Effective AI Use Cases for Web Content Management Teams

There’s a lot of hype around artificial intelligence and how teams can benefit from incorporating it into their workflow and adopting it as part of their digital strategy. But despite the vast potential, many people never get past asking ChatGPT to draft content for them, often with less-than-stellar results. 

Understanding How to Work with AI 

To get the most out of AI, it helps to understand its strengths and weaknesses and set realistic expectations for how you can use it. Because AI technology can encompass so many things, I use it broadly to describe the ecosystem of tools, platforms and open-source apps that exist. 

What AI Can and Can’t Do Well 

So, what is AI good at? AI excels at analyzing patterns, retrieving information, and performing data analysis and visualization. It also offers valuable editing and feedback, can perform discreet technical tasks, and can generate various types of content.  

Although it may seem like generative AI is creating new ideas and unique content, it’s all just material that’s been recycled and combined in different ways. AI also struggles with discerning the quality of content; when AI produces something, it doesn’t know if it has done a good job or poor job. Machine-learning models possess a staggering amount of data and information but lack judgement. Though the same can be true with people, large language models can be prone to error and sometimes perceive patterns where none exist, resulting in nonsensical or inaccurate outputs (referred to as “hallucinations”).  

Thinking of AI as Your Assistant 

I find it helpful to think of AI as your assistant. As such, it's crucial to remember that it starts without any context about you, your business, or your customers. You'll need to onboard it by providing thorough information and prompts and maintain ongoing training to guide its actions. AI is there to extend your capabilities, not replace the critical human touch and expertise you bring. 

AI as Your Research Assistant 

Research can be very time-consuming, especially when sifting through all the noise published on the Web. AI can supplement your own research efforts by taking on some of the legwork of finding relevant content and data, allowing you to focus on analysis and decision-making. Here are some examples of how AI might supplement your research activities: 

• Discover case studies or noteworthy examples surrounding a specific topic 
• Generate statistical information based on data published from different sources 
• Provide information about frequently asked questions (FAQs) people have about a particular topic 
• Build out personas of your customers, by compiling existing data 
• Point you towards reference material and credible sources of data 

Whatever research aids you use, judgement and critical thinking are still necessary to interpret and validate insights. 

AI as Your Writing Assistant 

Writing is hard work and can be daunting, from generating initial drafts to organizing content and ensuring readability. The skills and practice required to write well discourages many people from publishing their thoughts. AI can assist by creating drafts, suggesting outlines, and optimizing content for SEO, all while helping you maintain a consistent voice and tone. Just remember that the quality of its output is highly dependent on various factors, including the quality of prompts used, the training and information you have provided it about your brand and your audience, and the quality of information it has access to. Subject matter expertise is still necessary to discern the quality of the written work and to edit and revise the output or the prompts until it is satisfactory. 

Here are some ways you can use AI to help you with your writing: 

• Build out a style guide that captures the voice and tone of your brand, so that content sounds consistent and cohesive when written by different team members 
• Develop an initial outline for an article on a topic you have chosen, based on persona pain points 
• Help you organize your content, breaking it up into logical sections 
• Performing copy-editing to help you improve the readability of your content  
• Optimize your headings and content for SEO or click-throughs 
• Reword or rephrase things to help you expand on, or summarize chunks of text 

Tip: Avoid asking AI to draft content for you. While the results may appear impressive in terms of grammatical accuracy and clarity, the content itself will be suboptimal. Because of how GPT works, it simply recycles common content that already exists on the internet—meaning your content will be shallow, redundant, and non-unique. Use AI to help you expedite your writing workflow, not to do the content creation for you.  

