Can ChatGPT Write a Better Safelist Ad Than I Can?

Jerry faces a ChatGPT robot in a boxing ring beneath the words “Who Writes It Better?” in a safelist advertising showdown.

I’ve been writing safelist advertisements for more than 20 years.

That’s a lot of emails.

Some were pretty good. Some weren’t. And after digging through a bunch of my old advertising swipe files recently, I’m not entirely sure what I was thinking when I wrote a few of them.

There were song parodies, questionable celebrity impressions, and an alarming number of references to the handsome administrator of Mister Safelist.

Which is odd, because I’ve always been remarkably humble.

Anyways…

These days, ChatGPT can write an advertisement in a matter of seconds.

So I started wondering.

Could ChatGPT write a better safelist ad than I can? And could it write one that actually sounds like me?

Some of My Old Advertising

I’ve always had a habit of saving advertising copy.

Subject lines, emails, splash page ideas, promotional messages. You name it, I’ve probably got it stuffed into a text file somewhere.

Looking through some of those old files was quite entertaining.

Here’s how I opened one of my Mister Safelist advertisements:

Ho hum!

Another day another ad for Mister Safelist.

It’s Ok.

My members appreciate the hard work I do every day to bring in new members for them to advertise to.

They’re cool like that.

No big promises. No amazing secrets.

Just me, advertising my website again.

In another advertisement, I encouraged people to give Mister Safelist a try with this reassuring message:

Try it, track your results, and if you don’t like it simply delete your account.

But I hope you won’t :)

I’m not sure encouraging potential customers to delete their accounts is considered good marketing advice.

But there it is.

Of course, I wrote plenty of normal advertisements, too. Sometimes I just wanted to explain what I was offering and get people to click.

Other times, I apparently had a little too much time on my hands.

Whether my sillier advertisements performed any better than the serious ones, I couldn’t tell you.

But I enjoyed writing them.

Okay, ChatGPT. Your Turn.

I uploaded seven of my old swipe files to ChatGPT and asked it to look through them.

I wanted to see whether it could figure out how I write just by reading my old advertisements.

It came back with a surprisingly detailed analysis.

Apparently, I like short sentences, casual conversation, self-deprecating humor, and occasionally making fun of whatever I’m trying to sell.

Fair enough.

Then ChatGPT took a shot at writing a new advertisement for Mister Safelist.

Here’s what it came up with.

Subject: The Marketing Department Has Spoken…

Hello!

I asked the marketing department if we should send another ad for Mister Safelist.

They said yes.

The marketing department is me, so it was a pretty short meeting.

Anyways…

Mister Safelist lets you earn credits and use them to send your own ads to other marketers.

It’s free to join.

And we’ve got a very handsome administrator.

At least that’s what the marketing department tells me.

Click the credit link below and come check it out!

Enjoy!

Jerry Iannucci

Well, I’ll be darned.

I actually like it!

The marketing department joke is something I could easily see myself writing.

And naturally, ChatGPT recognized the importance of having a handsome administrator.

It’s learning.

So, Is It Better Than Mine?

I don’t know.

And I’m not going to pretend I do.

We haven’t tested these advertisements, so I can’t tell you which one would get more clicks or generate more sign-ups.

I’ve spent plenty of time thinking about what actually makes someone click, and I’ve been doing this long enough to know that the advertisement I like best isn’t necessarily the one that performs best.

Sometimes the simplest, most boring email produces the best results.

Which is unfortunate, considering how much time I’ve spent entertaining myself.

But I have to admit, ChatGPT did a pretty good job.

It picked up on my writing habits, recognized the sort of things I tend to say, and produced something that sounded surprisingly familiar.

And it did all that without spending 20 years writing safelist emails.

Show-off.

But Is Sounding Like Me Enough?

Here’s what I find interesting.

When I wrote those old advertisements, I wasn’t trying to develop some carefully crafted personal brand.

I was just writing.

Sometimes I had a feature to explain. Sometimes I wanted to try something different. And sometimes I apparently thought rewriting a popular song about Mister Safelist was a productive use of my afternoon.

