How Performance Marketers Should Talk to AI

Artificial Intelligence has many charms. Its most powerful one is appearing sentient (AI isn’t conscious, of course). There are many examples of AI demonstrating a lack of a soul and/or common sense, but often it hides it well by delivering cogent answers to questions and problems, rather than pages and pages of links.

It’s no crime being a computer, but it does mean AI’s brain is still binary. And in a binary world, the structure of your queries shapes the answers you receive. If you ask broad questions, you get broad answers that read like high school book reports. To make AI truly relevant and helpful, you need to speak its language.

The Ground Rules for Talking to AI

1. Use clear, specific questions that define what you want.

• Good Example: “Review this script for a short video and suggest ways it can be changed to help improve customer acquisition for a target TikTok audience of women aged 18-34 living in the Northeast.”
• Bad Example: “What are ways this script can be better.”

2. If possible, break down your query into single or simplified tasks.

• Good example: Rewrite this copy without going over 30 words and try to make it more appealing to Gen Z. 
• Bad example: What are some ways to get better engagement from Gen Z?

3. If you can’t simplify, give your complex question some structure.

• Good example: Write a history of tennis broken out into sections, each no more than 200 words: 1) Rules. 2) Gameplay. 3) Top five greatest female players. 4) Top five greatest male players. 5) Tennis’ most memorable matches. 6) When tennis and politics crossed paths.
• Bad example: Write 800 words about the history of professional tennis.

If the AI answer seems a little fishy to you and you’d like to know more about how it arrived at the conclusion, just ask, “Explain your thought process step-by-step and defend your answer.” This way you’ll know whether the AI is going in the desired direction.

Pro Tip: Try different AI chatbots. If you are only using ChatGPT, there are plenty of other companies with competing AI. Find the one that works for you. For this article, we tested a lot with Perplexity.ai, created by some Google defectors. We like it because it provides links to its sources, making it convenient to click through for more information that can help round out your findings.  

Use Cases for AI

AI can fix your code and formulas

The written word is AI’s second language – its native tongue is code. Lean into this and save countless hours scrubbing the web or waiting for answers from the Reddit community by just asking AI to fix your code. Got an excel formula that isn’t working? Looking for more efficient ways to write a string of code? Can’t figure out why your ad is appearing all wonky? AI has an eagle eye for fixing tiny coding mistakes.

Example: We deleted a colon in this WordPress code and asked Perplexity to find the problem. Here is the answer: 

There is more to the answer but you can see that the AI identified the problem and even wrote the code correctly. From here, just copy and paste the proper code and you are back in business. 

AI can also make code more efficient and explain its reasoning. Below, a functional but somewhat inefficient example of Python code was presented along with a request to write a more efficient version. 

AI makes big data more manageable

1. While researching, some bots (like Perplexity) can narrow the search to more reliable sources by using the “Academic” search scope. This will focus the search on credible research sources and industry reports.

Example: “Use the academic search scope to find data about microwave oven purchase trends in North America since 2020.”

2. As a follow up, ask the AI for charts supporting the findings. Some won’t actually make the chart for you but it will share the relevant images it finds. Even better, you can ask the AI to create a table. From there, you can paste the table into a slide program which will easily convert it to a chart. You can also ask the AI to convert the data to a CSV if that format works better for you.

Example: Provide a table showing the projected growth rate of the microwave oven market in North America from 2020 to 2030:

3. Artificial Intelligence can apply statistical and analytical techniques that will help identify trends, patterns, and anomalies within your data for some valuable insights into the underlying factors and relationships. Input your data in one of a few ways:

• Upload your Excel, CSV, PDF, txt, or JavaScript JSON file (this may require an API key)
• Paste a link to a spreadsheet stored in the cloud
• Paste the whole CSV text into the chat.

4. Ask general questions like “What trends do you see in this data?” and then drill down, “why did the test group underperform?” “What opportunities are there to scale?”

