My (dumpster fire) pseo casestudy

programmatic seo traffic

Often, I’ll test something just to see if I can do it.

Hottest chili contests, hang gliding…

Bulk creation of landing pages targeting long-tail keywords with the intention of drastically increasing awareness.

(Quick definition- Programmatic SEO is a way to create loads of keyword-optimised pages en masse.)

image 15
A rising tide lifts all the…keywords

It all began with a glint in my eye and an expired domain in my cart. A domain that when I visited WayBackMachine offered no red flags (just quality snooping).

It had a Moz DA of about 10, a PA in the high 20’s and a good Citation and Trust Flow (aka it wasn’t spam-city). I had a game plan.

Failed project case study

1)Buy an expired domain- reason being, an expired domain (with appropriate niche history and clean backlink profile) would expedite the indexing process.

image 12
Domain graveyard

2) Source some data- I decided on affiliate posts/comparison posts as this was my area of comfortability working with “mommy bloggers” in the past. I went about data sourcing/scraping in a few ways.

3) Upload using ??- This is where the wheels fall off. And this is where we’ll start.

Freak in the sheets

In the first iteration of data collection, I equipped myself with SmartSuite (and Google Sheets). I manually created a product comparison database. It was my Frankenstein’s monster, only less green and more SaaS.

smartsuite for programatic seo
Smartsuite is the 🐐 btw

I decided I wanted to go more niche. As an antidote to digital overload, I’d create an “auto-reviewer” of ProductHunt’s new releases.

This would be excellent- if I knew how to scrape with anything other than a Chrome extension.

I tested out various no-code scraper tools and pre-made “recipes” (Bardeen, BrowseAI, BrowserFlow, OctoParse, ScraperBee) and ended up settling on Apify.

The rented Actor on Apify didn’t give me all the datapoints I wanted (not enough to programmatically template to be valuable to readers).

failed project case study
producthunt v1 dataset

After 2 days of violently Googling the difference between puppeteer and playwright- I used GPT4 to help me create my own Apify actor to scrape ProductHunt for data points.

produchunt data

I tested the copy creation and sourcing of additional data points using a beta feature from Clay (Claygent) as well as their webscraping function + free Clearbit domain logo finder.

image 13
Messing with Claygent to grab data points

I asked for things like “integrations” and “number of plans” & “features”.

The sentiment of G2 scraping, logo finding en masse and bulk data was best to do with Clay. Linking scrapers together, I decided Apify was the way to go.

We didn’t stop to think if we should

While I was all gung-ho about my data collection, feeling like Mr Robot,I was missing basic pSEO commandments and considerations.

Turns out, creating a logical site structure once you’ve bulk uploaded 800 posts sans tags/categories is more complex than convincing a cat to take a bath.

Then there was the design layout. If I had a dollar for every time someone didn’t call my layout “the visual embodiment of Shakespeare’s work,” I’d be… well, broke. (Please give me a dollar).

I bulk imported using WP All Import, and function.php scripts to associate feature images to images already in the media library (helped along by my friend chatgpt).

image 8
my best friend, doing heavy lifting for another pSEO site I’m running

I’d technically done it. I had built a WordPress site that could run on autopilot-

  • use bulk sentiment analysis from reviews to see how people felt about the product
  • crawl/scrape the website attached to the PH URL and find additional pricing info/emails.
  • Internal linking on autopilot

What I’d missed was — user experience, functionality on the site, and any kind of design/templating.

My baby was ugly. And people weren’t converting.

I harp on about un-gating knowledge. To grow, we need to share, unabashedly. It’s important to share failed experiments and iterations. Growth doesn’t happen in a vacuum.

image 10
Site domain purchased, indexed 10/5/23

I’m not thrilled with the design and technical set up of the site. Nor the results 30 days in.

I’m still experimenting with the actual content and what I think is most valuable.

At the time of posting, RankMath is saying it is ranking for 143 keywords. Keep in mind, these are longtail, in most cases “zero search volume”.

There’s a lot to remember and a lot for me to gain from my first pSEO site.

Things I’d be focusing on-

SCALABLE DATA COLLECTION-starting with intent (top-of-funnel, shitty AI written content is everywhere), what dataset can you collect to create conversion, and hold reader interest.

DESIGN MATTERS- My first full pSEO site taught me a number of things about myself. One is that I rush. I do not possess that “attention to detail” we all lied about on our resumes. Go back to intent. What information do you want to display, what’s important? How are you going to showcase that?

The goal of programmatic SEO isn’t just to flood your site with traffic. It’s about providing memorable experiences for targeted users.

The only real mistake is the one we don’t learn from.

What about you? Have you had any programmatic SEO mishaps or triumphs? Connect with me here