Track more than basic keywords

Andy says: “In a world of generative AI, keyword tracking as we know it is no longer good enough.”

Why is keyword tracking no longer good enough?

“The way we track performance in SEO has historically been through organic sessions, organic revenue, organic lead generation, and that kind of thing. Obviously, that is still very relevant, but it's also been about keyword position tracking as well – tracking your individual single keywords, or long tail keywords, or whatever it is you're tracking.

To an extent, that's still fine. However, you can’t ignore what’s going on. If you're not tracking outside of that, you're not tracking your performance as a whole. With AI overviews, AI Mode, and ChatGPT, people are searching in so many different ways now. Tracking keywords is not everything anymore.”

Is tracking keywords not as big a part of the picture as it used to be?

“Exactly that. At Reflect Digital, we're doing a project called SearchPulse, where we're looking at how people are searching over time.

With ChatGPT and other methods of searching, we want to see how people are changing their ways of searching over time, whether they move away from Google or search in different ways depending on their intent.

The intent is the most important part. If someone's looking to make a purchase or a transaction, they typically still use a normal Google or Bing search, using normal keywords like ‘blue jeans’ or whatever they’re looking to buy.

However, when it comes to the awareness/informational stage of the journey, before they've made that decision of what they want to buy, people make that journey through many different means, many different search engines, social media, etc. Monitoring that is the most important part.

SEO is about generating a lot of traffic, not just for transaction keywords. A lot of it is from the awareness stage as well.”

What does the awareness stage of the buyer journey look like now?

“When you look at fashion or holiday searches, where someone's looking for inspiration, people are going on social media or an LLM (more often than not) to look for inspiration on what to wear or where to visit next.

Once they've narrowed down that information, through searching on ChatGPT, TikTok, etc., they'll come back to Google and make that traditional keyword search that we're used to tracking.

The bit beforehand is what’s hard to track, but it's super important.”

Does that journey differ across different types of businesses?

“Absolutely. It is going to totally depend on your audience.

We have a couple of clients in the travel sector, and they need to be monitoring TikTok, social media, etc. On the other hand, if you're in the financial sector, that's probably not going to be necessary, because people aren't going to be searching for financial advice on TikTok.

However, they might be searching for financial advice on ChatGPT. They may be asking a bigger question and looking for advice and information on where to go for this problem or what to do in a particular financial situation.

It's definitely worth doing your research into who your audience is and what they're searching on, and then it's about monitoring your presence in each of those platforms.”

What research do you need to do in order to determine where your audience is definitively residing?

“Speak to your customers, first and foremost – whether that be follow-up communication after a purchase or a survey on website exit. Ask people how they found you.

There's no getting away from real customer user data. It’s the most truthful thing you can get. If they've made a search on an AI result and then come back to your website by typing your domain name into the browser, you're never going to get that information without asking the question.

If you don’t have enough of a customer base yet, you could survey a specific set of people in your particular niche to find out that information and make those inroads. There are a couple of options, but just ask the question. People are happy to share their information if you ask them.”

How do you report on your presence in places like AI overviews, ChatGPT, and social media?

“There are a couple of ways. Reporting on AI overviews, ChatGPT, etc., is totally doable. There are lots of tools at the moment that are doing that. Some of them are really expensive, so do your research because there's a lot out there, and a lot of new platforms are coming up. Every single day, there's a new platform that can track your presence in something.

The most important part is, when you're tracking, it's not about tracking your domain name. If you’re Adidas or Nike, you want to show up for ‘What's the best shoe brand?’, but it might not be your brand domain name that's being referenced in that search result. It might be a review website that was mentioned in that AI overview or in ChatGPT.

What you need to be tracking is your brand name mentions as opposed to your brand website mentions, if that makes sense.”

If you’re losing a lot of the user journey, how do you measure the ROI of brand mentions or citations in AI search engines?

“That's the tricky part, especially at the moment.

AI overviews are currently really prominent, and Google's not even giving information about how many impressions or clicks we're getting from those. You can't filter that down in any way, which is frustrating. Hopefully, in the future, they’ll be giving us that information. Fingers crossed.

The hard part is obviously translating awareness stage traffic back to transactional performance. It's about getting your clients or the people in your business on board with the importance of that stage of the journey.

If you're used to reporting on last click attribution, where you only report on the clicks that generated a transaction, think about how you can change your internal reporting to also be about first click or multi-click attribution, where organic has been responsible for a conversion at any point of the journey.

If someone comes in through a paid ad, then goes back a few days later through an organic search, then finally converts directly a few days after that, you still need to make a point to your team that, in the middle part of the journey, organic contributed towards that conversion at the very end. That will help demonstrate the importance of the awareness stage.”

