My 30 Day AI Writing Challenge: A Public SEO Experiment

The following is a public experiment to revive a stagnant website (CopyAdsContent.com) using AI-assisted blog content.

In September 2025, I published an article stating I was done with freelance copywriting. By October, I was getting an influx of new leads:

• 2 were spam offers
• 5 were real opportunities
• 2 converted into $15K worth of work

From November 2025 onward, my website stopped generating leads. I also saw my organic visitors dropping and referrals from AI drying up.

Was it because I stopped posting content? Or because the last post I published told the world I was no longer available for copywriting?

I don't know the answer, but what I do know is that I want my copywriting website to be a lead gen machine again.

To achieve this, I'm starting a 30-day writing challenge with a twist.

I intend to publish one AI-drafted, human-edited blog post per day for 30 days. Every post discloses how it was made and links back to this page.

At day 90, which is 30 days of publishing plus 60 days of search maturation, I'll publish the final results.

Start date: May 4, 2026
End date: June 2, 2026
Final results: August 1, 2026

Keep reading to see what I'm testing, why this site is the test environment, and what I think (well, hope) will happen.

Baseline Metrics

A few notes on how to read the table above.

The baseline is a fixed snapshot taken on May 4, 2026, before I started publishing AI-assisted blog content.

Every number in it represents my site's performance over the 28 days (Google Search Console) to 30 days (Fathom Analytics) prior. This is the standard reporting window in GSC and Fathom which will provide me with a clean comparison against any 28/30-day window I pull later.

Day-90 numbers will be added to the same table on August 1, 2026, using the same window, the same sources, and the same methodology.

About the data

• Google Search Console (GSC) for anything that happens in Google Search — clicks, impressions, queries, page-level performance.

• AHREFS (free version, connected to my GSC data) for the deeper view: who's linking to me, what my DR is (a vanity metric, but worth tracking), and the longer historical window GSC doesn't give you.
• Fathom Analytics for on-site visitor data instead of Google Analytics (I use it because it doesn't use cookie tracking, which means I don't need a bloated popup/cookie banner. The sites I build for clients also run on Fathom, you can use my link for $10 off your first invoice).

My top performing blog post

Back in 2024, when ChatGPT was the "cool new AI for writing content" I ran my first AI experiment by creating a 10,000+ word article:

• 55 Copywriting Formulas for More Clicks, Conversions & Customers

Any copywriter worth their salt should be testing and experimenting with AI, especially if they can cut costs (without cutting corners) by producing words that lead to measurable results (i.e., more clicks, leads, and revenue).

I'll also be tracking this post to see whether or not existing blog content will improve over the next 90 days. 

One thing to mention about this insanely long post is that it was never meant to drive leads to my business.

I targeted the keyword: Copywriting Formulas.

My ideal client isn't searching for ways to write their own copy. 'Copywriting formulas' is a term that new and beginner copywriters look up to learn the trade.

However, this kind of article positions me as a copywriter who knows his stuff (at least, that's what I tell myself before I go to bed each night).

I have used most of the formulas in that post across various projects since I became a copywriter in 2016. Such as the PAS Formula, it's forever one of my favorite copywriting techniques and still proven to convert when used correctly.

This raises the obvious question:

If my best-performing post is one that targets a keyword my ideal clients would never search, where are my actual leads coming from?

From "steady stream" of leads to 6 months of crickets chirping 

The start of 2025 was pretty brutal.

I lost a couple of stable, retainer clients as they opted for either:

1. Local talent/agencies (versus a freelance copywriter in Chiang Mai, Thailand who was living their dreams).
2. Or cheap, AI-content.

If I'm being honest, I started feeling obsolete which had me questioning my self-worth and abilities in a career that was my bread and butter for nearly a decade.

I moved back to Canada in the summer to live in my parents' basement to reassess my future. I made the decision to QUIT copywriting and shift all my time and attention into BAOB.ca, my web design agency (which was mostly just a side-gig until then).

In September 2025, I wrote an article titled: Goodbye CopyAdsContent.

After that published, the weirdest thing happened:

I got an influx of new, inbound leads for my copywriting services from AI and Google search.

There were a few other inquiries, but it was your typical spam offer from web design agencies pitching "white labeling."

I also got an inquiry from a fake "@adobe.com" email address. The Adobe team (doubtful it was real) was hoping to submit a guest post in exchange for 1-year free access to Adobe Express.

Uhm, no thanks.

Anyway, I haven't had a single lead come through this website since November 2025.

This is my theory as to why my leads went dry

I don't have concrete evidence but I believe my leads went dry because I was actively signalling that my copywriting business was closed.

