799 reads

Enhancing Programmatic SEO with ChatGPT: The Good and the Bad

by Nebojsa (Cookie Duster)November 22nd, 2023

Too Long; Didn't Read

Programmatic SEO is often defined as a process or a strategy that aims to help you publish unique, optimized, high-quality pages at scale. But in reality, as most SEOs will confirm, it is not always like that.

featured image - Enhancing Programmatic SEO with ChatGPT: The Good and the Bad

Programmatic SEO did not come on the wings of AI (although it can benefit from it, as you are about to see). It is a practice that ruled affiliate marketing gigs for some time, adopted by different legit industries strategically targeting long-tail keywords to jumpstart the topical authority of the site, and it’s best described by example.

Understanding Programmatic SEO with Example

Say you run a travel blog and live from selling ad space on your website. The calculation for you is simple: more traffic = more $. There are so many places one can visit, but you’ve visited only a handful. Still, you want to have pages for all those places you haven’t visited yet (or at least those popular places) because they can earn you $.

Surely, you can add a couple of pages by hand. But if you want to add more than 100 places as fast as possible, what then? Then, you opt for programmatic SEO to leverage automation to handle the scale of optimization tasks while ensuring each place/page is search-engine friendly and user-centric.

The above means you find keywords to scale in and around TRAVEL destinations. Build a template/s full of modifiers, parameters, and variations that aim to diversify text. Have data in a sheet or a script that will build the pages with a click of a button.

Come back in 3-4 months, check analytics, and see for which TRAVEL term you rank well and build a hub around it (to enforce even more authority) or update the page manually with proper content. Sit back and let ad $ roll.

Easy right? Once, maybe. Today, the game is slightly different thanks to E-E-A-T (Expertise, Experience, Authoritativeness, Trustworthiness).

Any website that publishes a large number of similar pages risks being labeled as a thin content provider, i.e., a website with content that provides little to no value to the end user. Once in that box, getting out of it is an arduous task.

So, what can you do? Well, that’s easy: produce quality content for those programmatic pages.

OK, but how? One way is with ChatGPT; we will explore that, but first, I need to add a few things.

Jumpstart the Topical Authority

Topical authority is real. In practice, that means you’ll be able to rank faster and better for a topic G dreams you're an authority on.

Examples.

During my 24-year career in web and SEO, I worked as an affiliate guy only to move to work with a big publishing company and now (mostly) SaaS business. While with the publishing company, we had two interesting websites, one in the gaming industry and one in movies.

The gaming website ranked for PS and Xbox keywords with ease. But we bumped into a block when we started covering iPhone/Android games. Poor results. No amount of backlinks could help us move mobile game pages to rank better because we were, first and foremost, a go-to place for PS and Xbox news/topics.

The movie website ranked incredibly well for MOVIE TITLE trailer, review MOVIE NAME, coming soon MOVIE NAME, etc. keywords, but nothing happened when we started covering the series. The Big G and Bing recognized us as an authority on movies.

With a previous web dev company, we started to make waves in Jamstack (web dev architecture) communities, and with the recognition came good ranking on all Jamstack-related keywords (I even made a case study out of ranking for a term and explaining how I did it). Ranking for other web dev terms was harder or more challenging than for Jamstack terms.

I’m trying to say here that, like it or not, your website is (or will be) associated with a specific topic/industry/keywords for which you’ll rank much easier (than others). After all, you won’t see Disney rank well for adult entertainment keywords, right?

So, to move or associate your website with another industry/segment/topic, the topic needs to be close to the one you are known for, and it’s best you aim at low-traffic keywords and/or long-tail keywords.

The appeal of the programmatic SEO approach in building topical authority is in its ability to deliver a lot of pages quickly, helping you make a footing (so to speak) in the topic quickly.

The problem here is content or the lack of it.

Role of ChatGPT in Programmatic SEO

A lot has already been said about ChatGPT and its usefulness. Like any other tool, it is only as good as you are. Two things I mean by that:

How good you are at explaining what you need, and
How knowledge of a topic you already have.

The quality of your prompting will ensure the quality of the ChatGPT output, and your knowledge guarantees you’ll easily spot mistakes and hallucinations in outputs when you proofread it.

And you have to proofread it because Google and Bing will do it as well, and if the content is not up to par or better than what is already in search results, you won’t rank.

Having said that, let’s now focus on where and how ChatGTP can help.