Looking through those emails now, I can see the habits. Casual introductions, ridiculous subject lines, and the occasional joke at my own expense.

I suppose that’s how people came to recognize my ads.

And now ChatGPT can learn those habits in a matter of seconds.

But if AI can write something that sounds like me, does it really matter whether I wrote it?

I’m not sure it does.

As long as I read it, agree with it, and would have been comfortable saying it myself, I don’t see the problem.

It’s still my name at the bottom of the email. And it’s still my responsibility.

If ChatGPT starts promising people they’ll earn a million dollars by joining Mister Safelist, I doubt it’s going to stick around to answer the support tickets.

So, What Have I Learned?

I started out wondering whether ChatGPT could write a better safelist advertisement than I could.

I still don’t know the answer.

But I discovered that it can do a remarkably good job of sounding like me.

And that’s both impressive and a little strange.

I spent years writing those advertisements, developing habits, and occasionally making questionable creative decisions.

ChatGPT figured out how to imitate much of that in a few moments.

I’m not sure whether to be impressed or insulted.

Probably a little of both.

But I already use ChatGPT in my daily marketing work, and I can certainly see myself using it to come up with new advertising ideas.

And if it occasionally writes something better than I could have written myself, that’s fine with me.

I’m remarkably humble, remember?

What do you think?

Have you tried getting ChatGPT to write advertisements in your own style? Did the results actually sound like you, or did they sound like something everybody else is sending?

I’d love to hear your thoughts.

Enjoy!

Jerry Iannucci

Are We All Just Advertising to Each Other?

Marketers promote safelists and traffic exchanges while others look bored or earn advertising credits.

Hello!

I’ve been involved with safelists and traffic exchanges for more than 20 years, and every once in a while I stop and think about what we’re actually doing.

We all want people to look at our advertisements.

So we click other people’s ads to earn credits, which we use to show our ads to people who are clicking to earn credits so they can show their ads to us.

Round and round we go.

It’s like walking into a room full of salespeople where everybody is trying to hand out business cards, but nobody particularly wants to take one.

Everybody’s selling.

But who’s buying?

Seriously. Are we actually advertising to potential customers, or are we just advertising to each other?

And if that’s what we’re doing, why does any of this work?

Let’s Admit Why Most of Us Are Here

I don’t think there’s much mystery about why people use safelists and traffic exchanges.

They’re trying to promote something.

Nobody wakes up in the morning thinking, “Boy, I can’t wait to look at 100 advertisements for random affiliate programs today!”

Well, maybe somebody does. I’ve met some pretty enthusiastic surfers over the years.

But I’m willing to bet most people wouldn’t spend nearly as much time looking at advertisements if there weren’t credits involved.

I certainly wouldn’t. I’ve spent plenty of years earning credits myself.

And there’s nothing wrong with that. It’s how these systems are designed to work.

But let’s not kid ourselves. Getting somebody to look at your advertisement because they want credits isn’t the same as getting somebody to look because they’re interested.

That’s a pretty important difference.

So are we all just wasting our time?

I don’t think so.

We’re Not Exactly Strangers

One thing that makes safelists and traffic exchanges interesting is that a lot of us know each other.

Maybe not personally. I’ve never met most of the people I see advertising every day.

But I recognize their names.

I recognize their websites.

I know which programs they’ve been promoting.

And when somebody I’ve seen around for years suddenly starts advertising something new, I notice.

That’s especially true when a new program launches.

You start seeing it everywhere. Familiar advertisers begin promoting it. People get curious. Some join. Others sit back and watch to see what happens.

There’s an entire little community operating underneath all those credits.

And I think that matters.

We’re not necessarily looking at thousands of completely anonymous advertisements. We’re repeatedly encountering people and businesses we’ve seen before.

Familiarity doesn’t automatically create trust. Somebody can become well-known for all the wrong reasons.

But recognition gives an advertisement something that a completely unfamiliar advertiser doesn’t have.

A history. A reputation. A reason for somebody to think, “Hey, what’s this person promoting now?”

That’s not something a traffic counter can easily measure.