We all know AI is getting sophisticated and a post saying I've shut down my copywriting business is exactly the kind of context an LLM or a search algorithm absorbs and acts on.

I've seen the way AI uses a company's website as a source of information. If you've never pre-prompted AI to compare your business with another, give it a try. You'll see it scrape and reference every detail across your website.

Google's also been shipping a ton of algorithm changes over the last year. If "this brand is closed/inactive" exists in the content, why would an algorithm recommend it to people?

But again, this is just a theory.

My KPIs for the AI content published over 30 days

In January 2026, I deleted the post that was (probably) shooting myself in the foot. 

Because, despite all the volatility in the creative marketplace, I am — and will forever be — a copywriter.

Despite shifting gears and offering web design, I don't identify as a web designer nor do I want to "design" websites. I know that sounds weird, but web design is meaningless if you don't have the write right message, SEO strategy, and structure in place.

I've been part of one too many projects that fail post-delivery because the web designer was too focused on creating some flashy and fancy site that clients loved and approved but the end results never attract leads, convert visitors, or generate revenue. 

My primary goal is to revive CopyAdsContent.com into a lead gen website for productized copywriting services.

In other words, I will also be updating the entire site but won't be publishing anything until AFTER this experiment ends.

My hope is that after 30-days of AI-assisted blog content, my site will:

1. Reverse the "this brand is closed" signal that I think tanked my inbound. I believe consistent publishing should tell search engines and LLMs that this site is active again.
2. Lift the existing copywriting-formulas post from position 18.9 into the single digits by reinforcing it with a cluster of related deep-dive articles.
3. Grow total impressions by 30% or more within 90 days, measured against the May 4, 2026 baseline.
4. Grow total clicks by 25% or more in the same window. Impressions without clicks is just noise.
5. Get at least one new post into Google's top 10 for its target keyword.
6. Increase referrer traffic from ChatGPT, Perplexity, Claude, and Gemini. This is the metric I'm most curious about — if AI search citations grow faster than Google rankings, that'll reshape how I think about content strategy entirely.
7. Earn at least one editorial backlink from another site that's not part of the spam SEO ecosystem currently polluting my AHREFS report.
8. Bring back inbound leads. Even one real inquiry within 90 days of the experiment ending would be evidence that the theory holds. New leads are the real KPI I care about, everything else is just a vanity metric I can use to brag about on LinkedIn.

If this site can't recover or generate meaningful traffic again through AI-assisted articles, I'll likely stop trying to revive it and stick to BAOB.ca or possibly go back to teaching English.

However, I refuse to give up being a copywriter, even if AI can generate slop faster than my 5 km race pace.

Why I'm running this experiment in parallel to client work

Beyond my hope to revitalize a career in creative services (which has been decimated by AI's ability to produce writing cheaply and at scale) I've already seen some early signs of success for a few of my clients.

To avoid the usual slop AI produced, this is my process for delivering AI-assisted content for a client:

1. Interview the founder, owner, CEO, or sales team.
2. Record transcripts and extract their actual language — the phrases they use, the way they describe their work, the way they talk about their customers.
3. Use that as the source material for AI-assisted articles. Hyper-specific to the business, infused with the client's voice, targeted at low-hanging keywords they can realistically rank for.
4. Build internal links from those articles to the money-making landing pages.

One of my clients started picking up do-follow backlinks from other businesses in their space — businesses that found the articles via AI and decided to reference them in their own (mostly AI-generated) content.

So, clearly marketers and decision makers everywhere are experimenting with AI articles (with many publishing slop and giving backlinks for free). I'm not complaining though, just observing.

The difference with my experiment on CopyAdsContent.com is that this is a controlled version of what I'm already seeing work in the field. But on a domain where I have full baseline data, no client politics, and complete control over what gets published and when.

On accountability

One last thing I want to mention is that writing is a lonely craft. I've attempted 30-day writing challenges before and gave up because:

1. No one was reading my writing; or,
2. I allowed other things to get in my way (like a new client project, I'd drop everything I was doing and focus 100% on that. I gotta pay my bills after all).

Luckily, this is also part of a 30-day writing challenge I pitched to WPCNX, the WordPress meetup group I organize in Chiang Mai. Within this community there's all kinds  of creatives using WordPress from website developers to travel bloggers to local businesses learning how to grow and succeed online.

So, this gives me the pressure to deliver alongside a group of fellow writers, developers, and small business owners doing the same.

Even if my experiment fails, at least I'll have failed in good company.

If it works, I'll have a documented playbook you and anyone else who still reads content online can either steal or apply to blogging.

Either way, thanks for taking the time to read about my 30 day AI writing challenge.

I'll update this on August 1, 2026, when the results are in.