5. AI can also help contextualize your data by considering relevant external factors, such as industry trends, market conditions, or historical events that may be impacting the results. Just ask it and provide any additional relevant data the program thinks could be missing.

AI will watch videos for you

Humans may be very good at watching videos, but AI is much faster. If you are short on time, AI can summarize, suggest similar content, and extrapolate highlights from long videos in a matter of seconds. A lot of the information comes from the YouTube description, but it also pulls from sources like Reddit, Wikipedia, etc. to add context, highlights, and perspective. 

Example Query: Can you summarize what happens in this video?

Note: This example posed a particular challenge for the AI since there was no dialogue – only music and images. 

AI can help you optimize campaigns

AI can give sage advice before, during, and after a campaign. Just input the data, ask specific questions, and see what it says. You might be surprised. 

• Audience Insights:
  • “What are some common interests and demographics of our target audience?”
  • “Can you suggest effective messaging strategies to resonate with our audience?”
• Content Creation:
  • “What are some trending topics or keywords in our industry?”
  • “Can you generate compelling ad copy or social media posts for our campaign?”
• Competitive Analysis:
  • “What strategies are our competitors using that we should be aware of?”
  • “Can you identify gaps in our competitor’s offerings that we could exploit?”
• Optimization:
  • “What are some ways we can improve our ad performance or website conversion rates?”
  • “Can you suggest A/B testing ideas to optimize our marketing campaigns?”
• Predictive Analytics:
  • “Based on past performance data, what can we expect for future campaign results?”
  • “Can you forecast the impact of changes in our marketing strategy on key performance metrics?”
• Customer Engagement:
  • “How can we improve customer engagement and loyalty through our marketing efforts?”
  • “Can you suggest personalized messaging strategies to enhance customer relationships?”
• Emerging Trends:
  • “What emerging trends or technologies should we be incorporating into our marketing strategy?”

AI is the best brainstorming partner you’ve ever had

Working from home and no one to bounce ideas off of? AI is a sharp brainstorming buddy and can even do a lot of the work that brings the ideas to life. 

Here’s an example scenario to help you get started:

• Step 1: “Brainstorm some ways to position retirement savings marketing campaigns in new fun ways.” 
• Step 2: “Create three video treatments based on those ideas.” 
• Step 3: “Write a script for the second treatment idea.”
• Step 4: “Bring the script to your boss and say you did it.” 

AI can fact-check itself

Not sure about the accuracy of the results? AI can fix its own mistakes. In this example, we asked it to tell us about the top five greatest male tennis players, but it wrote this about Novak Djokovich: 

“Novak Djokovic – Serbian player who has won 22 Grand Slam singles titles and is renowned for his exceptional all-around game and mental toughness.”

This didn’t look right, so we asked it to review that particular sentence:

“Novak Djokovic is a Serbian professional tennis player who has won an all-time record 24 Grand Slam singles titles, not 22.”

This mistake highlights AI’s inherent flaws as an internet scraper extraordinaire. Because Roger Federer and Djokovich were tied at 22 Grand Slam championships for some time, a great deal was written on the subject. A likely reason for the mistake is that the AI initially found more content referring to that period than the time since Djokovich added two more titles. 

While it’s encouraging that AI was able to definitively reach the right answer, it needed to be prompted. The takeaway here is obvious: be skeptical. Use your organic intelligence to keep the artificial intelligence in check.

Big Picture: Using AI to Optimize Campaign Performance

By using specific, conversational queries and taking advantage of AI’s advanced search and synthesis capabilities, you can unlock valuable insights to improve the performance of your marketing campaigns. Think of AI as an interactive research assistant to guide you through the analysis process. And don’t expect it to read your mind. Err on the side of over-explaining because AI is never going to “know what you mean.”

Learn how Fuent is leveraging AI to help advertisers segment, target, and personalize their performance marketing campaigns here.