AI Driven Digital Marketing

AI Driven Digital Marketing

The use of AI driven Digital Marketing techniques is booming in 2021. In this post we look at the impact of Artificial Intelligence in hosting, website design and advertising and discuss the market leaders in terms of the applications enabling the data driven approach.

We’ll begin with the website itself and trace the impact of AI in the hosting environment, the design, content and marketing.

AI Driven Digital Marketing

What is AI?

AI probably began with Alan Turing‘s paper Computing Machinery and Intelligence published in 1950 and has a lineage travelling through IBM’s Deep Blue, the supercomputer that was famously trained to beat the chess grand master Gary Kasparov and the same company’s innovative Watson, the system that combines analytics and AI to produce commercial systems that are used in healthcare, law and dozens of other sectors.

There are many definitions of Artificial Intelligence today but the one I like best reduces the term to a simple, clear recipe

“At its simplest form, artificial intelligence is a field, which combines computer science and robust datasets, to enable problem-solving”.

IBM Cloud Education

Deep Learning

In its current form, AI is a system that can be trained by analysing existing data, finding relating patterns, determining the impact of these patterns on an outcome and applying a behaviour to produce a desired outcome.

This is most easily exemplified in terms of computer security. A system learns from recording website request data, determining which requests are hostile and which are not. It detects patterns and associates these patterns with outcomes. Being able to tell the difference between a Computer Hacked or an Admin login.

AI in the Hosting Environment

The hosting environment is not a facet of digital marketing exactly, but artificial intelligence is a vital part of the modern website’s environment and so we’re including it here as contributing to a 360° view of the subject.

AI is used in modern hosting environments to provide lightweight, agile, security solutions that leverage offsite data to protect individual sites.

Of course no self respecting hacker is going to use a single IP address to launch their attacks from, but in addition to using global databases of known IPs these systems employ sophisticated pattern recognition techniques to differentiate new hostile attacks from bona fide attempts to log in to or edit a website.

By watching for the hostile patterns, helped by training in the form of local whitelists and global blacklists, plus dynamically added data in the form of historical and new patterns, an AI powered security system is able to either block any attempts to access a system at all or deflect the intrusion into a harmless replica of the system, an approach that is highly effective as it draws the hacker in, wasting their time and learns more from watching what they do.

These systems work at server level and application level.

Bit Ninja

Bit Ninja is a cloud based software used in hosting environments. It is comprised of the following components.

Realtime IP Reputation

IP addresses soon pick up a reputation for nefarious goings on and so most of the leading security applications refer to global databases of blacklisted IP addresses to shut out this line of attack instantly.

Honeypots

Honeypots emulate existing services on decoy ports so that the hacker thinks they are accessing something they are in reality blocked from. Any IP address accessing one of these ports is automatically blacklisted from accessing the real applications.

Web Application Firewall

Web Application Firewalls work as a reverse proxy between a web browser and an application, governing access to applications. filtering all incoming requests.

Log Analysis

Bit Ninja analyses server logs in order to pick up patterns of potentially malicious intent. Such as repeated attempts to login to the same user account from different IP addresses.

DoS Detection

DoS attacks are detected by monitoring the number of connections to a server from the same IP address in real time. When too many concurrent connections are spotted the IP address is initially blacklisted to stop the attack progressing and then grey listed to allow human intervention from the administrator.

Malware Detection

Malware detection analyses filesystems for files with signatures matching known malware and analyse the structure of the code itself to find what is known as obfuscated malware – a malicious program embedded in code that may be doing something perfectly normal at first sight.

Wordfence

Wordfence is a popular and very effective WordPress plugin that applies similar techniques at application level.

Wordfence provides another layer of endpoint protection meaning that if an intruder manages to bypass server level or cloud based protection, they have to jump through a different set of hoops all over again. The effort involved in doing this would inevitably trigger the server level security so the whole effort would be rendered useless.

AI Driven Digital Marketing

Recently, the digital marketing community have discovered AI in a big way. This is important because it will transform the industry from one that at its current worst advocates a fluffy DIY approach to digital marketing into a more sophisticated industry whose best practitioners use data to inform every aspect of their practice from website design to pay-per-click marketing.

My own company Helter Skelter Digital use AI throughout the digital marketing cycle, from keyword analysis and content planning through website design to outreach in the form of advertising.

Data Driven Website Design

What do we mean by data driven website design? Websites are fundamentally marketing machines. There are two ways in which Artificial Intelligence can be used to enhance website design.

  • Choice of Content
  • Content Generation

Choice of Content

If a website is the public face of a business, then it needs to be visible and that means SEO. The art of choosing content for a website is about optimising that site so that it can be found in Search.

What this means in practice is breaking out the structure of the site into silos aligned with business offerings, and analysing the competition in each silo. This has traditionally been dome manually, using spreadsheets and tools such as Google Keywords.

More recently, tools have become available that use AI to do the research, identifying competitive keywords and generating suggestions for both the silos and the keywords within them.

Surfer SEO

Surfer SEO is a tool that includes

  • Keyword Research
  • SERP Analyser
  • Content Planner
  • Post/Page Audit
  • Content Editor

Natural Language Processing is supported and the tool’s most radical innovation is it’s method of analysing the competition in terms of the top ranked posts on the same keyword in Google.

