Artificial Intelligence

“If you build it, they will come” worked well in “Field of Dreams.” But, like a lot of marketers, I took a more cautious (yet still curious) approach toward AI integration. When ChatGPT first hit the scene, it was exciting to see all the buzz but it left me hungry for more: What is AI actually good for? How does it work in a business environment? And can it save me time, or am I going to rewrite absolutely everything it produces? Well, like 75% of marketers, I believe AI will become a workplace staple in the next couple of years. And that will have a lot to do with how AI-powered tools help us automate our daily work. Let’s explore the idea of AI workflow automation and how marketers stand to gain their time and lives back from repetition. Table of Contents What is AI workflow automation? Why use AI workflow automation? What You Need to Start AI Workflow Automations How to Implement AI Workflow Automation for Your Team 5 Best AI Tools to Automate Workflows What is AI workflow automation? AI workflow automation involves using artificial intelligence tools and resources to make work processes easier and more efficient by handling repetitive tasks, informing decisions, and letting teams focus on higher-impact tasks. Something to consider in workflow automation is what types of AI can work within your pipeline. Most of the online chatter has focused on generative AI (tools like ChatGPT or Claude). We’re now seeing agentic AI emerge as a discussion point in company boardrooms. Instead of creating something as generative AI does, agentic AI accomplishes specific tasks autonomously. You’ll see these subtypes as you explore workflow automation. Each has its uses, and they typically work together to automate workflows. But know that they are designed for slightly different use cases and be aware of it in your planning. Why use AI workflow automation? AI is poised to help marketers automate significant chunks of their workflows. And marketers are starting to embrace it: Our research shows that 64% of marketers use AI in some form at work. Yet, only 21% have integrated AI extensively into their workflows. There’s room for growth. So, why do I think you should join the top 21% of AI-powered marketers? Saved Time Our research recently found that marketers using AI save an average of 12.5 hours per week. That’s nearly 26 working days per year. What could your team do with an extra month? Timesaving benefits vary depending on where you incorporate AI into your processes. For instance, using AI to automate parts of content production can save you time drafting and editing — often the most time-consuming pieces of the process. I’ll discuss setting goals and objectives of AI workflow automation later, but know that it’s important to identify time-consuming tasks before you integrate AI significantly into your process. Effective Data-Driven Decisions Data is drowning marketers. Amid oceans of information, what data matters to your team? And how can you use it to plan your next moves? I think a key to conducting effective marketing is to find which data is most relevant to your needs and understand how to deploy it. AI is especially attuned to ingest data from the many sources in your organization’s workflow, analyze it for patterns, and deliver actionable insights to make things happen. Targeting and customer experience personalization are ripe fields for AI disruption. Imagine AI handling user data collection, process, and insight generation. You receive a list of “here’s what to do next” and can develop and execute campaigns to match. Real-world example: Yum Brands (which owns KFC and Pizza Hut, among other fast-food stops) is seeing double-digit increases in consumer engagement and more purchases with AI-driven marketing decisions. Scalability Demands on marketers’ time and energy are soaring — I know I feel it, and I’m sure you do too. Recent surveys found that over 60% of marketers feel overwhelmed by what their jobs ask of them. Marketers could always use an extra pair of hands — and that’s now started to include AI hands. Marketers can use AI to automate repetitive or predictable tasks like data collection and analysis or social media post drafting. Or, AI can use all that data it analyzes to automatically personalize outbound messages, helping you reach people more effectively without manual effort. AI doesn’t replace the human marketer (I’ll add more on that later), but it does help us do more — even if your marketing budget is stagnating. What You Need to Start AI Workflow Automations So, where do you begin? I recommend you give the following points some thought as you plan your AI workflow automation. Throughout my exploration of AI workflow automation, you’ll hear from fractional CMO Tim Hickle. He’s invested significantly in understanding AI’s role in marketing for companies of many sizes and structures, and he shared a lot of great insight with me. Team Needs and Pain Points What’s actually bogging down your team? What tasks frustrate them or suck away their time from the higher-level, strategic work you need done? Answering those questions takes quantitative and qualitative data. Sit down and ask your team what they feel eats at their time. Their responses won’t be exactly the same, but you’re likely to see patterns emerge. From those patterns, you can define the workflow steps ready for automation. For instance, when we first explored AI integration, I sat with my content team to hear about their struggles. Each had their own interest, desire, and need for AI, but items like repurposing our long-form work for social media distribution felt time-consuming and frustrating. We could then form a hypothesis that AI integration could help us automate that process. Pro tip: It’s also good to give yourself a few numbers to help — especially if you need executive buy-in. Have your people time-block their calendars or track using a specific hours-tracking tool (project management tools like Monday have these features built-in). See if their feelings

