What happens when you deliberately break all the testing rules in public. Including what it taught me about what big newsletter sponsors are actually getting for their money.

This week’s newsletter design experiment broke a lot of rules, here’s what happened and what I learned.

First, the recap.

This is Week 9 of my rebuild-in-public series. If you’re new here:

Week 1: Several people questioned my sanity. Briefly, so did I.I de-indexed 1,300 pages from my website.

Week 2: because the entire email industry changed its rules and forgot to tell anyone. Well, technically they told me. It went to spam.My open rate crashed to 11%

Week 3: Five unglamorous technical SEO tasks that sit on everyone’s to-do list until the damage becomes visible.

Week 4: A content rescue operation across 1,219 pages Google had visited and quietly rejected.

Week 5: I got the open rate to 70%. And argued that the list you already have is worth considerably more than the one you’re chasing.

Week 6: Sirens and a husband with broken ribs. Nothing went to plan. I sorted my inbox instead and wrote the guide on doing the same.

Week 7: Sponsored post enquiries started coming back. I wrote about the income stream that’s quietly saved my business more times than I care to count.

Week 8: I let Claude loose on my own website. Found things I really didn’t want to find. Published them anyway.

Right. Week 9.

Why I broke my own rule. On purpose.

The rule, which I have stated explicitly in this series, is this: test one variable at a time.

Week 9, I changed three things simultaneously.

I knew exactly what I was doing.

Here’s the honest version of why.

I’ve been planning to bring sponsors on board. And the big sponsored newsletters, the ones charging serious money for placements, are not plain text. They’re designed. Long. Category labels. Branded headers. Sponsor boxes woven between content blocks three times before you hit the unsubscribe link. That’s the format the market has decided a monetisable newsletter looks like.

Before I start selling that format, I needed to know how it actually performs with my readers. Not a hypothesis. Not a case study from someone else’s list. My data, from my list, on a real Sunday morning.

I also had a backlog problem. I’d published a lot of content over the previous two months and had never properly surfaced any of it to newsletter subscribers. A big roundup felt like a way to fix both problems in one send.

So yes, I changed the design, the volume, and the link format all at once. And I’m not rewriting history about that decision.

Here’s why.

Testing one variable at a time produces cleaner data. It also, when you’re doing it live in public, produces less interesting reading. The scale of this experiment was part of the experiment. I went big enough that readers had strong opinions and shared them immediately. I doubt I’d have generated a tenth of the replies I got if I’d changed one small thing and written a careful post about it.

And here’s the other thing, based on the feedback and the results together, the picture is actually surprisingly clear, despite my terrible methodology. The open rate held. The click rate collapsed. Multiple readers said the same things independently. I’ll get to all of that.

I have learned things from this that I couldn’t have learned by doing it correctly. Including some things about newsletter sponsorships that I think most big operators would rather you didn’t think too hard about.

We’ll get to those too.

Week 9 experiment overview: new design, 28 articles, 3 sponsor boxes and the results

What went into this newsletter design experiment

The design

Full HTML newsletter. First time I’d ever sent one.

What went into the Week 9 newsletter: 28 articles, 6 categories, sponsor boxes and reply mechanics

Orange header with my logo at the top. Six content categories, each with its own orange category label. Every single article linked from an orange CTA button. Three grey sponsor boxes with orange buttons woven between the categories. Three reply mechanic boxes throughout. A forward box. And a 600-pixel gap before the unsubscribe link. (That last one is deliberate. Unsubscribing should require a small amount of scrolling effort. Not impossible. Just not frictionless.)

The logo nearly ended it before it reached anyone.

My first approach was embedding it as base64 image data directly in the HTML code. Kit’s editor froze entirely. The file was 576KB, which turns out to be quite a lot for a newsletter builder to process. The fix was a reference the logo via its WordPress URL instead of embedding it, which keeps the loaded file size under 25KB. This is not in any tutorial I found. It cost me twenty minutes and some fairly uncharitable thoughts about email editors.

If you’re building HTML newsletters in Kit and the source code editor keeps freezing when you add an image, that’s almost certainly why.

Newsletter design built for value, not vanity: every element serves a purpose

Twenty-eight articles across six categories

I’d published a lot of content over the previous two months. None of it had been properly surfaced to newsletter subscribers. This was that roundup.

