (group) bought are the only relevant segmentation criteria.

We ran an A/B test on email length in one of our campaigns. We tested a dense email with multiple content blocks against a short version with one clear message and one primary click path. We assumed the longer format would perform better because it gave more context and options overall. The shorter version won by a wide margin in clicks and revenue per recipient.

This result reinforced a lesson we now use across our campaigns for our team. In a crowded inbox we learned that clarity beats completeness every time in practice. When people have too many choices they often do not take action. We now test subtraction before adding more content and we keep messages simple for better results.

We tested plain text emails against our designed HTML templates for a fashion e-commerce client. The plain text version outperformed by 340%.

Background: the client was sending polished, branded emails with custom headers, product photography grids, and stylized CTAs. Open rates averaged 18%. Click-through sat at 1.2%. Standard for the industry but nowhere near profitable for their 12,000-subscriber list.

The test: we sent the exact same offer (25% off selected items, 48-hour window) to two equal segments. Group A got the usual branded template. Group B got a plain text email that looked like it came from a friend. No images. No HTML formatting. Just text with a single link.

The results caught everyone off guard. Plain text: 34% open rate, 5.3% click-through. HTML template: 19% open rate, 1.4% click-through. Revenue from the plain text segment was 340% higher.

Why it worked: email providers increasingly filter heavily designed emails into Promotions or spam tabs. Plain text lands in Primary. Gmail treated our plain text email like a personal message. And the psychology is different too. A designed email signals "this is marketing." A plain text email signals "someone is talking to me."

The lesson we took away: we now send the first email in any sequence as plain text. Welcome email, abandoned cart recovery, post-purchase follow-up. The opener is always plain text. We save the branded HTML for the third or fourth touchpoint when the subscriber already knows and trusts us.

One caveat: this works best for offers and relationship-building emails. Product launches with visual elements still benefit from good design. But for anything where the goal is a click or a reply, plain text wins consistently across every client we have tested it with.

I'm always amazed at the little changes that will sometimes affect results. We had one campaign for a new business where simply putting the word "SALE" in all caps yielded a 10% higher open rate than the same subject line without the caps. This wasn't exactly intuitive to me. I've seen the opposite result, too, where too much urgency can turn people off.

We ran a simple A/B test in our newsletter after polling our audience: one version emphasized practical case studies and marketing wins, the other kept our previous, more general content. The version with real-world examples produced our highest open and click-through rates. The surprising part was how clearly audience preference for concrete examples translated into measurable engagement. The takeaway for others is to let reader feedback guide content choices and to validate those choices with a straightforward A/B test measuring opens and clicks.

The most shocking A/B test we ever ran, and why outrage metrics aren't real. This one was an email A/B test aimed at insurance agencies. Version A was a normal email campaign, and Version B was high outrage, high "industry outrage" based on a series of recent controversial compliance changes.

The clickthrough rate on Version B was great, something like 2.1% to 6.8%, but the amount of booked demos at the bottom was completely unchanged.

What happened? It turns out when you dig into the server logs, what you find is that around the peak the email sends, like 70% of the clicks on the Version B come in within 3 seconds, come from duplicated IP clusters, etc. Basically, they're automated clicks coming from the email security bots and stuff that are scanning the links, triggered by the outrage keywords.

We learned quickly that outrage generates bot clicks, and if you act on the data, you're putting your business at great risk. This is, of course, what we've seen in the media continually, as the Wall Street Journal recently reported on a major restaurant brand's rebranding backlash: nearly 50% of the outrage was automated, and the brand pulled its stunt and lost 10.5% of its stock value.

The same thing is happening in your email A/B tests. If you act on the initial data, you erroneously teach the algorithm and your marketing team that outrage works.

One must incorporate bot detection into the optimization strategy, strip out the < 3 second clicks, cross-reference IP clusters via the data vendors, etc. Don't let a few dozen automated clicks pretend to be thousands of real ones. Verify, then scrub. And then make the decision.

The most common type of campaign that produces unusual A/B test results is one that has been designed or created for campaigns.

One of the best examples of this can be seen when an edited version of a refined e-mail message was used to compare a raw draft.

More people responded to the raw, unedited version of the e-mail.

