David says: “The key thing is to stop trying to track Gen AI results the same as you used to track traditional keywords.

I see a lot of people talking on LinkedIn, and they want to take 100,000 keywords, which they’ve been tracking in traditional rank-tracking tools, and then drop them into their new Gen AI trackers.”

Shall we start at the beginning by asking you to define Gen AI?

“We have the AIO/GEO stuff. There’s a joke around “AEIOU/Old MacDonald Had a Farm”, but I think it's very much anything that is more advanced than traditional search results. We could even look back at things like rich snippets and featured snippets. They were some of the early Gen AI results.

What we're seeing now is maybe a bit more of an aggressive takeover in terms of the Google AI Overview, and then they have the deep AI Mode, but we are seeing a lot of people starting to test various other platforms – whether it's ChatGPT for answering some of those questions – and then comparing them against Google.

So yeah, I would say the Gen AI results are just the more advanced summaries.”

You say that people are making a mistake by trying to track this in the same way that they do more traditional SEO.

What kind of mistakes are people making, and why is this a mistake?

“As I said, looking at what people are sharing on LinkedIn around this topic, it's very much that people are taking what they used to track and dropping it into some of these Gen AI trackers. Typically, a lot of these Gen AI trackers are more expensive, they can't track as much, and they can't track as frequently.

An example: if you're a used car dealership, you would take your ‘used car’ keyword and drop it into the rank tracking tools and see, ‘How are we ranking for the Gen AI results around ‘used car’?’ People need to fundamentally review what types of keywords they really need to track and maybe be a bit more selective.

Instead of trying to track the entire keyword universe, they should just track a handful of more specific keywords related to their business and areas of concern.”

How do you know the keywords that you need to track in the new world of AI?

“I mean, that's the tough thing. Bing has obviously rolled out its AI Insights within Bing Webmaster Tools, but one thing I do see (it feels very much like tracking voice search) is that there are a lot of unknowns, and people tracking things for the sake of tracking them.

There are a few tools that try to estimate the amount of searches and stuff, but I would say, typically, by focusing on the core topics that used to drive a lot of traffic, in terms of traditional search, there's a fairly good chance that those are similar terms within Gen AI.”

Okay, so not too different necessarily for keywords, but what about the content itself?

How do you change the way that you produce content in order to optimize for Gen AI?

“I think there are a lot of people trying to take shortcuts. They've started doing listicles and stuff, and there are some programmatic, or ‘PSEO’, platforms out there that have tried to spam Google with tens of thousands of pages.

The one thing is, there are some penalties that have been going out over the last couple of weeks/the last couple of months. A lot of people were taking shortcuts and just producing low-quality garbage content to fill the gaps. It's coming back to bite them a bit.”

Okay, so what's the difference then between low-quality content and high-quality content?

Is high-quality content original content, or can you use AI to collate content to assist with the production of content?

“I think about the old school buzzwords, like the ‘10X’. Take existing content that's maybe outdated, it's not written with the most authority, and make amazing content.

It's not guaranteed that you're going to rank automatically, but if you are creating content, make sure it is good so that, if humans do actually find it, it does actually add value.

People say you're a Google lackey by saying some of these things, but fundamentally, if you're producing good content, it should be good enough that you would share it with your friends or family. If the content's not the quality you would actually publicly put your name to, or share with your friends and family, maybe you shouldn't be investing resources in creating that content.

It's about getting subject matter experts, and it's about doubling down. If I were to write content around trips to Antarctica, maybe I'm not the best person to do this. But if I were to write content about travelling around the Netherlands or Europe, I’m in a better place to do that because I live in Europe. It’s generally trying not to go too far outside your sphere of influence.

I would still say there's a huge opportunity for hiring journalists and subject matter experts and creating good content because, again, you can take that and you can also share it on social media. You can transfer that content into video content, which is doing really, really well.

That's the core thing that annoys me the most. The downside of Gen AI (the tools like ChatGPT and stuff) is that it allows content to be created that no writer would ever want to write. I could write ‘the best spatula for serving lasagna,’ but no content writer is ever going to say, ‘Yep, that's a great article. Let's go ahead with this.’

That's the downside to a lot of these Gen AI tools: it allows really, really subpar topics, that no one cares about, to be written about.”

You say video content is doing really, really well.

What kind of video content, on what platform? And how do you know if you should be publishing video content?

“If you look across some of the channels for video content, YouTube is still probably the second biggest search engine in the world. You've also got TikTok, and you've got Instagram. There's a lot of demand from consumers seeking video content.

The type of content that makes sense is the type of content where you can be a storyteller. For example, ‘How to pick a rug’ is maybe not the best video content. However, ‘How to correctly lay a tile floor’ (because I'm currently doing some renovation), that is amazing video content because, when you read an article, you can't really get your head around it, versus if you actually see someone doing an activity, you can really visualise what steps they go through.

Maybe Gen AI can help with some of these things, by making quick, easy visualisations to see which content makes sense, but I would say that for anything that's slightly difficult or slightly nuanced, video content works quite well.”

How much content is enough? You see brands that are publishing maybe one blog article a month or something like that, and there's so much content that can be published nowadays (partly because of AI). You alluded to the fact that it's not necessarily predictable whether or not a piece of content is to be a success.

Is it just a case of publishing as much as possible as often as possible? When do you know enough is enough when you're publishing?

