AI Digital Marketing Insights - What's Working And What's Missing

AI has really changed things in the last couple of years. Kieran Flanagan, who is the SVP of Marketing at HubSpot and also hosts the Marketing Against the Grain podcast, recently talked about how AI is impacting marketing. He shared what's been working, what's still missing, and how businesses might be losing money by not using AI effectively.

The Shifting Landscape of Digital Marketing

Digital marketing has always relied on the open web, where creating valuable content could help you rank high on Google and bring traffic to your website. Google, in fact, accounts for about 64% of all web traffic. But things are changing. AI is disrupting this, especially in search. AI tools like Perplexity and ChatGPT offer direct, concise answers, making the traditional "blue link" search experience less appealing. This shift is leading to what some call a "closed web," similar to what happened with social media.

Social media used to be great for promoting content and products. However, over the last five years, it has become more of a closed system. If you try to promote your own stuff too much, you often get penalized with fewer impressions. For example, putting a link directly in a LinkedIn post can result in six times fewer impressions. This is why people often say "DM me for the link" or put links in the comments. We're starting to see a similar trend in search with AI Overviews, where Google provides answers directly, meaning users don't even need to visit your website.

Another big change is the stagnation of new marketing channels. While there are 30% more B2B startups now than five years ago, there haven't been many new ways to reach customers. This means more competition for ad spend and less leverage for optimizing ads, making budget the main competitive factor. Plus, there are just more vendors in every market, increasing competition across the board.

AI's Impact on Expertise and Competition

AI is also changing how quickly people can become good at something. In the past, having deep expertise in digital marketing gave you a big advantage. Now, AI can help people learn and become very skilled in a short amount of time. This means more competition in all channels. The old ways of doing things might not work anymore, and we need a new approach.

Key Takeaways

• AI is changing how people search and find information, moving towards a "closed web" experience.

• Traditional marketing channels are becoming more saturated, with increased competition and fewer new avenues.

• AI accelerates learning, increasing competition by making it easier for more people to gain expertise quickly.

Adapting Content Strategy for the AI Era

We used to publish a lot of informational content to rank in search and grow businesses. Now, we need to make sure our content is "AI-proof." This means using large language models to get 10x value where we can still stand out. We've categorized our content into risk levels based on AI's impact:

• High Risk: Content that answers simple informational questions that AI can easily answer better than a blog post. If people start using AI assistants for all their search content, this category is most at risk.

• Moderate Risk: More complex questions that require deeper thought and research. There's still a chance to stand out here by offering unique insights that AI overviews or ChatGPT can't provide.

• AI-Proof: Transactional keywords, where people are searching for a product, service, or action. These are still working well, which is why premium services continue to perform strongly.

From what we've seen, AI is already causing a 10-20% reduction in traffic for many content engines. This is happening even before new AI search engines like ChatGPT's gain widespread adoption. AI assistants often provide a better, faster, and easier user experience, which users tend to prefer.

Integrating AI into the Content Process

We're integrating AI into every part of our content process to find areas where we can differentiate. Traditionally, content teams handled research, drafting, editing, and performance analysis. Now, AI can do a lot of the heavy lifting:

• Research: AI can do extensive research, even integrating with your existing tools.

• First Drafts: AI can build initial drafts using style guides you provide.

• Editing: This is where human nuance and competitive advantage come in. Humans can refine and add unique perspectives.

• Performance Analysis: AI can optimize content by running countless tests and learning what works best, similar to how AI-driven ad tools create and test endless creative variations.

This means that if AI can optimize and speed up many steps, you can spend more time on making your content unique. For example, Lenny Rachitsky, a newsletter writer, uses his network to gather unique data points, examples, and case studies that AI can't access. This "example-based content" is a great way to stand out. Instead of just creating educational posts for keywords, businesses need to develop a content strategy that focuses on differentiation and making their content 10x better than competitors.

AI as a Brand Research Tool

Many people wonder how to optimize content for large language models to appear in AI search results. While there's no simple trick, AI can act as a brand research tool. Tools like HubSpot's AI Search Grader can analyze how the internet perceives your brand and services. It shows sentiment analysis and brand awareness within the AI's training data. Since LLMs are trained on the internet, more positive mentions of your brand for specific products or services can increase your visibility in AI search results. This tool can also highlight reasons for negative sentiment, helping you address issues proactively.

