Improve your ability to handle large amounts of data and perform data analysis at scale

Pedro says: “Learn how to deal with large amounts of data and do data analysis at scale. That involves developing insights into MySQL, querying databases, or even dealing with spreadsheets. You can also use tools like AI to help you.”

Does every SEO need to do this or is it just for certain roles?

“As you move and your business grows, you will have more data to analyse and scale suddenly becomes important. Of course, this is more aimed at SEOs that work in-house because not many agency-side SEOs have this kind of technical approach.

It's important for SEOs that deal with big businesses and businesses that are either growing exponentially or have the potential to grow exponentially. You should be prepared to gather data and analyse data at scale because that will uncover trends, the behaviour of crawlers, internal linking, and anything else that you can pull into a database.”

Why do you think this advice is so important right now?

“I started doing this four or five years ago. For some of my clients back in Brazil, we would go in-house and work as a consultant on their premises, working with their teams and helping them establish these connections.

Now, with AI becoming mainstream, it has opened a lot of doors to make better use of the data you can analyse by querying databases and pulling things into a spreadsheet. Using AI that is able to interpret code, such as ChatGPT, you can easily feed it a spreadsheet and ask it to slice the data the way you want or find trends that you would not see just by looking at it.

This opens the door for a lot of potential in terms of what's coming in the future because it's not only giant websites that can do this, medium-sized companies will be able to do it now as well.”

How would you summarise best practice and/or bad practice when using automation in the production of content?

“One of my business partners in previous ventures is an automation engineer, and he always told me that the tricky part of automation is controlling the output; it’s not the input itself.

Bad practice with AI and automation is letting it run loose without any kind of curated control over it. Let the AI identify what you want to analyse but be the ultimate guardian of what goes out, what gets published, and what gets seen. Use it for the surface-level things that you would otherwise miss, but don't use it to replace you or use it without supervision.”

Do content creators have to use AI and automation, if they don’t want to get left behind?

“Yes and no. There will always be a need for a good content creator. AI can give you ideas, it can take what's already been done and rehash it, but it lacks human inspiration and creativity. It lacks the literacy that humans have. AI can only build on what's already been created; it cannot come up with something completely new.

Sometimes you need that in the creativity process, to bring something really good and really different to the table. Otherwise, it's just going to be marginally different, and sometimes that's not enough.

You can use AI to create content at scale – and when I say, ‘content at scale’ that’s content based on data. You can create content that is based on APIs and data, and you can use AI to create that at scale. However, it gets tricky when AI has to be creative. That's where AIs usually hallucinate but humans excel. We can be creative, and yet make sense of the creativity and make it understandable for other humans. The machine doesn't understand its own limitations in that regard.”

What shouldn't be automated?

“Anything where the output is unpredictable. If you cannot predict the quality of the output, then you probably shouldn't automate it. It's a matter of testing. Test the output for a while to see whether it always comes out clean. If there's a chance that it starts to present things that don't make sense, then don’t automate it.

The tricky aspect of using a fully automated system based on AI is that, when it starts to output things at a certain scale, you might be blind to what's coming out wrong.

There will be a percentage of things that come out wrong that are covered up by things that come out right. In the end, when they reach a certain level, they will hurt your business, your brand, and the way people perceive you. People will see that you didn't put effort into making these things look right. That is why you need to be the curator of what comes out of a machine rather than just letting it run loose.”

Is there anything that we need to do differently to optimize for AI-driven search results?

“I don't think so. We are using AI for very specific things.

Two days ago, I was looking for a formula in Google Sheets that I couldn’t remember. I was looking for a tutorial on YouTube on how to use this formula and I couldn’t find one. YouTube presented me with videos that were using this formula to do other things, but not what I wanted to do. In the end, I went to ChatGPT and input what I wanted to do for my specific case, and I got the answer that I needed.

That is a very specific thing; there are maybe 100 people in the UK who want to do this. It's not something that an SEO or content publisher would go after because it doesn't have the volume to justify creating a piece of content around it.

For these cases, where the usage is very niche or specific, an automated AI can help you. It offers things that it has seen somewhere else or that it has learned by looking at things and understanding what you are after. I don't think you have to use AI to create this kind of content, but that is the need the AI is aiming to fulfil.”

Is there any software that you use to help with automation?

“I use a few plugins that pull ChatGPT into Google Spreadsheets to help with formulas. You can say, ‘I want to do this or that in this spreadsheet’, and it does it for you. Other than that, I don't use many tools outside of the interface itself.

In my case, if I'm running an SQL query and I can’t fix the error, I ask ChatGPT what's wrong with it. It will tell me if I have used a curly brace or curly quotation marks, when SQL only allows straight quotation marks. Those kinds of errors can be almost invisible to the human eye and ChatGPT really helps to spot those. I don’t use any fancy software.”

Does the development of AI mean that we need to move past algorithm chasing and attempting to cheat the Google system?

“I think we should have done that a long time ago, even before AI. Ever since Google moved into doing core updates, they have been putting so much into them and it all becomes part of the main algorithm. It's almost impossible to chase an update and tell what has changed.

We all know that Google has algorithms aimed at sites that abuse keyword stuffing or use thin affiliation – sites that live off affiliation but put very little effort into producing added-value content. If these kinds of sites see ranking changes after core updates, it wasn’t necessarily the core update that was responsible. These algorithms have been working for a decade or more, and they have been working well. Google can improve them, but they don't really change. It should not come as a shock for a thin affiliate website, or a website that has other problems, when the algorithm is not favouring them.

What you need to understand is that the systems and algorithms that Google has in place, and always has had in place, don't stop working and are not replaced by core updates. They work in parallel. If your thin affiliate site was hit now that a core update has been released, that might just be a coincidence, and the website has already been hit by the algorithm countless times in the past. Correlation is not necessarily causation.

The industry is getting better overall, but many people who don't have good information tend to be derailed by chasing after the algorithm, which is usually a waste of time. If you’re running a thin affiliate site, you should align your expectations with how much you're going to get out of it and for how long.”

If an SEO is struggling for time, what should they stop doing right now so they can spend more time doing what you suggest in 2024?

“Most of all, stop chasing Google algorithms. That's certainly a waste of time. Besides that, we are often focused on what Google is changing, but we forget to work for what we think the future is. Be where the puck is going to be; don't chase after it.

For example, you might need to improve the structure of your website based on the vertical it is operating in because the vertical, the products within that vertical, and how people search those products, dictate how the site should be organised. We organise information differently according to the context, and the same applies to websites. Not all websites should be structured the same way, which is certainly not news to anyone.

We should be more vocal about teaching other people and mentoring newer SEOs in areas that are the base of the industry, like information architecture, user experience, and web accessibility. Sometimes, SEOs don’t understand why they are doing what they are doing. Why are you adding alt text? Why is it a web accessibility requirement? Why do search engines see value in alt text?

Once we understand the foundations of the different areas of SEO, our mindset should work according to those foundations, not according to our own whims or according to Google. That’s what you should do, and that's how you put yourself where the puck is going to be because that's what Google wants to aim for.”

Pedro Dias is Founder at Visively, and you can find him over at Visively.com.