Health tools, guides and an app for women in Nigeria and the diaspora

What It Is

Wholesome Girlies helps women through the years between dating and raising a child. It covers five stages: relationships, fertility, pregnancy, postpartum and parenting. Each stage has its own tools, including an ovulation calculator, a due date calculator, a hospital bag checklist and a baby milestone tracker. It also has guides written for Nigerian women, with a source for every medical point. There are six paid programs, one for each stage and a bundle, and an app called Girlies that you can install on your phone straight from the site.

Why I Built It

Over the years, I've learned more about the health problems women face simply because they have reproductive systems. It can be hard to get care made for women in Nigeria, where there are too few doctors who specialize in pregnancy and women's health, and even basic healthcare can be difficult to get. I built Wholesome Girlies to help in the small way I can.

What It Does

Women can start with a tool, read a guide or buy a program. Each tool links to the guide that explains it, and each guide points to the program for that stage. Relationship tools and games such as Red Flag Radar and Is He Husband Material? end with a result card you can send to a friend. The site checks where you are and shows prices in naira across Africa and in dollars in the diaspora. It also remembers what you've bought and takes you straight to your own dashboard.

The Girlies app is the same site, installed on your phone. It works offline, keeps the tools you save, and lets you back up and restore your entries. It also walks you through installing it in whichever browser you use. When I push an update to GitHub, the site and the app update together, so there is no separate app store version to maintain.

What It Taught Me

This is where I learned to build for search engine optimization (SEO) from day one and measure it properly. Seven weeks after launch, search was sending almost nobody to the site, and the dashboards could not tell me why. I pulled the numbers directly from Google, using its application programming interfaces. Google had never processed the sitemap, and 34 of the 60 pages did not exist as far as Google knew. I then ran the site through the SEO, GEO and AEO Playbook, the 52-step audit I built for Marketing In Action. It covers search engines and AI assistants. GEO stands for generative engine optimization, and AEO stands for answer engine optimization. The site scored 30%, and the AI results were the worst part. I asked four AI assistants to recommend a site like Wholesome Girlies. None of them named it, and two named a competitor. When I asked what Wholesome Girlies was, three of the four described a different company.

The audit changed how I build. I researched 4,322 keywords and grouped them into 115 topics based on the pages Google returns for each search. I then read the top five results on 220 Nigerian search results pages. The difficulty scores called some topics easy even though WebMD, the National Health Service (NHS) and government health sites already owned them. Forty-three topics were closed to a new site, and Nigerian websites held about 1% of the results across the whole category. The opening was in Nigerian words. A question about tiger nut in pregnancy or goron tula for fertility was open, while the same question in generic English belonged to the big health sites. So the plan focuses on pages those sites do not cover, like omugwo, Nigerian foods in pregnancy and the immunisation schedule in Nigeria.

I then built the system that publishes against that plan on a fixed schedule. Every Monday, a scheduled job pulls Google Search Console data and updates the strategy workbook. It rebuilds the content roadmap from what is live on the site and scores every guide and tool against a written checklist. I choose the week's pages from that roadmap. A script builds each guide from a data file and adds the trust signals the audit found missing: a named author, cited sources, and schema markup that tells Google and AI assistants what the page contains. On Friday, a release script checks every page, blocks anything that fails, and adds each page to the sitemap on the day it goes live. A second check runs on GitHub before every deployment and blocks any page that breaks the site's rules. Pages Google already knows get up to ten updates a week. New pages go out one or two a week, each submitted to Google by hand, because a new site that publishes in bursts can be treated as a content farm. The schedule repeats the AI assistant questions every month in the same words, and the full audit every quarter.

The traffic has not moved yet. Google had indexed 31 of 73 pages at the last check, 19 of the 110 planned pages are live, and the first full re-audit is on 24 December. I measure every change against the same starting numbers in an SEO tracker in Obsidian.

It is also where I learned to make content with AI from start to finish. I built a Copy Engine, a ChatGPT project that writes ad ideas and full ads, and a Design Engine, a Claude project that turns the copy into a plan for every slide. A Claude Code skill then turns that plan into files ready to post. For video, I created a cast of 10 cartoon characters for the tool videos and 22 realistic personas for the story videos. Each of the 36 tools has its own short video in three sizes for different platforms, and the videos play on every program's sales page and thank-you page. Getting AI video good enough for an ad took a lot of trial and error with my own money. The video system has its own page: AI Content Engine.

I also learned from a pricing bug. The first version sometimes showed a buyer outside Africa the price in naira, because it trusted one location check. I rebuilt it so two separate checks must agree. I found the problem by sending live traffic to the site and watching what broke.