f you have ever stared at a blank screen trying to write the perfect ad headline, you are not alone. Even a bank with a marketing department the size of a small city runs into the same wall. JPMorgan Chase, one of the largest financial institutions in the world, decided to stop guessing and start running proper AI ad copy testing instead. It partnered with AI marketing platform Persado to find out whether artificial intelligence could write better performing ad copy than its own experienced marketing team. The results say less about AI being smarter than humans and more about something every small business can copy: the power of structured testing.
Why Ad Copy Testing Matters for Every Business
Writing ad copy is part art and part guesswork. A marketing team can be experienced and still not know, in advance, which headline, tone, or call to action will get more clicks from a specific audience. Chase’s challenge was not unique. It is the same challenge every business faces when writing a Google or Meta ad. You only find out what actually works after you have already spent the budget. That uncertainty is exactly why AI ad copy testing has become such a useful tool for marketers who want answers backed by data instead of opinio
How Chase Used AI Ad Copy Testing Against Its Own Team
Chase entered a multi year partnership with Persado, an AI platform that generates and tests marketing language using data on how different word choices, tones, and phrasing perform with real audiences. Instead of relying purely on instinct or brand style guides, Chase used Persado’s AI to generate multiple copy variations for its campaigns and ran them head to head against copy written by its in house marketing team. The goal was simple: measure actual performance rather than internal opinion about which version sounded better.
The Reported Results
According to Persado’s own partnership announcement, some of the AI generated ad copy produced during this testing achieved click through rate lifts of up to 450 percent compared to the human written versions in that pilot. This is a reported figure from the technology provider and Chase, describing results from a specific pilot campaign. It is not a guaranteed or average outcome across all of Chase’s advertising.
Results were reported by the company or technology provider and may not represent outcomes achievable by every business.
What the Results Actually Mean
It is tempting to read “450 percent lift” and assume AI is simply better at writing ads than people. That is not really the lesson here. The real story is that Chase ran a controlled comparison instead of assuming its existing copy was already working well. The AI tool’s advantage was less about creativity and more about scale. It could generate and test many more copy variations, much faster, than a human team writing by hand. The lift came from testing, not from AI being inherently more persuasive.
This distinction matters for anyone considering AI ad copy testing for their own campaigns. A business that turns on an AI copywriting tool but never tests the output against alternatives is not really doing what Chase did. The value sits in the comparison, not in the tool alone.
There is also a human factor worth noting. Persado’s approach does not remove marketers from the process. It changes their role. Instead of writing every line from scratch, the team’s job becomes choosing which AI generated options to test, setting the brand tone the AI should work within, and interpreting the results. That is a useful model for any business weighing up AI copy tools. Use them to widen the pool of options, but keep a human making the final call on what actually gets published.
How Small Businesses Can Apply AI Ad Copy Testing
You do not need a Fortune 500 marketing budget to put AI ad copy testing to work. Here is how a small business can apply the same principle Chase did.
1.Never run a single version of an ad. Write at least two to three headline and description variations for every Google or Meta ad campaign, even if it takes a bit more time upfront2.Test one variable at a time. Change the headline but keep the offer the same, or the other way around, so you know exactly what caused a change in performance.
3.Let data decide, not opinion. The version your team likes best is not always the version that gets more calls or form submissions. Let click through rate and conversion data settle the argument.
4.Use AI tools to generate options, not final answers. AI copy tools, including features already built into Google Ads, are useful for producing variations quickly. Treat every AI generated line as a draft to be tested, not a finished ad.
5.Track quality, not just clicks. A high click through rate means nothing if those clicks do not turn into leads. Always connect copy testing back to actual enquiries or sales.
6.Give it time before judging results. A test needs enough clicks and conversions before the results are reliable. Ending a test after two days of data usually leads to the wrong conclusion.
7.Keep a record of what you have tested. Even a simple spreadsheet noting which headline won, and why, stops your team from testing the same idea twice or losing useful insights over time.
Where This Approach Has Limits
Chase’s result is a single reported pilot outcome from a large financial brand with an enormous amount of traffic and testing budget behind it. A 450 percent lift is not a realistic benchmark for every business, every industry, or every audience. Financial services campaigns, audience size, and testing volume all shape a result like this. Small businesses with lower ad spend and smaller audiences will see smaller, slower, but still meaningful improvements from structured AI ad copy testing. The lesson worth taking is the method itself, testing copy deliberately, not the specific percentage.
Ready to Test Your Own Ad Copy?
If your Google Ads campaigns are running on a single set of headlines that have not changed in months, there is a good chance you are leaving performance on the table. Wesetads can set up structured AI ad copy testing across your campaigns, track which versions actually drive calls and form submissions, and refine your ads based on real data rather than guesswork. Get in touch with Wesetads to talk about testing your ad copy properly.
Want more breakdowns like this?
We regularly cover real world AI marketing case studies and share practical AI ad copy testing tips you can apply to your own Google Ads and Meta campaigns. Follow Wesetads on YouTube for weekly insights, or explore more case studies on our blog to see what actually moves the needle for small businesses in Melbourne’s west.
Frequently Asked Questions
Does AI generated ad copy always outperform human written copy?
No. Reported results like Chase’s come from specific, controlled pilot tests. Performance depends on audience, industry, offer, and how rigorously the copy is tested, not on AI alone.
How can a small business start AI ad copy testing without expensive tools?
Start simple. Write two or three variations of your headline and description, run them at the same time within Google Ads or Meta Ads, and compare click through and conversion rates after a few weeks of data.
Sources
Persado, “JPMorgan Chase Announces Five Year Deal with Persado for AI Powered Marketing Capabilities”