“You should be able to track some type of feedback metrics from search engines. What percentage of the content that you're publishing is getting indexed or discovered, and what percentage of your content is getting traffic?

If you get to that tipping point, where the majority of your new content is no longer being indexed or discovered by search engines, or is not getting traffic, that's probably where it is.

One of the things I've seen in the past is that, sometimes, it doesn't come down to how much content has been produced; it's the effort that was taken. For example, in the past, I worked looking across around 800 websites, and for some of the websites, the content was great. They had five or six really good-looking guides, in terms of the metrics. They had a lot of time on site, they led to conversions, and all these types of things. Other sites had hundreds of pieces of content that no one ever visited.

There is some aspect of using a bit of time and effort to look at your existing content that doesn't make sense anymore, but also looking to see some of your content that can probably do with a bit of a refresh.

There's no perfect number, where you have to produce 300 articles a day. It depends on the size of your site. If you're the New York Times, you can produce 300 articles a day, and Google won't blink. If I’m Dave's Travel Blog and I produce 300 or 400 pieces of content a day, Google's probably going to nuke my site pretty quickly.”

You also suggest focusing on the things that are either hurting your brand or hurting your revenue. How do you define those things?

“It's one of those areas that I've talked a lot about in the past. You basically have what you could call ‘Dr. Google.’ People go to Google when they have problems or pains, and they basically want to see if they can do self-help remedies before going elsewhere.

Also, let's say you saw an Instagram ad for a really great knife sharpener. Okay, great. I'm going to go online. I’ll find out about the company. I'll look at your Trustpilot reviews. Is this a legitimate company? You will often find that Google gets it wrong, or even ChatGPT as well. Maybe its sources included a troll on Reddit who talked about putting glue on your pizza.

You want to look for the standard types of stuff: reviews, whether your company is legitimate, refund policies, etc. What you'll find is that, a lot of the time, most sites don't have many of these pages. They don't have histories of the company, they don't have a ‘Who are the founders?’ page, they don't have information about shipping, they don't have any information about complaints, etc.

The problem is, the way these Gen AI tools work is they fill in the gaps. They will find a source that answers the question, which may not be positive for what you're trying to do.

You may have a really great return policy, but because some YouTube influencer said something terrible a couple of years ago, that is now the result that shows up in Google.

It's very much looking at the particular topics, and how much is too much? For example, ‘Can I bathe in vodka?’ If you go to Google, it says: ‘No, you shouldn't. You definitely should not bathe in vodka.’ But if you look at the first organic result, it's for Quora, where some random person has said, ‘Yeah, actually, you can bathe in a tub of vodka. It's probably not advised, but yeah, you can do it. You might be a bit itchy.’

That's the problem. You get those types of results feeding into the Gen AI results, which can put you in a very precarious position. If you're a corporate SEO, before you put any content on the website, it needs to be approved by legal/regulatory/safety teams, brand teams, etc. But the problem is, the Gen AI results don't have that filter, so if you don't have that content on your website, these Gen AI results will actually try and fill in the blanks.”

I think that many brands can see the value in what you're suggesting there about ensuring that they have more reviews, more case studies, and that they have a nicer story that represents positive aspects about what they offer to AI, but it's hard to measure the value of that, isn't it? It's hard to prioritise that because you don't necessarily know what impact it's having.

How do you actually go about measuring the ROI of doing that?

“One of the ways I've done this several times in the past is we've worked with customer service teams.

If you're a larger business, and you've got a customer service team or you've got a complaints department, you can see what types of common things come into their call logs.

Say, for example, 30-40% of all your inquiries are about how to reset your password. Then, you go to the website, and you have no page about how to reset your password. You can create that simple content, and you can use a proxy metric of the number of calls and the number of email inquiries that had to be answered by some of the marketing team. How many were saved by having that page capture a lot of that traffic?

With some of the projects I've worked on, businesses estimated that they saved hundreds of thousands, even millions, by just adding one or two pages of content.

Also, mining your Trustpilot reviews or your Google Maps reviews. What are the common topics that people are complaining about? What are the common issues they have with purchasing or buying from your company? Look at those types of ones, create some great content around those topics, and use that as a proxy metric.

The number of people who go to that page, that equates to a saving where they would have emailed or contacted your customer service team.”

How do you incorporate all of this into an ongoing content marketing strategy, and how do you persuade other marketing teams to work with you on this?

“It can be a bit of a tough one. It depends on how much of a pain point it is for particular teams.

If you have a customer service team with 200 or 300 people, and you speak to them, and they're really annoyed because they get all these inquiries about how to reset a password or how to verify their account, you can do a simple check. Again, it's not going to drive huge amounts of traffic volume, but it will save the company money. It will improve the trust and authenticity of the company because people will be able to answer their questions at any time of the day. They don't have to wait for your call centre to open.

Then, from there, you look to expand it. Let’s say you sell glue, and there are a lot of pain points around whether your product is suitable for tiles. You can start to see the common questions people are having, and then use that to create product-led content – which, ideally, should drive more product sales in the longer term.”

David, what's the key takeaway from the tip you shared today?

“Focus on the things which are causing pain points for your team or your customers.

Seeing how you rank for ‘used cars’ in Gen AI, you're not really going to do anything with it – but I would look to see where there’s misinformation, false information, or potentially illegal content, where the Gen AI results have given consumers advice on how to incorrectly use your product that may lead to serious harm or death.”

David Iwanow is Marketer at The Lost Agency. Find out more over at DavidIwanow.com.