28 articles across 6 categories in one newsletter design experiment send

Sales and leads (7 articles):

→ How to break the feast and famine cycle

→ How to use Reddit for lead generation

→ Why sales feel hard (and what actually fixes it)

→ How to close more sales with sales psychology

→ 25 unconventional lead generation ideas

→ 10 B2B lead generation tools that will fill your pipeline in 2026 

→ What a no-brainer business really is

AI (6 articles):

→ Keeping up with AI without losing your mind

→ Stop making ChatGPT sound like a corporate robot 

→ When to use ChatGPT, Claude, Gemini, Perplexity and Copilot 

→ How to build a one-person AI business in 2026

→ How to use ChatGPT agents to save 10 hours a week

→ Best free and paid AI courses in 2026

Weekly experiments (6 articles):

→ I built my career on luck and timing 

→ My open rate crashed to 11%

→ How I took my open rate from 11% to 70%

→ I’m de-indexing 1,300 pages to save my website

→ I let Claude loose on my website

→ How to make money from sponsored posts 

Tools and websites (3 articles):

→ 50 websites you didn’t know existed

→ 30 free business tools that are actually free 

→ Best content repurposing tools in 2026 

LinkedIn and social (3 articles):

→ How to beat the LinkedIn algorithm in 2026

→ How to use AI for LinkedIn content strategy in 2026

→ How to make your brand memorable

Productivity (3 articles):

→ 25 productivity habits for entrepreneurs that save real time

→ How to keep your business running when your brain has left the building 

→ How to organise your email inbox properly

That’s twenty-eight articles.

Twenty-eight orange buttons.

Every design element in the Week 9 newsletter had a specific purpose

The reply mechanic

Four options, all specifically about the design:

1. This is better. Keep it exactly like this.

2. Bring back plain text immediately. The buttons are doing something to me.

3. I noticed nothing. I read this on my phone at 6am and I am barely a person at that hour.

4. All of the above and I did not expect to have feelings about email design today.

How long it took

Four to five hours, across three separate working sessions because context windows kept dying mid-build. Every section of copy was scored against five editorial voices and rewritten multiple times before being locked. Then the HTML was built and uploaded via Kit’s source code editor.

Significant effort for the lowest click rate of the entire rebuild.

That’s the experiment.

Before I get to the results. Here’s what the previous week looked like.

For eight weeks, I’d been sending a plain text newsletter. Four articles per send. Two arrow-format text links per article. No buttons, no design, no HTML. Just words.

Open rates had been running consistently above 70%. Click rates had been sitting at 7 to 8%. Those are the baseline numbers. Keep them in mind for the next section.

The results

The open rate didn’t move.

The click rate dropped 80%.

Week 9 newsletter design experiment results: 70% open rate held, 1.69% click rate dropped 80%

Twenty-eight buttons. Fifty-six clicks.

That’s the number. Say it again slowly. I put twenty-eight separate orange buttons in a single email. And I got fewer clicks than four plain text links had been producing for eight consecutive weeks.

The unsubscribe rate was 0.48%. Normal. No alarm there.

So people opened it. The subject line worked. The sender reputation held. And then they got inside the email and… mostly didn’t click anything.

This is not the catastrophe it looks like at first glance. Here’s why.

The replies were something else entirely.

This newsletter got more replies than anything else I’ve sent in the entire rebuild series.

Not a little more. I’d estimate ten times what a typical newsletter would produce. People had proper opinions about the design, and the reply mechanic gave them somewhere specific to direct those opinions.

Nobody replied with option 2 asking to go back to plain text. Multiple people chose option 1 asking me to keep the design. There were strong feelings about the volume, the buttons, reading it at 6am with half a brain. I replied to every single one personally. If someone takes the time to write back, they get a real response. That’s not negotiable for me.

Then came the Facebook comment.

I’d shared a post promoting the newsletter. A few positive comments came in, which I’d expected. Then one subscriber commented publicly, great content, but too much of it, I don’t read it because I prefer bite-sized.

I hadn’t anticipated a negative one being public.

My assumption was that any negative feedback would arrive via email reply, privately, where I could respond without an audience. The idea of someone commenting critically on a public post promoting the newsletter hadn’t really crossed my mind. It probably should have.

I handled it the way I’ve had to learn to handle things like this over twenty years of doing this publicly, calmly, warmly, acknowledging the point, thanking her for the honesty. No defensiveness. No explaining myself.

I’ll be straight with you, t requires more composure than it sounds like on paper. Twenty years in this industry doesn’t make you immune to criticism under your own post. It just means you’ve built up enough scar tissue to know that the worst thing you can do is react. You respond. You don’t react. You do it with as much grace as you can manage, even if internally you’re doing something slightly less graceful.

Two decades of public work teaches you that responding calmly and warmly to public criticism is not weakness. It’s the only strategy that doesn’t make things worse. And privately, you assess whether the feedback is useful signal or just noise.

In this case? Useful signal. She wasn’t wrong.

Want to follow how this plays out in week 10?

Results land in the newsletter first, before the blog post. Join 15,000 subscribers.

What I actually learned. And no, I’m not rewriting history.

Here’s where I could write that I learned the golden rule of always testing one variable at a time.

I already knew that rule. I wrote it in this series. I then chose not to follow it.

What I actually learned is more useful.

The paradox of choice is not a theory. It’s your click rate.

You can know the research and still walk straight into it. I’ve cited the paradox of choice in client work. I’ve explained it to founders. Then I put twenty-eight choices in a single email and watched it play out on my own dashboard.

Four plain text links: 7 to 8% click rate. Twenty-eight orange buttons: 1.69%.

The paradox of choice in email: more links produce fewer clicks

If your newsletter click rate is lower than you’d like, before you test subject lines or send times or content: count how many things you’re asking your readers to do in a single send. The answer is probably higher than one. It should probably be one.

