As discussed in his previous article, Dave revisits marketing applications of AI, this time with SEO.
As I explored in a previous article, there are many emerging marketing applications of AI. Within the article I looked at 15 options across marketing, but didn't drill down to look at the benefits of using AI in marketing in detail. Here, we'll focus on SEO...
The potential of AI in search marketing is highlighted by a change in Google's own positioning and strategic focus shown by Google's CEO Sundar Pichai announcing at Google I/O the change in emphasis as an important shift from a mobile first world to an AI first world. Of course, AI has long been part of Google's technologies, with RankBrain machine-learning artificial intelligence system used to help process its search results as part of the Hummingbird.
Since then Google has encouraged its engineers to apply AI and machine learning concepts throughout Google's services. Machine Learning is an important subset of AI for marketers to learn about, since this involves analysis of historic trends and patterns to identify future opportunities and problems. Many Google and vendor AI features are based around Machine Learning and Analytics so this is where I recommend search marketers focus on in the their learning.
For example, Smart Goals uses AI to identify the best quality sessions for marketers using AdWords who don't have the experience to set up goals (which is still the recommended approach).
In Smart Lists, machine learning determines which users are most likely to convert in subsequent sessions for use of remarketing in AdWords where retargeting hasn't been set up for future.
Smart Bidding is a set of conversion-based bid strategies based on Target CPA, Target ROAS and Enhanced CPC – that use advanced machine learning to help you tailor the right bid to each and every auction. It factors in a wide range of auction-time signals including device, location, time of day, remarketing list, language and operating system, which analysts may find difficult to assess.
So, what can we as search marketers do to apply Artificial Intelligence to organic search marketing?
Complexity of analysis is one of the biggest challenges for search marketers. As SEOs we need to understand which keywords to target, which entry pages attract organic and paid visits and how these compare to competitors. We also need to understand the quality of backlinks from other sites and where they link to in affecting ranking. Complexity increases when we need to review these through time to understand the reasons behind changes in our search performance to take corrective action.
AI can help here as a 'robo analyst', particularly for large, complex sites targeting many keywords. AI can potentially perform the time consuming, repetitive tasks traditionally performed by analysts. I say potentially, because it's still a work in progress if you look at some of the tools available like Google Analytics Intelligence.
I've been a fan of Google Analytics Intelligence features since they launched 10 years ago. I never found their early efforts at automated alerts to problems that helpful - they didn't have true intelligence - it's not helpful to know that smartphone visits in California to a lesser page is important. However, the customer alerts you could set up where you could apply your own intelligence were more useful. For example, to create an alert for when month-on-month organic traffic is down by more than 10%.
With time being short, it's rare for teams to put these manual alerts in place, so it's better if you can be alerted automatically. Google's relatively new Automated Intelligence feature shows warnings in the top right of Google Analytics.
Today's Automated insights in GA can show spikes or drops in metrics like revenue or session duration, tipping you off to issues that you may need to investigate further. Insights may also present opportunities to improve key metrics by following specific recommendations.
For example, a chance to improve bounce rate by reducing page loading time, or the potential to boost conversion rates by adding a new keyword to your AdWords campaign.
You can ask questions in natural language which are then answered. For example, try this in GA:
- You asked: Biggest drops in page views for organic search?
- GA then returns with a list of the pages with the biggest drops in page views:
- Bottom Page by Week over Week Growth of Page views for Default Channel Grouping of Organic Search.
- If you find a useful insight you may want to see regularly, you can save the insight to pin the query to the Insights bar.
It's a huge improvement on previous versions of GA Intelligence, but it's a long way off a 'robo-analyst'.
For an analyst, one question usually follows another to dig into a problem or opportunity, but the questions aren't integrated currently. In search you're often looking for changes in different categories of blog or product pages, but Intelligence can't help you there. However, automated insights are intelligent enough to understand concepts of different time periods, so month-on-month and year-on-year drops are reported on.
As well as understanding changes in organic search performance, AI can potentially help identify the cause of the change. The right type of system which has complete data on competitors also like SEO analysis platforms, like AHREFS, BrightEdge, Majestic, Moz or SEMRush, can help determine whether it is competitor action that may have caused disruption. AI can help create alerts and give recommendations which answer questions like: "Are competitors now performing better for certain keywords, is this change due to changes in site authority based on link-building or some other reasons?"
The larger enterprise organic search reporting platforms are most likely to innovate first, since they have the biggest R&D budgets and their clients' large-scale enterprise clients can reap the benefits best.For example, at ClickThrough we use BrightEdge to report and analyse our clients' SEO performance, since it is one of the most powerful tools available. BrightEdge Insights uses DataMind, an AI solution. BrightEdge explains that:
Acting like a marketer’s personal data analyst, Insights continually works in the background, analyzing billions of data points, filtering out the noise, and identifying changes outside the normal behaviours. BrightEdge Insights then prioritizes the most relevant findings, so marketers can focus on the initiatives with the biggest impact to their business. All of this is done at a scale and at a speed not possible for an individual.
This tool also highlights a pre-requisite of other capable AI solutions, which must combine data from different sources including website analytics, Google Search Console, and link-databases.
More granular insights can be provided if the tool shows the types of SERPs entries and how they differ on mobile against smartphone, e.g. featured snippets and images since these have now become so important. Technical problems resulting in 404s or declines in traffic can also be identified.
In future, it's likely innovation in AI use for organic search will be provided by third-party tools which are more focused on search marketing. The benefits are clear, as Jim Yu of BrightEdge explains:
The unifying thread through all of this is the fact that AI can deliver highly relevant insights automatically, at huge scale, and in a manner we can easily share with other departments in our organization. Without the right technology, we could only achieve this with the support of hundreds of analysts and an infinite budget.
It is worth noting that the difference between a valuable insight and a simple observation is incredibly significant for any business. A true insight illuminates something new and guides future action based on the moments and metrics that matter.
If you'd like to find out more about applying AI Intelligence to search marketing, get in touch with our SEO team. We're always happy to share our insights and discuss them with you in detail!