Automating Keyword Research with ChatGPT

Keyword research allows you to tap into what's already happening in your industry (and keywords) to get a clear picture of what and how your target audience is searching for and interest (volume) for those keywords.

Usually, you’d use tools like SEMRush or Ahrefs and search results to evaluate the above and build your keyword sheet. With ChatGPT, you can do it a whole lot faster.

If you are lazy, use one of the pre-made prompts as the ones AIMPR offers. For our TRAVEL example, let’s do it with a single keyword: Spain (yes, I know it’s too broad).

And the output and the results are:

But it pays off not to be lazy and do a bit more prompting. You can introduce a persona or role for your output, even share your blog as an example of writing, suggest modifiers, parameters, and variations (as you would in the classic programmatic SEO approach), etc. Check the output and iterate any part of the output till you are happy, and then request a sharable Excell file.

Is that all? Well, not really. In the early days, you would make a generic text copy with SPCAES for keywords and modifiers (keyword being a destination, i.e., Spain or Italy, modifiers being words like beach, restaurant, culture, etc.). What you want to do today is do the same for another TRAVEL destination (ex., Italy) and make one giant sheet with all keywords.

How Can ChatGPT Generate SEO-optimized Content?

Content is where the actual fun is :-) And the pain. Let’s say we opt out of creating a 600-word article for each title in the Excel sheet. It would be pretty extensive if you try to run it as a single prompt.

Instead, you’ll have to run each TITLE independently and copy/paste the results into the Excel sheet. But it allows you to personalize the writing even more by assigning personas and roles and pointing out what you already have in terms of SEO.

Again, the lazy way to do it would be using one of the pre-defined prompts or do something like this:

Remember, the more details and explanations you can offer, the better and more unique the output will be.

Proofread

There are two ways to proofread the output. Take a couple of sentences from the output and check it in Google for duplication. Take names and places mentioned to see if these are real and explained adequately (if you don't know them already).

Tools like QuillBot and Grammarly can help you as well. Or, why not use one of the available trained GPT Agents like Copy Editor?

Adding results in two new columns, one for the article and one for the image (besides proofreading), is the most time-consuming manual part of the work.

Publishing

Hitting that publish button depends on the tech behind your website. Say you are running a WordPress website. You can implement the programmatic SEO pages with a plugin or create custom WP template files.

Running with Wix? You can do it with Python and Velo, for example.

Is a modern web development tech stack in place? There are a couple of excellent solutions for those into headless architecture and modern frontends. The team at BCMS shared their programmatic case study here as an example. Commerce Layer explained how they generated an e-commerce glossary with AI here.

At Crystallize, we’ve built our knowledgebase section aptly named Answers because the topics rather than keywords came from answering sales inquiries and client glossary questions. The primary purpose of the knowledge base is to deliver quick answers to both sales and clients and potentially rank for WHAT IS type of queries. You can tell we’re just starting.

While Crystallize is a headless commerce platform built for product storytellers with PIM, eCommerce, and CMS functionalities out-of-the-box, adding pages via Crystallize is pretty straightforward.

Finally, you can use something likeModern Next.js Assistant GPT Agent to help you build pages with Next.js, for example (or any other framework).

BTW, this is also an excellent way to learn modern web development.

Challenges and Drawbacks

Poor execution, whether research, content, or publishing, can seriously hurt website growth. As I mentioned, the risk of creating thin content - pages with little substantial information or value to the user is genuine. Search engines may view such content as an attempt to manipulate rankings rather than provide useful information and penalize you for it.

Even with ChatGPT, automating content creation increases the risk of producing duplicate content across multiple pages and even domains. The mass generation of content can sometimes border on spammy tactics, which might harm your website/brand's image and credibility.

Then, there are nuances of language and cultural context, potential biases in AI-generated content, and the need for human proofreading and oversight to maintain brand voice and authenticity.

Finally, there's an ethical dimension to consider. According to the latest estimates, 328.77 million terabytes of data are created daily. Do you really want to help overpopulate the web with shitty content?

The Two Cents No-one Asked For

While programmatic SEO presents a unique opportunity to scale SEO efforts and capture long-tail traffic, it requires a careful approach and meticulous planning. Balancing automation with quality content creation, ensuring adaptability to search engine changes, and maintaining the integrity of your brand are critical for the success of any programmatic SEO strategy. ChatGPT can help that, but be careful.