Utilise Google Cloud’s APIs

Lazarina says: “Start utilising Google Cloud’s publicly available APIs to help you execute SEO. Google Cloud - and other giants like AWS and Microsoft as well - have a bunch of APIs that can be used by SEOs as part of our role.

To name just a few within the Google Cloud ecosystem: you have Natural Language API, Speech-to-Text API, Vision API (which is quite popular for captioning), Translation API (that can be very useful when you’re working with large-scale, internationalisation products), and Knowledge Graph API that we should be incorporating into what we do.

All of these can provide an additional layer to the work that you are doing and to the analysis that you are conducting, that will help you to better understand your work.”

At cloud.google.com/apis you find many different APIs that are available for SEOs to use. Why is Google giving us access to all of these?

“Profit is the first reason; they are making money out of it. They’re not 100% free. Personally, I feel like it makes your life a lot easier if you know how to utilise them in different tasks.

It would be interesting to also know how these APIs are used in some of Google’s other products. We know that these APIs have not been developed with the sole purpose of being released to the public, and at least some have definitely been developed as part of another product.

The Google Translate API has been released but, originally, it was used as part of Google Translate. The entity recognition Knowledge Graph is being used in search and the Vision API is being used in image search as well, as well as a bunch of other products that Google has.

It’s very interesting to see how your content, or maybe your website, looks from the point of view of the systems that Google is using. It helps you to understand how they see your website a little bit better - if they are using these APIs in the same form that they are being released.”

Should every SEO be aware of this and understand how these APIs actually work?

“‘Everyone’ is a very optimistic scenario, but yes. The reason why I am an advocate of everyone experimenting with them is because I think they give an additional layer of understanding of what is happening. It’s a very good perspective for seeing how your website looks from the point of view of a model or a system.

Also, these APIs have been developed by a massive organisation with a very big budget, and by the best researchers in the world. They are very advanced for what you get out of the box as a working model. It’s very easy to create custom reports, based on a certain use case that you have in SEO.

There are a number of different use cases for entity identification, for instance. Typically, it would be very hard to carry out entity recognition yourself - if you’re just starting to code that algorithm or you’re using an underdeveloped API. It’s much easier to start with something that has been developed by a corporation like Google or Amazon because they provide you with the code to start with, and they provide you with different use cases for different platforms. You can use this in Python or using Apps Script in Google Sheets, you name it.

You can also apply the API to different things like internal link work, identifying anchor text opportunities, SERP analysis, content analysis, and much more. That’s just a few examples of many.”

Which API would you use to identify anchor text and links?

“The Google Cloud Natural Language API allows you to identify entities and also do sentiment analysis and text annotations. It is very useful for content analysis, SERP analysis, internal link opportunity identification, and anchor text identification. On its own, it can only identify entities but, as SEOs, we collect a large amount of data.

You can export the report from this API and pair it with crawl data and with search volume data, and you can build a very interesting picture of the correlations between different data points. From that, you can come up with a lot of scenarios about what you’re doing day-to-day.”

Why do SEOs need to be using these APIs directly? Are SEO tools and platforms not using these APIs themselves?

“I hope that they start incorporating some of these APIs at some point, but the cost is obviously a barrier to having this implemented. I also think that being able to push this to production for a live system or tool might be a little bit complex and may take more time.

The time barrier is a big one. These deployments even take a long time at Google. Daniel Waisberg recently spoke at BrightonSEO, and he walked us through the process of releasing a Google API. It takes about a year, and that’s for an API that was already available to them. It’s a lot slower to wait for a tool to integrate that API.

It’s important to understand that you can start using these APIs without having any coding experience. There are Google Sheets templates available for almost all of them and you can quickly incorporate a very advanced level of understanding and analysis to your reports - and your day-to-day work.”

Why should SEOs choose to use Google’s Speech-to-Text API?

“From my personal testing, Google’s Speech-to-text API has been a lot more precise at recognising speech. They also do more as well. Let’s say that you have 500 YouTube videos as an SEO. If that content exists in video form, there’s a good opportunity to have that published in a textual form on your website as well. Having a transcript for that - or converting it to a blog post - is a very easy way to repurpose that content, depending on the type of content that it is.

It’s very common to have Speech-to-Text problems in that kind of scenario, and having an API that actually understands sentences and the complex concepts that might be mentioned in those videos is very rare. It’s even more rare to have proper punctuation being put into the sentences that are being transcribed.

If you have this problem, and you’re working with a client like that, it’s obviously important to find an API that does all of these things: understands text, understands complex entities, and understands punctuation - and uses those to the best of its ability. Google’s Speech-to-Text API is good at solving all three problems.”

Is this more relevant for larger, enterprise websites with thousands of pages than it is for smaller websites with fewer videos and fewer tests to carry out?

“I would argue that these APIs are especially useful if you have a smaller website because you probably have a very limited budget as well. The budget that you would need in order to commission someone to transcribe your videos would be quite big in comparison to having an automated API do that for you with a click of a button.

There’s a common misconception that using advanced, technical APIs and machine learning is something that should only be done for large-scale enterprise organisations. Actually, it is even more useful for people that don’t have the budget. Just five videos could be costing you the budget of a full PPC campaign, or maybe even publishing five new content pieces. You’ve taken that budget and given it to someone transcribing videos.

It’s all about removing the barriers to how difficult people perceive it to be. People think that machine learning with APIs is challenging, but once you know that everything can be executed fairly easily then it can replace a very large chunk of your budget - even if you’re in a smaller organisation.”

If someone hasn’t worked with APIs before, is there a lot of technical knowledge required? Are there things like plugins available for popular CMSs, to get things up and running quickly?

“I give a lot of props to Google Cloud in terms of the documentation that they provide for all of their APIs. They not only provide documentation on how to get started, but they also provide documentation in different languages. One of the languages is Apps Script, which allows you to build custom functions in Google Sheets. Once it’s been built, the only thing that you need to replace is your API key.

I don’t know which CMSs have managed to make these APIs readily available yet, but Google themselves have already done so. You just have to copy-paste the code from the Apps Script that is on their website, provide your API key after you have registered, and then you can start using it in the data that you already have in Google Sheets. There’s not really anything further that you need to do – it’s literally as simple as one or two steps.”

What shouldn’t SEOs be doing in 2023? What’s seductive in terms of time, but ultimately counterproductive?

“Stop keyword stuffing and optimising content based on a singular keyword. We have very much moved past keyword-based optimisations - in terms of content and also in terms of how we should be tracking the performance of the content that we write.”

Is it okay to have a core target keyword on a page, but then write naturally around that particular topic with a view to ranking for as many related keywords as possible?

“You just need to focus more on what the topic is covering, even if it doesn’t include the keyword that you’re writing for. A lot of SEOs still get really hung up on the concept of keywords in their strategies and research, without incorporating the topics, subtopics, and other keywords and entities that make that content come to life.

Your research shouldn’t just include the main keyword or a related keyword, it should also include the questions that people ask and anything else that makes a website or piece of content an authority on a topic.

That kind of research can go both ways. It can be something that is used to identify an opportunity in a marketplace, but it can also be used to identify a place where you shouldn’t create content as well. It’s important to start doing that type of research, as opposed to collecting keywords and optimising based on just one of them.”

Lazarina Stoy is SEO and Data Science Manager at Intrepid Digital, and you can find her over at lazarinastoy.com.