AI as Your Production Assistant 

AI tools can simplify your workflow by generating text, images, code, and formulas based on contextual information or prompts. They are also pretty good at converting content from one format to another. The ability to create, manipulate, and transform content can streamline how you work, enabling you to focus on the creative or strategic aspects of production. Here are some tasks GPT can help you with: 

• Convert spreadsheets or data into data visualizations like charts and graphs 
• Manipulate spreadsheets using conversational language instead of complicated formulas 
• Generate formulas that you can paste into platforms like Jira (JQL) or Google Search Console (Regex) to develop custom queries or views 
• Use generative fills to alter photographs to suit your needs 
• Generate AI images based on specific terms to reduce time spent scouring stock sites 
• Generate metadata, captions, or alt text for images 
• Convert text documents into branded slide decks 
• Transcribe videos 
• Create templates or scripts to automate repetitive image editing or content entry tasks 

Note that not all AI models are equal. Some AI image models are better at certain types of image creation than others. Similarly, your results may vary with programmatic or content-optimization tasks depending on which tool or platform you use. Fortunately, Adobe, Microsoft, Atlassian, and other large companies are rolling out AI-driven features continuously, putting all the examples above within reach.  

AI as Your Data Analyst 

Handling and analyzing large datasets can be overwhelming. AI can organize and clean data, identify patterns, reveal correlations, and provide predictive analytics allowing you to make data-informed decisions more efficiently. The beauty of this is that it allows subject matter experts and business stakeholders to interact with massive data sets in real-time using conversational language, and without the need for a developer to query and manipulate SQL databases. Platform as a Service (PAAS) tools like Google's BigQuery ML and open models like ChatGPT, just to name a few, provide a range of capabilities, including:  

• Organize or clean large, messy data sets 
• Group or merge similar entries, or delete entries meeting specific criteria 
• Analyze large data sets to identify patterns and reveal correlations and trends or sentiment analysis 
• Crawl transcripts and recordings and highlight trends or patterns 
• Produce interactive tables and charts pulling data directly from files on Google Drive and Microsoft OneDrive 
• Analyze various data sources to predict outcomes such as consumer behavior, product demand, or financial trends 

Complex AI Use Cases for CMS Sites 

By combining AI's ability to research and retrieve information, generate content, and analyze data, there are endless opportunities for how you might implement an AI or automation strategy that helps your business. Here are some recent examples of leveraging AI to solve real-world business problems: 

Using AI to Improve Clicks and Conversions 

Scenario: Your landing pages aren't converting well and rank lower than competitors for target keyword phrases. You want to boost organic traffic and conversions but are unsure which content edits will improve your search engine rankings and engage users. 

Solution: Use ChatGPT or another Large Language Model (LLM) alongside SEO tools to streamline page optimization. First, use a tool like SEMrush or Google Search Console to collect keyword performance and ranking data for your landing pages and competitors, then export this data to a CSV file. AI can help clean and consolidate similar keyword phrases and prioritize them.  

Next, upload or link to your landing page content. For any recommendations to be relevant, you will need to provide AI with context about your target audience; you can instruct it to build a persona around this profile. Supply AI with links to brand guidelines or previously published content to analyze your organization’s voice and tone so that it can replicate it. 

With a prioritized list of target key phrases, an audience persona, and an author profile, ask AI for specific recommendations to improve your page’s ranking and conversion rate. AI can suggest content changes, optimized headings, and call-to-action copy. If AI recommends adding supportive imagery, it can also assist in finding relevant data and creating visuals. For example, when producing strategic content about Web Accessibility, AI helped find data on accessibility lawsuits and created a line graph visualizing the dramatic rise over time—a 670% increase since 2017. 

Caveat: AI helps optimize and enrich content, acting as a collaborator. But for it to be effective, you need to know what to feed it, have patience for trial and error, and recognize high-quality output from poor quality. You also need the tools in place to measure the impact of the changes you are making, both in terms of SEO and user engagement. 

Using AI to Remediate Missing Alt Text in Images 

Scenario: You have a website with thousands of pages of published content, and many of the images on those pages have missing alt text. The missing alt text is negatively impacting your site’s accessibility and SEO. Manually going through each page and updating alt text in the CMS would be a tedious and time-consuming process. 

Solution: Using a tool like Screaming Frog SEO Spider, audit your site (or a section of it) for all images that are missing alt text, and export a spreadsheet that includes the file names, the URL of the webpage it lives on, and where it is stored within the CMS. You can then use AI to clean up the spreadsheet, removing duplicate or redundant data.  