Somebody Must Be Paying Attention

I’ve run enough advertising campaigns over the years to know that most visits don’t turn into anything particularly exciting.

But some do.

People sign up. They join mailing lists. They discover new programs. Sometimes they even spend money.

Last July, I ran my Safelist Attention Project through 45 safelists to see whether people were actually paying attention.

Instead of promoting a product, I sent visitors to a page where they could voluntarily click a button confirming they’d noticed it.

After cleaning up the data, I ended up with 8,392 visits and 143 deliberate button clicks.

That’s a 1.70% interaction rate.

Not exactly a number that makes you want to run out and buy a yacht.

But here’s the thing.

Nobody earned any extra credits for clicking that button. They’d already done what they needed to do to get their credits.

Yet 143 times, somebody took that extra step.

Does that mean the other 98.3% weren’t paying attention? Of course not. Some may have read the page and moved on.

But it does tell me that at least some people were doing more than blindly clicking their way through another batch of advertisements.

They were paying attention.

And that’s really all I’m getting at.

People can be clicking for credits and still notice something that interests them.

It happens.

Maybe We’re Advertising to Exactly the Right People

Here’s another possibility.

Maybe advertising to other advertisers isn’t nearly as ridiculous as it sounds.

Think about the things most of us promote.

Safelists. Traffic exchanges. Affiliate programs. Advertising tools. Autoresponders. Lead capture systems.

Who’s the ideal customer for something like that?

Probably somebody who already does online marketing.

And where do you find a concentrated audience of people actively involved in online marketing?

Well…

Safelists and traffic exchanges.

Imagine that.

Of course, this doesn’t mean every offer belongs here, or that every advertiser is going to succeed.

If you’re promoting the same tired opportunity that everybody has already seen 500 times, you’re probably going to have a hard time getting noticed.

But the same thing that makes this audience difficult to advertise to might also make it valuable.

These people understand online marketing. They need advertising. They use marketing tools. And many are always looking for the next interesting program.

Maybe the audience isn’t the problem.

Maybe we’re just not always giving them something worth noticing.

So, Are We All Just Advertising to Each Other?

Yes.

To a considerable extent, I think we probably are.

I’m not going to pretend otherwise. I own advertising sites myself, so I’m certainly part of the system.

Most of us are here to promote something, and the credits are what keep the whole thing moving.

But that doesn’t mean nothing worthwhile happens along the way.

People recognize names. They discover new programs. They follow what other marketers are doing.

They join things. Sometimes they buy things.

Most of the time, they’ll probably just collect their credits and move on. That’s the reality of this kind of advertising.

But every once in a while, something catches their attention.

Maybe the real problem isn’t that we’re advertising to each other.

Maybe it’s that too many of us forget there’s an actual person on the other side of the advertisement.

Someone with interests, problems, preferences, and money to spend.

They just happen to be earning credits while they’re doing it.

And that brings me to one final thought.

If you found this article through a safelist, there’s a pretty good chance you clicked the link because you wanted credits.

Yet here you are.

Reading an article about people clicking advertisements for credits.

There’s something wonderfully ridiculous about that.

And maybe it proves at least a small part of my point.

Yes, we’re advertising to each other. But “each other” isn’t nobody.

Thanks for reading,

Jerry

My Daily Mailer: Year 1

My Daily Mailer Year 1 featured image with a morning workspace, coffee, laptop, and sunrise view

My Daily Mailer is officially one year old.

That feels a little strange to write.

A year ago, I had an idea I thought was pretty cool. I knew how I wanted it to work. I had spent a lot of time thinking through the logic behind it.

But there is always one question you can’t answer until you put something in front of real people:

Will it actually work?

After one full year of My Daily Mailer, I finally have an answer.

I think it did.

And looking back through the numbers, I may actually like the idea behind it more now than I did when I launched it.

The Idea I Really Wanted to Test

I’ve been using safelists since 2003.

I’ve promoted on them, tracked them, written about them and operated them. I’ve watched the industry change quite a bit while the basic mechanics of a safelist stayed surprisingly similar.

One thing kept bothering me.