The Evolution of Social Media and Marketing Measurement

Social media has changed. The old way of promoting everything on your posts doesn't work well anymore. Instead, you need to build a real audience and act more like a creator than a brand. Channels that are still growing, like YouTube and podcasts, are often more individual-centric. This forces brands to think about how to engage audiences more personally.

Even with AI optimizing many processes, human connection remains important in the buying process. People trust other people more than brands. However, the future of marketing might become less measurable. HubSpot's acquisition of The Hustle, a media company with a popular newsletter and podcast, shows a shift towards a "media-first" approach. The idea is to build real media assets across various platforms to influence buyers before they even need your product. This long-term strategy, though hard to quantify for a CFO, is about building an audience and influence, which AI can help with.

Practical AI Applications and Personal Experiments

AI is not just a copy-and-paste tool; it's an accelerated learning and iteration tool. Here are some practical examples:

• Building Style Guides: AI can reverse-engineer what makes good content. By feeding it examples of successful content, AI can create style guides. For instance, you can give Claude examples of effective outbound emails and ask it to create a style guide for outreach emails. This allows you to quickly create content in different styles and see what performs best.

• AI Agents for Content Creation: You can train AI agents with your domain expertise and writing style. For LinkedIn, there are three types of posts that perform well: educational posts (single, clear takeaway), spicy takes (counter-intuitive ideas that spark debate), and articulate positioning (expressing widely held beliefs in a new way). By combining style guides with these content templates, you can significantly increase impressions.

• Templatizing Knowledge for Video: AI can templatize knowledge from long-form content into short-form videos. By analyzing successful short-form video creators, AI can learn the structure and psychological triggers that make videos engaging. This allows you to upload long-form content and have AI extract talking points to create effective short-form videos.

AI in Personalization and Sales

AI is a huge boost for personalization in B2B marketing. Marketers are likely to own more of the customer journey, from acquiring contacts to closing deals and even some customer support functions, because AI can automate many of these tasks.

• Email Personalization: AI-generated emails have shown impressive results, with conversion rates increasing by 80-100%. To get the best results, separate your prompts for subject lines, body copy, and calls to action. You can also use AI to analyze your best-performing emails and create templates. It takes several iterations to see significant results with AI in email.

• Automated Chat: Over 80% of HubSpot's chat interactions are now automated through AI. While AI is well-known for customer support, it's also showing promise in sales conversations, converting prospects into customers or upgrading existing ones. The quality of AI's performance depends heavily on clear onboarding documentation and effective annotation (feedback) from skilled human experts.

• Business Development (BDR): AI can significantly boost prospecting efforts, especially for inbound leads. It can generate thousands of incremental meetings per month. However, for mid-market and corporate segments, email alone is often not enough due to AI saturation. A multi-channel approach, including calls and social outreach, is needed. Unique data is crucial for crafting personalized messages that stand out. Integrating AI into BDR teams and having them provide feedback is essential for continuous improvement.

Voice AI is also emerging, with some companies using it for lightweight business calls, particularly in B2C. While legal complexities exist, voice integration is expected to grow in the coming year.

The Future of AI in Go-to-Market

We are in the early stages of AI in go-to-market strategies. It's like the "Ask Jeeves era" for AI, with much more advanced tools yet to come. Experimenting early is critical because saturation will happen quickly. First-mover advantage is more important than ever. For example, HubSpot is piloting an AI avatar for product walkthroughs. While initial reactions might be mixed, the current version is the worst it will ever be, and early feedback is valuable for improvement.

AI's ability to take unstructured data and turn it into valuable insights is a game-changer. For instance, by analyzing unstructured data from customer interactions, AI can infer reasons for customer churn, even when customers don't explicitly state them. This allows businesses to proactively address problems and improve customer retention.

When training AI models, while open-source options are emerging for specific use cases, starting with established large language models like ChatGPT and Claude is often a good approach. The key is to continuously experiment and integrate AI into your teams, allowing them to provide feedback and fine-tune the models for better results.