Why should you be tagging prompts?

“If we're tracking AI overviews, ChatGPT, or whatever it might be, you will want to track a lot of different prompts – with prompts obviously being the questions or queries that you want to track.

The example we had earlier, ‘What's the best shoe brand?’, is a really generic example. You want to have some much more niche examples for your business. You will have a range of different prompts that you want to report on.

As a business, you also probably have different products or services that you would want to tag in some way. As much as it's a good idea to look at your visibility or performance in AI overviews for your website generally, you need to tag them or group them in some way, like you would do for keywords.

That way, you can report to your team and say which products you're doing really well for, and which products you’re not doing so well for. That will give you some information for your reporting.”

What software do you use for reporting?

“There are lots of different platforms available. Peec AI is a very good tool at the moment. By the time this gets released, this information will probably be completely outdated, and there'll be something new and amazing on the market.

Profound is good as well. Semrush is also a good platform for it.

It’s about finding a platform that's doing the searches for you. It's going to search a prompt on ChatGPT or whatever it is that you're tracking and report back on what it saw on that day, and it will repeat that search every single day and aggregate that data for you. There are lots of platforms, and it's then about how you utilise that data.”

Is sentiment analysis increasingly important, and what do you do with that data?

“We've had lots of clients asking how many mentions they have got in the AI overviews, or how many times they are listed there. It could be as many as you want it to be, but if all of those are saying, ‘This is a terrible brand, don't buy from here because of X, Y, Z customer feedback,’ you need to have that information to hand.

Going back to the earlier point about tagging, if you can see that sentiment for one product is generally very positive, but your other product is generally poor, you could use the reporting to look at why AI says that product is poor. Where's it getting that information from? Is it from customer reviews? Is it from a product review somebody did?

If it is, can you influence that? Is it that you need to improve the product, or do you need to do some positive PR and try and change that narrative? Monitoring sentiment analysis is really important because just having the number of mentions is not what you want.

A lot of the tools I mentioned have sentiment analysis. Some do, some don't. If you want to monitor your performance on AI overviews or otherwise, make sure they do have sentiment analysis because it will really allow you to dig deeper into your data.”

How do you benchmark your performance against your competitors?

“Just as you would with keyword tracking, consider who your competitors are going to be and track those in whatever tool you are using. You can then monitor that over time.

My recommendation would be to track that as soon as you can, because the longer you've got the historic data, the more you can do with that information.

It's the same as any other platform. Once you put in those competitors, you can then monitor their visibility over time, but also their sentiment over time. There's visibility in different platforms as well, so you might see better performance in ChatGPT than in AI overviews, for example, and maybe vice versa for a competitor.

It's about where your competitor is getting positive sentiment, reviews, and mentions that you're not.”

What software do you use to monitor this, and how does that inform your regular SEO strategy decisions?

“It depends on how you want to visualise it, in terms of reporting. We use Looker Studio so that we can brand it properly from our side. Bear in mind, we're a digital agency, so we want to provide a good experience for our customers. We want to provide them with that information in an easy-to-use dashboard.

That said, a lot of tools allow you to see good visuals over time in the platforms themselves. It's whatever you're happy with.

In terms of using the information, it's about drilling down into the detail, and that's where you need to find a reporting solution that's good for you. The details will give you information on where X brand is mentioned in Y article, and that Y article is cited however many times across however many different prompts.

For ‘What’s the best show brand?’, maybe there's a particular article that's really powerful because it's in a really strong domain written by a really influential author who says this is the best footwear you can get. Maybe that's where it's getting its information from.

If your brand isn't in that article, you need to find a way to get in there, be cited, and be referenced from that article because AI is using it. That's what the reporting would help you to do.”

What does an effective SEO strategy for this look like?

“It is going into the detail, but it's more of a branding or digital PR exercise, really. SEO is blurring the lines between SEO and digital PR with this kind of performance and the tracking that's required.

It's not just a case of giving that link to your client and saying that they need to go reach out to that publication and ask them to put your name in it. Maybe it's not suitable, or it's not relevant. Maybe you need to go with a different strategy, and approach that person or publisher with some information or a takeaway that means they will want to rewrite the article based on the information you've given them, or publish a whole new version.

There's an exercise to be done with your clients, your brand team, or your PR team regarding how you can position yourself to be referenceable in those kinds of articles.”

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

“Avoid regular keyword tracking. Look at your brand performance overall, especially for generative AI platforms.

If you're not monitoring how you're performing in generative AI now, you'll be falling away in 6-12 months' time. You'll be wishing you had done it earlier.”

Andy Mollison is Search & Innovation Director at Varn. Find out more over at Varn.co.uk.