FALCON AI: our in-house AI for outstanding SEO Content

FALCON AI is our in-house AI specifically trained to increase the performance of SEO Content on WordPress topics.

FALCON AI is Ellipsis’ proprietary tech. It’s trained and built in-house, and we use it to improve SEO performance for our Content Growth clients.

On average FALCON AI improves the performance of SEO Content on WordPress topics 30%, but we’ve seen as high as 350% improvement.

FALCON AI consists of a number of modules. All of these are our own in-house tools. This page gives you a sample of what FALCON AI can do. If you’d like to know how FALCON AI can help improve your SEO Content results, do get in touch to arrange a chat.

AI can predict whether a post works or not — so we only do posts that will work

Keyword selection is vital for SEO Content. If you pick the wrong keyword, you’ll never get the rankings or sales you want from your content.

FALCON AI makes sure our Content Growth clients always get posts with a high chance of success. We use a custom machine learning model to predict if a given keyword and title combination will succeed.

Because we can predict if keyword and title a will succeed, we can also predict if keyword and title b is a better alternative. The AI can generate multiple titles based on what’s currently ranking top for the keyword, and then we predict which will work best.

Not only are we only selecting keywords with a high chance of success, we’re also making sure our titles perfectly match the search intent too. The process looks like this:

FALCON AI can do this at scale, letting us rapidly evaluate every single possible keyword that might fit a client’s product.

FALCON AI will also automatically group keywords by topic cluster, letting us easily build out areas where we want to improve a client’s subject matter expertise. We do quarterly SEO planning for Content Growth clients, and this is an important part of that. Building tools in-house lets you personalise them exactly for your needs 🙂

Google rewards content that answers a searcher’s question; FALCON AI makes sure we cover EVERY question

SEO in the 2010s was all about the keyword: select a good keyword, and then include it as many times as you can in the content. SEO in the 2020s is about comprehensively answering questions the searcher has about a topic.

FALCON AI makes sure we answer EVERY question a searcher has.

Once you start creating content this way, it’s impossible to go back. We want to make sure we’ve done a better job than competing content of answering questions a searcher has. We do this by looking for the topics covered in competing content, and covering all of it.

Once we’ve chosen a target keyword, FALCON AI looks up the top ranking content and runs its custom-trained Natural Language Processing (NLP) on each result. The NLP extracts the key topics each result covers. We repeat this on all the top ranking results for the keyword, and then collate the list of topics.

The output is a list of every topic that top ranking content for the keyword covers. We then create the content outline and write the content to match the list — and we’ve guaranteed we’ve answered all possible questions.

Google rewards this content with the highest rankings. This is how we can have confidence that our Content Growth clients will ultimately win with Google.

To be clear: we’re not AI-generating blog posts

There’s a lot of hype around “AI copywriter” tools like Jarvis. These tools are typically powered by GPT-3 from OpenAI. GPT-3 is a huge technical achievement, but AI generating entire blog posts is a really terrible idea.

Google rewards authoritative, expert content. AI-generated blog posts are the opposite of this: the AI has no notion of what “truth” is. It’s not an expert in anything and does not understand what it’s writing.

At a time when AI-generated content is going mainstream, we’re spending more on writers and subject matter experts. We will continue to produce the highest quality content, and we’ll do that with expert humans.

We are using GPT-3, but it’s fine-tuned GPT-3 for really specific tasks (like title generation) where we can tightly train and control the model. We’re not generating blog posts. By understanding and leveraging the technology where it can truly add value, we’ll win for our clients.