Choosing the right large language model can feel overwhelming with so many options out there, especially if you’re not exactly living and breathing AI But as we’ve worked through each one, we’ve gotten a real sense of what they’re good at (and where they fall short). So, let’s talk about what to use, when. ChatGPT & OpenAI-o1: The Reliable All-Rounders Let’s start with ChatGPT and OpenAI-o1. OpenAI’s latest model is impressive, and people are hyped about its “reasoning” abilities — basically, it’s designed to tackle more logic-heavy stuff alongside the creative tasks that ChatGPT has always been great at. Why We Like It Big on Logic: OpenAI-o1 uses something called chain-of-thought reasoning. In simpler terms, it’s better at walking through complex problems step by step. Custom GPTs: This feature lets us create models that remember instructions specific to our work. If we need it to think like a project manager or a social media assistant, we can set that up with just a few clicks. Where It Falls Short Overkill for Basic Stuff: Most of the time, GPT-4 can get the job done. OpenAI-o1 shines with complex tasks, but you might not notice a huge difference for more straightforward use cases. Not a Quantum Leap: The big improvements are behind the scenes. If you’re expecting to see massive changes in day-to-day use, you might be underwhelmed. When to Use It: Anything involving more complex logic, or when you need tailored responses, like for coding or detailed content editing. Claude by Anthropic: The Summarizer & Storytelling Champ Claude is our go-to for summarizing and making sense of long documents. It’s also fantastic at storytelling, which is helpful if you’re in content creation or need to simplify dense information. What Makes It Stand Out Document Summarization: Claude is amazing at boiling down information, so it’s perfect when we’ve got huge documents m and need a quick summary. User-Friendly Customization: Anthropic’s Projects feature lets us set up custom instructions for repeat tasks. It feels more intuitive than ChatGPT’s setup. What to Watch Out For File Size Limits: If you upload a big file (over 20 MB), Claude sometimes throws a fit. We usually compress PDFs to work around this, but it’s worth knowing. Best Use Case: Summarizing or creating content when you need a straightforward, reliable tool that’s easy to navigate. Google Gemini: The King of Context (and Podcasting) Google’s Gemini feels like it’s in a league of its own when it comes to handling tons of data. We love that it has a massive context window, meaning it can hold and process entire books if needed. Plus, it has a quirky new tool called Notebook LM that turns docs into a mini-podcast for you. Why It’s Cool Handles Huge Data Loads: With a 10-million-word limit, Gemini can keep track of massive documents all at once, so we can load entire libraries if we need to. Notebook LM: This feature actually turns documents into audio summaries in a conversational podcast format. It’s a great way to get the gist of something while multitasking. Drawbacks Limited Customization: While it has “Gems” (Google’s answer to custom GPTs), they’re pretty basic. You can’t connect it to other tools or APIs like you can with ChatGPT or Claude. When to Turn to Gemini: When you need to process a mountain of data at once, or if you’re in the mood for an audio summary while I’m doing something else. Llama by Meta: Privacy & Flexibility Llama isn’t necessarily the most advanced, but because it’s open-source, it’s our go-to when privacy is a concern. Unlike the others, Llama can run offline on your computer, so it doesn’t share data with a big tech company. Why I’d Recommend It Keeps Things Private: Since Llama runs locally, we can be sure our data stays off the internet. Highly Customizable: Llama’s open-source, meaning we (or any developer) can modify it for unique needs. We don’t do this much, but it’s nice to know it’s an option. Weak Spots Not the Most Powerful: It’s not as good as Claude or ChatGPT for high-quality content or problem-solving. But for basic use cases, it’s solid. When It Makes Sense to Use: Anytime privacy is key, like with sensitive internal data, or when you just need a quick local solution. Grok by xAI: Twitter Data & Realistic Image Generation Grok is a fun one — it’s a social media native, integrated with X (formerly Twitter). It’s a decent model and comes with a strong image generator, Flux One, that can make super-realistic visuals. But where it really shines is pulling in Twitter data in real-time. Why We Use It Live Twitter Insights: Grok lets us see what’s trending or analyze popular Twitter profiles on the spot. Image Generation: Flux One can create realistic images of people, scenes, and more, with few limits on topics. Downsides Niche Use Cases: It’s great for Twitter data and images but doesn’t stand out in general tasks like summarization or storytelling. Ideal Use: Social media research and generating realistic visuals for content. Perplexity: A Researcher’s Best Friend Perplexity isn’t technically an LLM in the traditional sense. Instead, it’s an AI-powered research tool that pulls information from the internet and then uses a model to organize it. It’s our go-to when I need quick, accurate information or a second opinion on a topic. What Makes It Indispensable Web Search Capabilities: Perplexity searches the web and summarizes content, making it perfect for research-heavy tasks. Choose Your Model: we can use GPT-4, Claude, or even OpenAI-o1 as our “engine” within Perplexity, so we always get the model that fits our needs. Caveats Double-Check for Accuracy: Sometimes it mixes up similar names or pulls outdated info, so it’s good to cross-check important facts. When I Use Perplexity: Anytime I’m in “research mode” or need up-to-date insights for blog posts, presentations, or meetings. Finding the right LLM can be as simple as matching a tool’s strengths to your needs. Our advice? Try out a