The economics of a traditional mailer don’t naturally balance themselves.

Members earn credits. Sites give away credits. Upgrades create credits. Promotions create credits. Balances accumulate.

Eventually, those credits turn into mail.

But there isn’t necessarily any connection between the amount of mailing power being created and the amount of activity available to support all that mail.

That was the problem I wanted to play with.

With My Daily Mailer, one Mail Token has a very specific value: one mailing.

But that’s not the clever part.

The part I still think is coolest is this:

The amount of MT members earn for reading changes based on how active the mailer is.

The economy responds to the mailer itself.

Member activity helps determine how much future mailing power gets created.

As conditions inside the site change, the earning rate can change with them.

Instead of constantly trying to force balance from the outside, the system is designed to move toward balance on its own.

It’s a little nerdy.

I know.

But I still love it.

The Members ARE the Economy

That’s also why My Daily Mailer depends so much on participation.

When somebody reads an email, they’re not simply clicking to collect an arbitrary pile of credits.

Their activity is part of the system.

Reading creates the conditions that allow mailing.

Mailing creates new messages to read.

Those reads create more mailing power.

And around it goes.

That’s the experiment I wanted to run.

Could a safelist economy regulate itself based on what the members were actually doing?

A year later, we’ve had enough real activity to give that idea a pretty serious test.

Year 1 By the Numbers

I pulled a full 365-day report directly from the My Daily Mailer database for this anniversary.

During Year 1:

  • More than 2,700 members joined My Daily Mailer
  • Members sent 28,305 mailings
  • Those mailings produced almost 15 million emails
  • Members earned 32,531 MTs
  • The mailer delivered more than 1.76 million reading hits
  • More than 100 members made upgrade payments

Those are real numbers from a real year of people actually using the site.

And I’m proud of them.

But there is another part of the report I like even more.

Month 12 Tells Me More Than Month 1

Launch numbers are fun.

They’re also dangerous.

You can launch something new and get a flood of curiosity.

People join because their friends are talking about it.

Affiliates promote it.

Everybody clicks around to see what’s new.

Then six months later…

Crickets.

So when I went through the Year 1 report, I wasn’t especially interested in proving that My Daily Mailer had a good launch.

I wanted to know what happened after everybody stopped caring that it was new.

September 2026 was month twelve.

During that month, members sent 2,621 mailings.

Those mailings generated more than 1.35 million emails.

The system delivered 174,812 reading hits.

And on an average day, about 144 members were actively reading mail.

That’s month twelve.

Even more interesting, the mailing volume throughout 2026 was remarkably consistent.

Not one giant launch spike followed by a slow fade.

People found a rhythm.

They read.

They earned MTs.

They mailed.

Then they came back and did it again.

Some Members REALLY Use My Daily Mailer

The Year 1 report also reminded me how much of the site is driven by a relatively small group of extremely active members.

There were 270 members who completed at least 500 reads during the first year.

Together, those members accounted for more than 1.13 million completed reads.

And 243 of those heavy readers also sent mailings.

I love that.

Because that’s the system doing exactly what it was designed to do.

The people doing the reading aren’t trapped on one side of the equation.

They’re earning the ability to become senders themselves.

Read.

Earn.

Mail.

Repeat.

The economy keeps circulating.

And to those members who have spent hundreds—or thousands—of reads helping keep that cycle moving:

Thank you.

My Daily Mailer would just be software without you.

Then I Accidentally Tested It Against Everybody Else

One of my favorite moments of My Daily Mailer’s first year didn’t actually happen inside My Daily Mailer.

Earlier this year, I ran a safelist experiment for this blog.

I sent the same simple page through 40 different safelists.

There was no sale.

No opt-in.

No prize for clicking.

Visitors were simply asked to look at the page and choose whether they wanted to interact with it.

I wasn’t testing My Daily Mailer.

I wasn’t trying to prove anything about my own site.

I was testing attention.

And when I looked at the results, My Daily Mailer stood out.

By a lot.

Its visitors interacted with the page at a dramatically higher rate than the other safelists in the test.

