Artificial intelligence – AI for short – is being tested in more and more disciplines in online marketing.
The Techpally agency tested how AI can be used in search engine optimization.
We spoke to the group CEO Akin Chaktty about the first findings of his SEO-AI experiments.
Search engine optimization can be divided into different task areas and processed with SEO
In general, a distinction is made between on-page SEO and off-page SEO.
Table of Contents
Onpage vs. Offpage Optimization
While off-page optimization focuses on optimizations outside of your own website, we will now look at on-page SEO.
In addition to technical SEO, in which the website is optimized and accelerated by changing the code and hosting settings, content SEO is one of the areas that can be optimized with artificial intelligence.
The CEO explains that search engine optimization always starts with analyzing the status quo.
In the case of new web projects, this includes above all keyword research, which to date has quickly taken half a man-day to a whole man-day.
The result is a list of relevant keywords for which content must be produced.
In the last quarter of last year, the business ally boss started the “AI and SEO” experiment.
SEO experiment using AI
The SEO Scientist – as a technically experienced search engine optimizer called Biz Pal – has supplied an AI system with various input data.
The AI engine used is the BigML platform. The machine learning platform offers data models for supervised and unsupervised learning (supervised and unsupervised learning), such as decision tree (decision tree), random forest (random forest) and deepnet.
To operate such a platform, one should have a basic understanding of how machine learning works.
In addition, it is important for the interpretation of the results to master mathematical knowledge such
as linear algebra, discrete mathematics and statistics. Programming knowledge is not required.
The data should not contain formatting errors, the variable declaration must be unambiguous, and the data size should be 5,000 to 6,000 records for simple predictions.
For larger calculations and questions, more data sets can of course be used.
He imported a total of 80,000 data records as input parameters for BigML, including keywords, CPCs (costs per click) and search volumes.
The output of the artificial intelligence is a list of specific keywords for which the content is to be produced or optimized.
The agency needed six months from the initial idea to the ready-to-use model.
Other similar models would be usable in 1-2 months. In the meantime, with AI in the keyword analysis, a result is possible within a very short time, so speed and time savings are becoming a USP in search engine optimization from Deiwick’s point of view.
In the production of content, the Businesspally agency is always testing new solutions for AI and test automation.
So far, the company has not found anything, so the actual production is still carried out by editors.
The head of the agency sees a growing need for support among companies in the area of content production.
Deiwick also mentions podcasting and infographics as a growing service portfolio of the agency. (Read our article on Video SEO here ).
The agency develops the necessary skill set for SEO and other work internally.
New employees are trained in the agency using complex onboarding and induction processes.
The training takes place almost exclusively digitally. Most colleagues work from home.
Hartmut Deiwick tells us that online meetings are also preferred to physical visits by customers.