AI content SEO” is one of those terms we’ve all Googled at some point since AI became part of our workflow. No doubt it’s great for marketers — it makes us faster — but does it hurt our rankings? To what degree? We all want to know, but there’s no simple “yes” or “no” answer. However, I’ll joy you with real-life examples of AI content wreaking havoc and when it goes unnoticed by search engine algorithms. Let’s explore everything you need to know about AI-generated content — including how it performs in search results, its limitations, and tips for leveraging it. Table of Contents Can AI-generated content hurt my search ranking? How does AI affect content performance? The Limitations of AI-Generated Content for SEO 5 Tips for Using AI-Generated Content Can AI-generated content hurt my search ranking? At this time, Google has made it clear that AI-generated content will not impact search rankings. As long as your content is helpful, original, and relevant, you have the green light. In other words, Google is less concerned about how you produce content and more concerned with the quality of the content itself. What are the real results, though? We did some research on this, and 46% of respondents say AI has helped their pages rank higher. On the flip side, 36% feel AI hasn’t made a difference, and 10% have seen a drop in rankings. The first category of satisfied respondents employs the E-E-A-T framework because Google loves content that’s: Helpful Demonstrates expertise Published on an authoritative site Trustworthy But here’s the rub: AI-generated content may not check all those boxes. As Josh Blyskal, associate marketing technical manager at HubSpot, aptly points out, “Now, more than ever, the value of content hinges on the authenticity of its creator and the underlying value, meaning, story, and perspective of the content they’re creating.” As the internet becomes flooded with AI-written content, the real hurdle is standing out from the masses. AI’s Influence on User Engagement Do users actually engage with AI-generated content? That’s another million-dollar question. While some are skeptical, others are experimenting to see if it really moves the needle. I found one interesting case study on this matter. SEOwind ran an experiment by publishing 116 AI-generated articles in just 30 days. That’s what the study proclaimed. Using their CyborgMethod strategy, they tracked the results and saw impressive outcomes: a 77% increase in clicks and a 124% boost in impressions. Source The goal was to see how AI content performs and prove its ability to drive organic traffic. Throughout the experiment, SEOwind focused on topics like SEO, blogging, and AI tools, using AI to create and optimize high-quality content. But don’t be fooled by the numbers. The study was conducted in 2023. We don’t know how those pieces of content perform to this day. But more importantly, the team carried out a good deal of work besides spitting out AI-generated content. They did: Keyword research and content gap analysis Let AI create titles, meta descriptions, and outlines Optimized for secondary keywords Added quotes, numbers, external and internal links, descriptive alt tags to images, and product descriptions Trimmed long sentences (aka edited AI content) So it doesn’t sound like AI-generated content alone, does it? Let’s move on. How does AI affect content performance? The impact of generative AI on content performance is mixed. 34% say AI boosts performance, 19% see no change, and 6% think it hurts performance. Moreover, 29% believe AI doesn’t improve ROI but does speed up content creation. Speaking of content creation, it mainly helps with specific parts of it — e.g., brainstorming and certain aspects like coming up with headlines (every writer knows how often we hit a block there). For instance, a Danish news outlet, TV 2 Fyn, conducted A/B tests to improve CTR using ChatGPT to generate headlines. Over three weeks in late 2022 and early 2023, they ran 46 A/B tests. AI-generated headlines won 46% of the tests, while human-created headlines won 24%. Source The AI-driven headlines led to a 59% increase in CTR, outperforming traditional headline strategies. The results showed that while AI improved performance, refining its suggestions was key, with human input still crucial for optimization. I heard a similar experience from Edward White, head of growth at Beehiiv. He said: “A unique trick we’ve implemented is using AI for dynamic A/B testing of blog headlines and meta descriptions. This iterative process has helped us improve click-through rates consistently,” White says. The impact of generative AI on content performance depends on factors like the content type, the input you provide, the overall quality of the AI content generator, and how well it aligns with audience needs. And when it comes to content types, some formats will thrive while others might struggle. So, which ones will crush it, and which will fall flat? Let’s see. Content types that will crush it. HubSpot study shows that some content types are perfectly positioned to thrive with AI. 45% of people believe educational content — like “How to” guides and step-by-step tutorials — will perform the best, while 37% think review and comparison content will also do well. So, if you’re creating this kind of content, you’re probably set to see some serious payoffs. Content types that might stumble. On the flip side, personal stories (30%) and opinion pieces (28%) are likely to face the biggest challenges. These formats depend heavily on a unique voice and personal touch, which can be tough to keep fresh and relevant with AI. The takeaway? Content that offers clear, real value is more likely to succeed, while more subjective, personal content will have a harder time keeping up. Almost impossible to generate with AI, to be honest. Curious if I used AI to generate this piece? An outright “No.” Just a few times for suggesting alternative wording. The Limitations of AI-Generated Content for SEO While AI can speed up the content creation process — which certainly has SEO benefits — it’s not

Think you’re good at spotting AI-generated content? Put it to the test. Between Midjourney, Lummi and the host of other AI image generators, it’s getting harder to detect AI imagery.  Take this quiz and find out how good you are at finding AI in everyday images, videos, and illustrations.