Now, there are several possible reasons for that.

Members know me.

Lucky Letters give people a reason to pay attention.

The member base is active.

And the structure of the mailer may encourage a different kind of reading behavior.

I can’t separate all of those things perfectly.

But seeing My Daily Mailer stand out in an experiment I hadn’t designed to make it look good was pretty satisfying.

Especially because one of the things I wanted from the beginning was a mailer where activity actually meant something.

Year 1 Wasn’t Just Running the Site

Of course, I didn’t launch My Daily Mailer and then spend twelve months admiring it.

I tinkered with it.

A lot.

Lucky Letters became part of the experience.

I built a better contest system.

We now have reader contests and weekly reader sprints.

I added clearer guidance for new members who hadn’t yet figured out how to earn their first MT.

I watched the activation reports.

I changed things.

Then I watched again.

Some ideas worked.

Some didn’t.

Some things that seemed obvious to me were apparently not obvious at all once a new member landed on the site.

That’s one of the biggest differences between designing something and operating it.

Real members find every assumption you made.

That’s a good thing.

They make the product better.

The Site Is Still Evolving

My Daily Mailer is not finished.

I don’t really expect it ever to be.

That doesn’t mean I’m constantly trying to reinvent it.

The basic idea has held up remarkably well.

The self-balancing economy is still the part I believe in most.

But there are always ways to make the experience clearer, smoother, more useful or more fun.

Sometimes that means a new feature.

Sometimes it means changing a paragraph.

Sometimes it means staring at a report and realizing new members aren’t doing what I thought they were doing.

And sometimes it means spending far too long adjusting something that nobody except me will ever notice.

That’s part of running a site.

If You’ve Never Tried My Daily Mailer

If you’ve never joined, Year 2 seems like a pretty good time.

You can join for free.

You don’t need to understand the economics before you start using it.

Just read some member emails.

Watch your MT balance.

Earn a mailing.

Then use it.

You’ll figure out the rhythm pretty quickly.

And underneath that simple process is a system that’s constantly responding to the activity of the people using it.

That’s the part I’m proudest of.

Join My Daily Mailer

To Everyone Who Made Year 1 Possible

I built My Daily Mailer.

But I didn’t create what it became during Year 1.

The members did that.

Every person who opened an email helped create the economy.

Every person who sent a mailing gave somebody else something to discover.

Every affiliate who referred a member helped it grow.

Every Prime member who decided the site was worth paying for gave me another vote of confidence.

Every contest participant, Lucky Letter hunter and regular reader added a little more life to the place.

That’s what I see when I look at the Year 1 report.

Not 15 million emails.

People.

People using something I built.

People coming back.

People making the system work.

A year ago, My Daily Mailer was an idea I wanted to test.

Today, it’s a working mailer with thousands of members and a full year of real activity behind it.

That’s pretty awesome.

Thank you for Year 1.

Now let’s see what happens in Year 2.

— Jerry

My Daily Mailer

Almost Half My Attention Clicks Came From Just 20 Visitors

Small group of repeat visitors generating a large share of advertising engagement

My experiment recorded 143 attention clicks.

But they didn’t come from 143 different people.

They came from 94 distinct tracked visitors.

And just 20 of those visitors generated 69 of the clicks.

Almost half.

That was the final surprise in my Safelist Attention Project.


Most People Clicked Once. A Small Group Kept Clicking.

Throughout this experiment, I used a persistent browser cookie to identify returning visitors.

That doesn’t guarantee every visitor ID represents a completely separate physical person. Someone using another browser or device could receive another ID.

But it gave me a reasonable way to distinguish between a tracked browser appearing once and one coming back later.

Out of 7,380 distinct tracked visitor IDs, only 94 ever clicked the attention button.

Of those 94:

Clicker TypeTracked ClickersAttention ClicksShare of Clicks
Clicked once747451.7%
Clicked on multiple visits206948.3%
Total94143100%

That’s the number that really changed how I looked at the experiment.

74 tracked visitors clicked once and generated 74 clicks.

20 repeat clickers generated 69 clicks.

Nearly the same number of clicks.