In today's world of e-mails, consumers will immediately disregard anything that looks like it has been made overly professional since almost all of their emails are sent via AI.

Therefore, here at MKB Media Solutions, I always advocate for using personal stories as opposed to the safe copy created by computers.

We need to connect with people in order to communicate.

If you don't include some human grit (real power) in your writing then you're simply going to be just another voice of the millions of voices on the Internet.

A revealing A/B test involved the first onboarding email after sign up. Version A opened with credentials, authority, and process. Version B started with what the next seven days would feel like, including uncertainty, waiting, and when to expect clarity. The second version delivered a 24 percent higher click rate and reduced early drop off. That result surprised many people internally because the stronger proof points seemed commercially safer.

We took from that test that people do not always need more information first. Often they need emotional orientation. If the message reduces ambiguity, trust builds faster than when the message tries to impress.

One surprising A/B result we saw was a plain-text style email outperforming a polished visual template on reply rate and downstream conversion. The visual version looked stronger, but it created too much promotional distance. The simpler version felt more direct and trustworthy.

The key learning was that audience intent should drive format. For decision-stage audiences, concise copy with one clear action and one proof point performed better than multiple CTAs and design-heavy modules. We also learned to evaluate results beyond open rate. Click-to-action quality and conversion lag gave a more accurate picture of campaign value.

Now we test one variable at a time and define success metrics before launch. That discipline has made our experiments more repeatable and easier to scale.

We ran an A/B test on a client's abandoned cart sequence that completely changed how I think about email copy. The control was our standard approach -- subject line mentioning the product they'd left behind, body copy with a product image and a "complete your purchase" CTA. Solid, sensible, what everyone does.

The variant stripped out the product entirely. No image, no product name in the subject line. Instead, the subject was "quick question about your order" and the body just asked: "Was something wrong with the checkout process? If you hit a snag, I'd genuinely like to know so we can fix it." Signed from the founder's name, plain text, no design.

The variant outperformed the control by 41% on click-through and 27% on actual completed purchases. We were stunned. It looked like a customer service email, not a marketing one. And that was exactly why it worked.

The lesson: people don't abandon carts because they forgot. They abandon because something stopped them -- price concern, shipping cost surprise, trust wobble. Reminding them what they left behind doesn't address the friction. Asking what went wrong does. It reframes the email from "we want your money" to "we want to help." And it turns out that's a much more effective way to get the money anyway.

We've since rolled this approach out across three other e-commerce clients. The plain-text, service-first variant beats the polished marketing version every time. Not close -- usually by 20-35%.

A/B Test: Curiosity-Driven Subject Line Boosts Open Rates by 30%

One of the most surprising A/B tests I ran recently focused on something as simple as the subject line—and it led to a 30% increase in open rates.

The Test:

We tested two subject lines for a new product launch email for our e-commerce client:

Version A: "New Product Launch: Check Out Our Latest Collection!"

Version B: "What's New This Week? (You Don't Want to Miss This!)"

Both versions aimed to promote the same product collection but took different approaches. Version A was direct and explicitly highlighted the product launch. Version B took a more casual and curiosity-driven approach, leaving the reader with a bit of mystery.

The Results:

Version B outperformed Version A by a 30% higher open rate. We also saw a noticeable improvement in click-through rates (CTR) and overall engagement. This unexpected result indicated that a more casual, curiosity-based subject line resonated better with our audience than a straightforward promotional one.

Key Takeaways:

Curiosity Drives Engagement: The A/B test showed that curiosity-based subject lines often perform better than direct, "sales-pitch" lines. Posing a question or implying something exclusive or interesting can increase open rates.

Tone Matters: The tone of the subject line plays a crucial role. While we initially thought that being direct about the product launch would grab attention, the more casual, open-ended version worked better for this audience.

Regular A/B Testing Is Key: Tools like Mailchimp or Klaviyo make it easy to test subject lines and optimize email performance. Regular testing helps you understand audience preferences and refine your strategies.

This experiment taught me that subject lines should not just reflect the content of the email but also engage readers right from the inbox. Curiosity and simplicity can make all the difference in boosting open rates and overall email engagement.