Models like GPT-4 Turbo with Vision and Azure AI Vision incorporate both natural language processing and visual understanding, meaning that they can effectively recognize and describe the content of images. By utilizing their APIs, you can analyze all the images in your spreadsheet and generate descriptions for use as alt text. This new information can then be added back to the spreadsheet, which, with some scripting, can be pushed back into your CMS database. 

Caveat: A person will still need to review the generated descriptions to verify their accuracy and edit those that don’t make sense or are biased—generative AI has a reputation for producing content that can be cringe, biased, and naively offensive. As with most AI solutions, you aren’t replacing the need for an expert but rather delegating tasks and allowing AI to aid the expert so they can work more efficiently. Significant human oversight and intervention are still necessary, but you eliminate the tedious and time-consuming task of manually going to each page, opening each image, and creating a description. 

Adding an AI Copilot to Your Site 

Scenario: Your website has a ton of useful information that your customers (or internal support staff) need to access often. Your navigation works well, as does your search. But because much of your content is interrelated, you see that people looking for information need to access multiple web pages and documents, found in different parts of your site, and often leave dissatisfied with incomplete information. 

Challenge: The power of open LLMs like ChatGPT or Google Gemini is also the problem with them—they scour the entirety of the Web, making their sources of information unreliable and generalized. In addition, there may be concerns about data privacy and cybersecurity. If you are handling sensitive information, the distinction between first and third-party data determines whether you introduce compliance problems with GDPR or CCPA regulations. 

Solution: You can overcome these obstacles by integrating with an AI Copilot platform. This type of Copilot is like a virtual assistant on AI-powered steroids, trained on your content and able to be configured in a secure and controlled environment. This content can be your published web pages, documents and files, images, and even information living in your other systems, like a PIM, middleware, or CRM. Additionally, you can supplement your model’s knowledge with reputable external content or data sources that you trust, if you're not under strict security requirements.  

Now, imagine how you can improve customer experience by enabling your users to interact with your site and it being able to retrieve and dynamically aggregate information from across your systems to provide an intelligent response—informed by all your web content, help desk content, product manuals, or policy documentation. This can be a great way to supplement your customer support team, helping them retrieve information from across your systems more efficiently, or providing better self-service tools for customers on your public-facing website. 

Because the Copilot keeps the context of past session requests by the user, it can provide more helpful and personalized responses. This goes far beyond the limitations of traditional conditional logic-based chatbots and has the potential to revolutionize, if not replace, your site’s traditional search. 

Caveat: This type of solution depends on the quality and volume of the content you feed the AI model. To ensure that the responses make sense and are useful, someone needs to be trained to work with the AI system to configure settings, improve prompts, and provide the AI with directions on how to behave in certain scenarios. The cost and effort to implement this type of solution may vary based on the platform itself, your choice of CMS and its composability, and how much usage the Copilot gets. If data privacy is a concern, verify that the model, your data, prompts and generated output are hosted in a secure environment that does not directly interact with services operated by OpenAI.  

AI is for Enhancement, Not Replacement 

AI is faster at writing, researching, analyzing, and generating content than humans. It is even better at those things than people who struggle with them. But qualitatively, AI rarely outperforms experts—it’s not going to write better than skilled writers. To leverage AI in your organization, you need to treat it like an assistant and use it as a tool to help expedite the work of people who know what they are doing and can distinguish quality from crap. In other words, it enables people who are good at what they do to work faster and more efficiently by helping them offload tedious, time-consuming, and often repetitive tasks. 

A (Funny) Moral of the Story 

Photoshop is a tool I learned to use when I was doing graphic and UI design, but the ability to use Photoshop wasn’t what made me a designer—it was simply a tool. Once, long ago, I had a task to color-correct 900 images of mattresses. They had been photographed under poor lighting and all looked urine-soaked and were unusable for the e-commerce site they had been photographed for. Requiring very little of my design expertise, and a lot of my time and Photoshop knowledge, that was a long and painful week that's still stained in my memory. I sure wish we had AI assistants back then. 

So, as you contemplate how to integrate AI into your organization, a good place to start is to consider the tedious pain points that hinder your ability to deliver value. What are the things that slow you down and get in your, or your users', way? Instead of asking, “What can AI do and how do I use it?” start with the problems and explore from there.