From dramatically different numbers of people.


That Small Group Was Really Small

Those 20 repeat clickers represented only 21.3% of the people who clicked at all.

Yet they generated 48.3% of every attention click I recorded.

Put against the entire experiment, the number gets even more surprising.

There were 7,380 distinct tracked visitor IDs.

Only 20 became repeat clickers.

That’s about 0.27% of all tracked visitors.

Now, I want to be careful with that number.

I’m not saying 0.27% of the audience created half the value of the campaign.

They didn’t.

This experiment measured one very specific thing: whether somebody deliberately clicked the button telling me they had noticed the page.

What the number shows is much narrower—and still interesting:

A very small group generated a surprisingly large share of the measured interaction.


Repeat Clicking Usually Meant More Than One Extra Click

At first, I wondered if “repeat clicker” mostly meant somebody saw the experiment twice and clicked both times.

That did happen.

But it wasn’t the whole story.

Of the 20 repeat clickers:

  • 9 clicked twice
  • 11 clicked three or more times
  • several clicked four, five, or six times
  • one tracked visitor clicked on eight separate visits

So more than half of the repeat-clicker group interacted at least three times.

And the result wasn’t being distorted by one extreme visitor.

The most active tracked visitor produced eight clicks.

The remaining repeat clicks were spread across a small group that kept interacting when they encountered the experiment again.

Why?

I don’t know.

Maybe they recognized the experiment.

Maybe they enjoyed participating.

Maybe the page became familiar.

Maybe they remembered the unusual image.

Maybe they simply thought, “There’s Jerry’s experiment again. I’ll click the button.”

The experiment never asked them why they clicked again, so I don’t want to invent an explanation.

The safest conclusion is simply this:

The same tracked visitors remained willing to deliberately interact on later visits.


Most Returning Visitors Still Never Clicked

There’s another number that keeps all of this in perspective.

In my previous post, How Many Times Did People See My Ad Before They Clicked?, I looked at when people first responded.

There were 288 tracked visitors who appeared more than once during the experiment.

But returning did not automatically mean engaging.

Of those 288 repeat visitors:

  • 247 never clicked
  • 21 clicked on exactly one visit
  • 20 clicked on multiple visits

So only 6.9% of all returning visitors became repeat clickers.

That’s important.

I don’t want to look at these numbers and say:

“Just keep showing people the same ad and they’ll eventually become highly engaged.”

The data doesn’t say that.

Most returners still never clicked.

What happened instead was that a relatively small subgroup behaved very differently from everybody else.


Ten Clicks Can Tell Two Completely Different Stories

This may be the most useful lesson I’ve taken from this final part of the experiment.

Imagine two campaigns.

Campaign A gets 10 clicks from 10 different people.

Campaign B gets 10 clicks from three people who keep returning and interacting.

If you’re only looking at the click counter, both campaigns say:

10 clicks.

But those numbers are telling very different stories.

Campaign A shows more breadth.

Campaign B shows more repeat engagement.

Which would I rather have?

That depends on what I’m trying to accomplish.

If I’m looking for new leads, I’d probably be more interested in reaching more different people.

If I’m trying to build recognition or familiarity, repeat interaction may also be useful.

And if I’m trying to make sales?

Then neither number tells me enough.

I’d want actual conversion data.

That’s something this experiment has kept reminding me:

The value of a metric depends on what you’re trying to learn from it.


Different Safelists Produced Different Patterns

This also helped explain something I had noticed earlier in the project.

Not every safelist produced the same kind of attention.

Mister Safelist generated 14 attention clicks from 14 different tracked visitors.

That’s broad participation.

SendCircle produced 32 clicks from 20 tracked clickers, including five repeat clickers who generated 17 clicks between them.

My Daily Mailer showed a similar mix: 30 clicks from 19 tracked clickers, with five repeat clickers generating 16 clicks.

Then there were sources such as Zodiac Mailer, where eight clicks came from two tracked visitors who both interacted repeatedly.

I don’t look at those patterns as better or worse.

They’re different.

One source may produce broader participation.

