13% of my users make half the pins and pay half the bills

I run Made to Spark, a Pinterest tool: paste a link or connect your blog or shop, and it turns your posts and products into a month of scheduled pins. Solo founder, a bit over a year in, under a thousand signups.

It didn't start that way. The first version generated articles for Pinterest: listicles with images, built to be pinned. Today it's a pin generator only. Here's what pulled me away from the original idea, what one SQL query later confirmed, and the SEO lessons along the way.

1. Biggest group ≠ best customer

At signup I ask what people do. Affiliate marketers were the largest group, so my homepage tried to speak to everyone, and a good chunk of my roadmap this year went into Amazon integrations built for them.

Then I joined signup answers to usage and payments:

• Bloggers are about 13% of users
• They make close to half of all pins
• They pay for about half of my subscriptions

Affiliates make a small fraction of pins and almost never pay. I had spent months building for the people who leave.

Do this: group users by their signup answer, then compare each group's usage and revenue. One query, thirty minutes. Do it before the next homepage rewrite, not after.

2. I stopped selling the feature the product was built on

Listicles made sense on paper: bloggers run out of posts to pin, so write them new ones. In practice they were expensive to generate, hard to make good, and people kept coming back for the pins, not the articles. In July I stopped selling them: new signups don't get article generation, existing users keep it, and it's gone from every marketing page.

The data later backed this up. Most of my paying subscribers never used articles at all. A small group still generates some every month, so I'm glad I didn't delete it.

Do this: separate "features people use" from "features people pay for". Sell only the second list, even if the first list holds your founding idea. The rest can stay for the people who already have it; you don't have to break anything to sharpen the pitch.

3. My old tagline was outranking the new one

Three months of Search Console, brand searches excluded, showed two things I hadn't expected:

• My best non-brand page is a tiny Rich Pins validator I built in a day. It beats every feature landing page I've sweated over.
• "Paste a link, get a month of pins" was a tagline I used in June and later replaced with something more polished. People still search it word for word, and it has one of my best click-through rates. I had deleted it from the exact page they were landing on. It's back.

The lesson underneath: people don't search for your feature names, they search for the outcome in plain words. Search Console will tell you which of your phrasings people already use. That's copywriting research you don't have to pay for.

Do this: before renaming anything, check Search Console for the exact phrase. And build the boring one-purpose tool your audience actually googles.

4. The skills I use for SEO

I build with Claude Code, and there are open-source "skills" (packaged instructions it follows) for exactly this kind of work. The three I use most:

• free-tools: which small utility to build from what your audience already searches, and how to route tool users into the product.
• seo-audit: diagnosing pages with lots of impressions and few clicks. My font generator page is next on the list.
• ai-seo: making pages citeable by ChatGPT, Perplexity and Google's AI answers (llms.txt, answer-first structure).

Do this: export your Search Console queries and pages, then point an audit at every page with high impressions and a low CTR before writing anything new.