Open rate and click rate are measuring completely different things.

This keeps coming up and Week 9 is now the most dramatic illustration of it I’ve ever produced in twenty years of newsletter work.

Open rate measures whether your subject line and sender reputation are strong enough to get someone to open. That’s the entire job of the open rate. Once someone is inside the email, open rate has nothing further to say about what happens.

Click rate measures everything inside. The clarity. The focus. The number of choices. Whether what’s in front of the reader is specific enough to make them move.

You can be excellent at one and terrible at the other simultaneously. A 70% open rate tells you people trust you enough to open. It tells you precisely nothing about whether what’s inside is working.

Open rate versus click rate: two different metrics measuring two completely different things

Volume and value are not the same thing.

More articles means more decisions. Twenty-eight decisions before reaching the end of an email is too many. Most readers won’t reach the end.

Each link requires a micro-assessment: is this relevant to me right now? Do I want to stop reading and click through? Is it worth losing my place in the email? That happens fast, mostly without the reader noticing. But it happens every time. Multiply it by twenty-eight and you’ve asked readers to make twenty-eight small decisions in one sitting on a Sunday morning.

Most of them chose not to decide anything.

Replies are their own metric. Not a consolation prize.

It would be easy to frame the reply rate as a silver lining on a click rate failure.

That’s not the right frame.

Clicks and replies measure completely different forms of engagement. Clicks measure willingness to leave the email for external content. Replies measure willingness to engage directly with the sender. Both are valuable. Neither replaces the other.

The replies from this newsletter told me things that no click data could, which design elements readers noticed first, whether the buttons felt editorial or promotional, how the volume made them feel. That’s qualitative data an A/B test doesn’t produce.

Build a reply mechanic into your newsletter. Give readers specific options to respond with. Make at least one of them slightly absurd. The responses will tell you more than your dashboard will.

The design wasn’t the problem.

Nobody asked for plain text back. Multiple readers explicitly said keep the design.

Visually, readers preferred the new format. The visual change was liked. The volume and the buttons were where the feedback consistently pointed.

Which means week 10 has a testable hypothesis: keep the design, reduce to four articles maximum, remove the buttons and go back to text links. If the click rate recovers above 5%, the problem was volume and buttons. If it doesn’t recover, the design itself has to go.

I’ll find out. And I’ll write it up in exactly this format.

What this tells me about the newsletter sponsorship market.

This was, in some ways, the real experiment from the start.

Most established sponsored newsletters are designed, long, and full of link positions. That’s the format. The click-through rates from those newsletters are almost never shared publicly. Not in media kits. Not in case studies. You get open rates. You get subscriber counts. You rarely get click rates.

My hypothesis, based on this experiment: their actual click-through rates are considerably lower than their open rates suggest.

Here’s the maths.

A sponsor in my plain text newsletter, during the eight weeks before this experiment, was competing with three other links for reader attention in an email getting 7 to 8% click rates. A sponsor in a designed roundup with twenty-eight link positions is competing with twenty-seven other things. If the total click rate of that email is 1.69%, the sponsor’s share of that pool is a fraction of a fraction.

The designed format looks more impressive in a media kit pitch. The plain text may deliver better outcomes for sponsors.

Newsletter sponsor strategy: 3 sponsor boxes placed naturally across the newsletter

And here’s what I’m sitting with now.

My open rate is 70%. I’m proud of that number. Getting a newsletter to 70% open rate is not a small thing. Most newsletters, including the big ones with hundreds of thousands of subscribers, don’t publish that number because it would make their own look bad by comparison.

The click rates I’ve been achieving in plain text are significantly higher than what this experiment produced with full design and maximum volume. Which means the value I can offer a sponsor is higher than what most big newsletters are actually delivering, regardless of what their follower counts look like in a pitch deck.

I haven’t started formally selling sponsorship packages yet. This experiment has clarified exactly what I’ll be selling when I do, and why the numbers will hold up under scrutiny in a way that most newsletter media kits simply don’t.

What’s happening in week 10.

The design stays.

The volume drops to four articles maximum.

The buttons go. Text links come back, the same arrow format I’ve been using throughout the rebuild.

If the click rate recovers above 5%, the problem was volume and buttons. The design format is safe and the path to a sponsored newsletter is clearer.

If the click rate stays low, the design itself has to go. Back to plain text. Which would be a shame, because the visual response was positive. But pretty doesn’t pay.

The goal is always conversions. Every other metric is secondary to that.

I’ll document whatever happens next week in exactly this format.

Keep testing.

Keep writing it down.

The useful findings are almost always inside the results that didn’t go to plan.

If you want to know more about partnering with or advertising in this newsletter, the details are here. I mention this every week. You’d be surprised how many people click it out of curiosity. A click is a click.

If what you’ve just read is the kind of thinking you want applied to your marketing, I work with founders and CMOs as a fractional CMO and AI automation specialist. Practical, specific, and built around real data rather than best practices.  Get in touch here.