Another may produce stronger repeat engagement from a smaller responsive group.

The important thing is that the total click number by itself doesn’t show the difference.


Why I Want Two Numbers From Now On

Before this experiment, if somebody told me an ad received 100 clicks, my first instinct would probably be to compare that number with another source that received 80 or 120.

Now I want another number sitting beside it:

How many different tracked visitors produced those clicks?

Because:

100 clicks from 100 people tells me one story.

100 clicks from 50 people tells me another.

100 clicks from 20 people tells me something else entirely.

None of those outcomes is automatically good or bad.

But without knowing both total clicks and unique clickers, I don’t really know what kind of response I’m looking at.

That’s a lesson I can use far beyond safelists.

The same idea could matter with email, content marketing, recurring advertising, social media, or anywhere else people can interact with something more than once.


One More Interesting Connection To The Eyes

There was one other result I couldn’t help noticing.

Earlier in this series, I looked at what people said made them notice the experiment page.

The eyes in the image barely won among people’s first responses.

But among the 29 additional follow-up answers submitted by repeat respondents, 21 selected the eyes.

That doesn’t prove the eyes caused repeat engagement.

It doesn’t even prove they were the reason somebody remembered the page.

But it does make me wonder whether that visual became especially recognizable once people had seen the experiment before.

Maybe the creative wasn’t only noticeable.

Maybe, for some people, it became familiar.

That’s something I’d have to test separately.


What This Doesn’t Prove

There are a few important limits to these results.

The visitor IDs were cookie-based, so they represent tracked browsers rather than guaranteed individual human identities.

A repeat click does not equal a new lead, a sale, or stronger buying intent.

The experiment didn’t ask people why they clicked again.

Visitors weren’t randomly assigned a certain number of exposures.

And this wasn’t a normal sales page.

The page openly told people they were part of an experiment and asked them to click a button if they noticed it.

Someone who returned later may simply have recognized the page and remembered what I was asking them to do.

That’s still deliberate interaction.

But I wouldn’t assume a commercial offer would produce exactly the same behavior.


What This Whole Experiment Changed For Me

When I started this project back in July, I really thought I was going to find out which safelists were best at getting attention.

I did learn quite a bit about the safelists.

But that wasn’t the most interesting part.

I learned that traffic and attention aren’t the same thing.

I learned there can be more than one kind of winner, depending on what I measure.

I learned that people notice different things.

I learned that some people don’t respond the first time they see an ad.

And now I’ve learned that even after people do respond, the attention itself isn’t necessarily spread evenly.

Some of it was broad.

Some of it came from repeat engagement.

Some visitors never responded at all.

And a very small group generated a surprisingly large share of the visible activity.

I started with a traffic question.

I ended up with a measurement lesson.

The number on the counter is only the beginning of the story.

From now on, if I see that an ad got 100 clicks, I don’t think I’m going to be satisfied with knowing there were 100 clicks.

I’m going to want to know one more thing:

Who produced them?

That’s probably the biggest lesson I’m taking away from the Safelist Attention Project.

Don’t just count the clicks. Understand who is producing them.

And with that, I’m going to call this the final core article in this experiment.

I’ve gotten far more out of that strange little page than I expected when I started sending it through safelists in July.

I’m not finished experimenting, though.

So now I’d like to turn the next one over to you.

If I run another marketing attention experiment, what would you like me to test?

Different headlines?

Different images?

Short emails versus long emails?

Curiosity versus benefits?

A plain page versus a designed page?

Whether attention actually predicts conversions?

Leave me a comment and let me know.

Your idea might become my next experiment.

Thanks for following along,

Jerry

How Many Times Did People See My Ad Before They Clicked?

Repeated ad exposures showing the second view getting noticed while later exposures fade.

In my last post, I looked at what made people notice the page in my Safelist Attention Project.

This time, I wanted to answer a different question:

What happened when somebody saw the page more than once?

If they didn’t click the first time, was that opportunity gone?

Or could another exposure give the ad a second chance?

Most People Only Saw It Once

The final analysis included 8,392 recorded visits from 7,380 tracked visitor IDs.

