Essential KPIs to measure generative AI’s success in marketing

No marketing technology series would be complete without discussing the ever-important measurement of success or KPIs. I have been writing about KPIs and ROI my entire career, with the topic becoming especially hot and heavy during the dawn of social media. I recall discussing the topic at a digital marketing dinner about 15 years ago, where after several glasses of wine, the bold consensus was to forget direct ROI and instead find inherent value in the new relationships we could garner on social. 

Today, you would get no argument on the importance of building relationships with influencers across social platforms, as social media marketing ROI and KPIs are now well-established within strategic frameworks. I wish it were that simple when it comes to AI. 

New tech always breeds resistance

There are similarities between the birth of social media marketing and the latest advances in generative AI. In both cases, brand-new technology caught on like wildfire, affording marketers the opportunity to reach audiences in new ways. 

As with anything new, some were skeptical of its long-term value. During the social era’s boom, these clients insisted on seeing a direct sale or a direct conversion to find value in social media campaigns. It was my job to gently educate these skeptics on why their target audience was on the social platform in the first place and see the opportunity to meet their needs in a new way — not just pushing the product. 

The dawn of social media drove a need for valuable content, and content marketing has been front and center ever since, along with standard content consumption and engagement KPIs to help judge its effectiveness. 

AI ROI solution: ‘It’s complicated’

AI and predictive analytics, on the other hand, are quietly playing a leading role in every facet of marketing, not just outreach and external communications, for decades. Recent advancements in generative AI and tools like ChatGPT have marketers asking how to prove ROI and measure success. The simple answer is that it depends. 

Are you getting the most from your stack? Take our brief 2024 MarTech Replacement Survey

If you followed this entire series of articles, you are aware that generative AI can be used across all areas of marketing and impacts every task we do daily. This is where AI dramatically differs from social media and why it becomes much more complicated to determine success metrics. 

Success measurements are directly tied to the job the AI machine needs to do and can fall into several categories, such as productivity KPIs, cost-saving KPIs, audience engagement and campaign KPIs and output and creative quality KPIs. 

Start with strategy

The golden rule for establishing AI success metrics is to start with developing a clear strategy. Clearly and concisely define the project objectives, the roles and/or users impacted by the objective and finally, a set of KPIs that will help you determine if the project was successful. In a previous article in this series, I provided more information on building an AI marketing strategy and a comprehensive list of marketing tasks that can benefit from AI. 

Let’s roll up our sleeves and look at a few real-world examples, shall we? 

AI for personalization and engagement: Intel’s Art of AI

Intel approached me looking for new ways to help engage attendees at an upcoming global conference. They wanted to reach both in-person and online conference attendees and show them the potential of new generative AI technologies powered by Intel.

In this case, AI needed to create a “surprise and delight” moment to capture the attention of conference attendees, leaving them inspired and enthusiastic about how AI could impact their industries. 

After pitching several creative concepts, the team decided to move forward with an AI art exhibit that demonstrated the power of AI through interactive AI art installations. 

Here is the short list of KPIs for the project:

• Number of conference attendees, both in-person and online, who experienced the exhibit.
• Number of attendees who shared their personalized co-created artwork on social media.
• Number of in-person attendee badge scans.
• Number of website visits on the exhibit microsite.
• Number of brand and experience mentions across social and digital channels.
• Share of voice during the conference for terms related to AI as compared to competitors.
• Analysis of feedback and comments capturing trends and overall impressions of the experience.

In this case, generative AI allowed for personalization and interactivity in the form of a co-created, one-of-a-kind AI piece of art that was offered as an experience takeaway. The three AI art exhibits demonstrated a different aspect of generative AI while offering a completely new and engaging experience. 

AI for productivity: PIA, my AI client-coach and tutor

This second example is all about extending and scaling my expertise, allowing clients and students to access my knowledge 24/7 without continuous monitoring of my inbox. In this case, AI needs to help save my time by automating the process of answering repetitive questions and my clients’ time by providing them with quick and easy information anytime and anywhere. 

The solution was a privately trained AI chatbot named PIA which is accessed via a web browser on any smartphone, tablet or laptop. PIA was trained on this entire series of MarTech articles along with transcripts from my AI Marketing Revolution Challenge video tutorials, content from my AI strategies courses, blog, website and entire work history. 

This tool has recently launched, so stay tuned for a full case study outlining PIA’s performance in the coming months. In the meantime, here are the KPIs for this project:

• Number of questions successfully answered.
• Number of client engagements with PIA.
• Number of hours saved weekly (i.e., time that I can spend on more meaningful client work).
• Sentiment analysis of conversations with PIA.
• Post-PIA survey results determining client satisfaction with the tool.

Explore, experiment and engage

These are two vastly different examples of aligning AI KPIs with a specific objective, and hopefully, they will help you understand my strategic approach. I recommend launching your AI projects with a sense of exploration and experimentation. We are very much in the preliminary stages of these powerful technologies, all of us pioneers navigating unfamiliar terrain.