Surfer SEO

Again, this requires some knowledge of SEO in that the user needs to appreciate that finding competitive keywords is important. The cliche that you “find your competitor’s content and improve on it” is often misunderstood. It’s easier to write better content than it is to face the fact that your competitor is not IBM but “Joe’s Software Emporium” – you won’t be successful by competing with giants. IBM will outrank you 99.9% of the time, even if your content is better.

The Keyword Research Tool and the SERP Analyser help you to establish the keywords that are realistically achievable. The Content Planner then breaks out the main keyword into associated silos that provide valuable pointers to possible content.

The audit and content planners feature an editor and a score section that goes way beyond what most people will be familiar with from their Yoast or Rank Math accounts. Where SEO analysis tends to focus on a primary keyword and offers little support for secondary and contextual keywords, Surfer analyses the content of the top ten ranking posts on the same keyword and analyses the structure of the articles, common words and phrases, headings etc.

I have found that using the Audit process on existing posts invariably produces an increased ranking and more readers.

I’ve used the software to devise content plans for a few of my own sites now as well as my clients, check out my in depth review of Surfer SEO.

Content Generation

Content comes in many forms, articles, blog posts, Facebook, Twitter and Instagram posts and so on. Typically these require totally different approaches. Pillar content is typically long form, up to 3000 words, Facebook posts often around 100. Twitter is measured in characters.

Gaining viewer’s attention in a world of scrolling requires different techniques ranging from reliance on graphics to eye catching headlines, outlandish claims and more subtly, well placed links in thoughtful and original content.

Jarvis

Jarvis is the market leading content generation software enabling copywriters to produce ad content, social media content and website content more quickly and persuasively than most humans can manage.

Jarvis - AI Copywriting

Focused on high converting copy, Jarvis is trained by established marketers in multiple conversion oriented frameworks such as AIDA and PAS. Multilingual (25 languages and counting) and reared on internet content, Jarvis bases new content on analysis of multiple related sources to avoid plagiarism.

Creating content in the newly introduced “Boss Mode” is an iterative task involving simple instructions and actions on the output which combined, form recipes for various content types. In skilled hands, this copy is indistinguishable from that produced by human copywriters.

An example of this approach, using boss mode to create a blog post looks like this:

  1. Write some blog post titles for the topic of goldfish (Choose a good title from the output)
  2. Write a content brief for the above blog post (Use the best part of the content brief)
  3. Write an intro about the above (User the intro to feed the next instruction)
  4. Write an outline for the above (Choose the best bits)
  5. Write a blog post conclusion or
  6. Summarise the above

It should be born in mind that Jarvis is not a subject matter expert, you provide that expertise, so where this approach produces good content for say, a post entitled “10 things to do in Granada” it is not so productive in technical niches such as Robotic Process Automation.

There is an enthusiastic and friendly community built up around the software and we have been using Jarvis for short content successfully for several months. Check out our website Helter Skelter Digital to see examples.

I will write an in depth review of Jarvis presently, so keep an eye on this space!

AI in Advertising

The complexities of mastering pay per click advertising have generated an entire industry of specialist service providers. Both Facebook and Google are now introducing artificial intelligence into their advertising systems so that AI can be empowered to compose bidding strategies, etc. based on realtime trend analysis and confined by human moderated budgetary restrictions.

Facebook

Currently Facebook uses AI to choose which ads to show to which user. This is governed by an algorithm that matches the advertisers chosen audience with a quality score attached to the ad to determine the chances of an individual clicking on a displayed advertisement.

Over time, the actual click rate attached to an ad is fed into the algorithm, producing an ever more accurate set of data.

I have found that the drawback to this approach is that in any given ad set, Facebook will start to channel clicks to a single ad, effectively excluding the other ads. If the intention of an ad set is to show different varieties of the same product, this result is not helpful. We will often temporarily suspend an ad that becomes too successful in order to force a more equitable share of the budget across the ad set.

Google

Google is actively introducing AI into its Google Ads interface with the intention of automatically optimising their customers ad spend. Again this works by analysing click-thru rates on real advertisements and combined with their dynamic ad content capability that scrapes content from the landing page is intended to deliver increased conversion rates.

It’s too early to tell how effective this will be or whether it will fall into the same trap that Facebook has, but there is huge potential here. The challenge will be around innovation and creativity. Advertising thrives on these two qualities and both are extremely difficult to model. If all advertising ends up looking and sounding the same then that will render ads invisible.

Conclusion

As AI becomes more and more ubiquitous in our day-to-day lives, it’s also becoming a real force for disruption in the world of digital marketing. The challenge will be around innovation and creativity as marketers try to stay ahead of the curve when it comes to using artificial intelligence (AI) tools.

One of the common complaints about AI across all industries is that it will put human beings out of work and to be fair we’ve seen plenty of evidence of that. The Corporate world is relentlessly growth focused so innovation is often seen as a means of replacing expensively salaried employees.

In the marketing industry, Facebook and Google will also put more pressure on the industry as algorithms replace human expertise in devising high converting ads and optimising CTR and Ad Spend. The challenge for humans is to drive the process rather than being driven by it. That will mean more focus on digital marketing strategies and on coordinating multi channel marketing to deal with a world where your audience roams across devices and channels at will.

What is your experience with AI? Let me know in the comments below!

Geek Friendly?
Add our Insight direct to your mailbox!

Subscribe..

Each Month..
Actionable tips to put your marketing and your technology front and centre..

Further Reading

Surfer SEO Review

If this was useful - give it a share!

Similar Posts

Leave a Reply

Your email address will not be published. Required fields are marked *