Most of them appeared only once.

  • 7,092 appeared once
  • 288 appeared more than once

So only about 3.9% of the tracked visitors returned for another recorded exposure.

For this article, an exposure simply means another recorded page load under the same visitor ID.

That could have come from another mailing, a return visit, a delayed click, or even a refresh. It does not necessarily mean somebody received the exact same email twice.

When Did People First Click?

There were 94 distinct tracked visitors who clicked the attention button at least once.

Here’s when they clicked for the first time:

Bar chart showing 72 first clicks on exposure one, 13 on exposure two, then 3, 3, 2 and 1.

The biggest takeaway is obvious.

72 of the 94 people who eventually clicked did it on their first recorded exposure.

That’s 76.6%.

So most of the attention happened right away.

But that also means 22 people — 23.4% of all eventual clickers — did not respond the first time.

Nearly one-quarter of the people who eventually clicked needed another opportunity.

That seems worth paying attention to.

The Second Exposure Was Especially Interesting

On Exposure #1, all 7,380 tracked visitors were eligible to become first-time clickers.

72 clicked.

That works out to a first-click rate of about 0.98%.

Now look at Exposure #2.

Among people who returned without having clicked the first time, 269 were still eligible to become first-time clickers.

13 of them clicked.

That works out to 4.83%.

So the first-click rate among those returners was nearly five times the rate on the first exposure.

That sounds dramatic.

But there’s an important catch.

The Second Group Was Different

I’d love to say the second exposure made people five times more likely to click.

The numbers don’t quite let me say that.

Everybody was included in Exposure #1.

Exposure #2 only included people who came back.

Those people may have been more active safelist users, more curious, more likely to revisit pages, or exposed to my campaign differently.

They were a self-selected group.

I didn’t randomly divide people into a “see it once” group and a “see it twice” group.

So this does not prove that repetition caused the higher click rate.

What it does show is that among the people who naturally returned without responding the first time, the second encounter created quite a few new responses.

That alone is useful.

A First Non-Response Wasn’t Always the End

This may be the most practical lesson for me.

It’s easy to run an ad, see that somebody didn’t respond, and assume the message failed.

Sometimes that may be true.

But in this experiment, 22 eventual clickers passed through once without clicking and responded later.

Thirteen responded on Exposure #2.

Another nine responded after that.

I can’t say exactly why.

Maybe the page looked more familiar.

Maybe they were less distracted the next time.

Maybe curiosity had another chance to work.

I don’t know.

What I do know is that one missed opportunity did not always mean the ad had lost them forever.

But More Repetition Wasn’t Automatically Better

The extra response also faded pretty quickly.

After the 13 new clickers on Exposure #2, the numbers dropped to:

  • 3 on Exposure #3
  • 3 on Exposure #4
  • 2 on Exposure #5
  • 1 on Exposure #6
  • 0 after that

In this dataset, every tracked visitor who eventually clicked had done so by their sixth recorded exposure.

I would not turn that into some kind of “six exposure rule.”

The later groups were small, and this was one unusual advertising experiment.

To me, the pattern is simpler:

The first exposure did most of the work. The second created another useful opportunity. After that, new response became increasingly rare.

So How Long Should You Keep Running an Ad?

I don’t think this experiment gives us one perfect number.

What it does suggest is a reasonable middle ground.

If you have an ad that you believe is strong, don’t necessarily abandon it just because someone didn’t respond the first time.

Give it another chance.

Maybe even a few.

But don’t use repetition as an excuse to run the same ad forever either.

Eventually, you may be better off testing another headline, changing the creative, rotating the ad, or bringing it back again later.

For me, the takeaway is:

A first non-response is not always the end of the story.

Most people who noticed my experiment responded right away.

Some responded only after seeing it again.

And eventually, the supply of new attention dried up.

That feels a lot more useful than trying to force the results into some magic advertising rule.

There’s one more strange pattern hiding in the repeat-visitor data.

A small group didn’t just return and click for the first time.

They kept clicking on later visits even after they had already responded once.

That’s what I’m going to look at next.