TL;DR
There’s nothing worse than pulling up a recipe page and having to scroll past about a bunch of background about how their recipe has been handed down four generations. So if you can relate, you can learn all the details about this service on my 404 No More services page.
The Inspiration
The internet is littered with broken pages. It’s not just an SEO issue; it’s a usability issue. When I learned about natural language processing (NLP) in grad school, it got the gerbils running in their cage. I wondered if it could be leveraged to solve one of the most daunting tasks for web managers: fixing broken pages.
So I started experimenting with using it, along with some automation and a lot of Pandas, to suggest redirects for one client’s site. I’ve been working on this suggestion engine since March. It’s given Chutes and Ladders in that I’d think I was close to a solid algorithm but then would get weird results for edge cases. So I experimented with different Python libraries, customized my dictionaries to exclude low-value keywords, and started applying weighting mechanisms. In the end I was satisfied with the results.
My Process
Sometime in September I felt like I had an algorithm that was runway ready. But I empathize with marketing teams who need to actually analyze their broken pages, so I set out to visualize this crawl data. One of my favorite charts for data that maintains a hierarchy but has a lot of data points is the treemap. It’s highly effective in giving you a 100-foot view of your hierarchical data.
To illustrate, I picked a site I’d feel good about helping out. Toms was a great candidate for a demo because it has a really clean architecture and a good mix of mostly clean directories with a few that have a higher concentration of broken pages. +10 for having a cool mission.
Looking at this chart gives you a good lay of the land as to the areas of their site that needed the most attention.
I made it highly interactive, so you can click on any node to drill down to that direct or page, while filtering the table below it, that contained the suggested redirects.
I excitedly gave people a heads-up that I would be launching a new, AI-assisted service to address broken pages but then started tinkering again. 🙄 At this time I was working a full-time contract, so I was doing this work nights and weekends. I basically stayed on my school schedule.
If I had been a client, I would’ve fired myself because the scope creep was unparalleled. And then there were those days that adding some new feature would completely break something that was previously working just fine. All the while I learned a lot about using AI for coding (for another post).
You can view annotated screenshots of these dashboards below and watch a demo of both highly interactive dashboards on my 404 No More page.
White Label Option
If you’re a consultant or agency and want to white label the service, I will align the branding to match yours and work through your team.
Referral Program
If you facilitate an email introduction and that organization closes within 60 days of the referral, you will receive a $1,000 referral fee for the Expert tier and higher and $500 for the DIY tier. Since it could get dicey, we’ll need that paper trail to minimize risk of colliding referrals.
If you make an introduction to an agency, you will receive a referral for each client the agency closes on within 60 days of your referral.
Let’s Chat 📞
If you’re interested in learning more, you can request a proposal, ask questions, or request a meeting on my 404 No More page.
Live Q&A
I will be hosting a Live Q&A on YouTube Mon, Nov 25th @ 1pm EST. You can submit your question ahead of time using this Google Form or ask your question during the live. I look forward to chatting!
Leave a Reply