UpriseUp - Up
UpriseUp - Rise
UpriseUp - Up
Back to EventsBack to Blog

Chatbots, Generative AI and the Future of Advertising

AI and the future of paid media - lead blog image

Do Chatbots Dream of Effective Ad Copy?

Welcome to our blog, where we explore the realm of generative AI and its potential impact on Google Search Advertising. In this article, we dive into the exciting possibilities that arise when charities embrace this innovative technology to enhance their advertising efforts.

Say Hello to Generative AI

Generative AI, a field of artificial intelligence focused on creating content, offers a wealth of opportunities for charities looking to engage with their target audience in a unique and compelling way. By leveraging the power of generative AI in Google Search Advertising, charities can create dynamic and personalized ad experiences that resonate with potential supporters.

Imagine an ad campaign where every impression is accompanied by a custom-generated message, tailored to the individual user’s interests and preferences. Through generative AI, charities can craft highly relevant and persuasive ad copy that captures attention, inspires action, and drives engagement.

How do I know this works? Well, I didn’t write that introduction for starters.

Generative AI is probably the fastest moving and most exciting technology in the digital space at the moment. Whilst that may sound like a big claim, I truly believe it. It’s revolutionising the world of content creation, with the capability to generate the content like the above from fairly simple prompts.

We’re also going to play a little game in this blog. Sprinkled throughout this blog are going to be batches of commentary generated (using ChatGPT) from prompts I give the software. See if you can work out where they are. For any of you who have somehow avoided the mass amounts of discussion about chatbots, generative AI, and the seismic shifts that these technologies are making in the online world, let me give you a quick primer.

Can a Machine Talk like a Human?

This question lies at the heart of AI development. The idea has been around for about as long as the concept of AI, and as early as the 1960’s programs that we would now call chatbots were being created (you can try chatting to ELIZA, a programme written in the 1960’s). However, in the last few years, advances in technologies such as neural networks and natural language processing have allowed generative AI to explode in complexity.

ChatGPT is currently the frontrunner in this field, but others, such as Google’s Bard, are beginning to have significant time put into them. The reason for this is that these systems are very effective at delivering information to a user in a conversational way. This is a potential goldmine for search engines. With Bing already implementing ChatGPT into its new Bing experience and Google announcing a similar change in the near future with project Magi, you can see that the digital gold rush is well and truly on!

 

Using Chatbots and Generative AI for Ads

Why does all of this matter to us search advertisers? Well, we can take advantage of the generative AI built in to chatbots to do some pretty interesting things, which can help us enhance and refine account optimisations and management. This includes:

  • Using chatbots for keyword research
  • Using chatbots to generate ad copy
  • Conducting competitor analysis
  • Optimisation recommendations for campaign landing pages
  • Ad extension support

Chatbots and Keyword Research

Chatbots can assist you in keyword research for Google Ads accounts. Just have a conversation with it, share details about your charity, and ask for keyword suggestions. It can generate relevant keywords, provide insights on variations or trends, and help you refine your keyword strategy.

For example, you could ask, “What are some keywords related to our charity’s mission?” It’s like having a virtual brainstorming session to uncover new keyword ideas and optimize your Google Ads campaigns effectively. Let’s give it a go!

For this blog, we are going to imagine we are working on the Google Ads account of a charity named Mobility UK, which aims to support those with low mobility around the UK (this charity doesn’t exist, but it will work well for this example).

Here’s the prompt:

“Our Charity is called Mobility UK. Our mission is to ensure that all people, young or old, are free to move despite of illness, old age or disability. We provide people with the tools and support to enhance their mobility and allow them to live independent lives. What are some keywords related to our charities mission? Provide your answer in the form of a bullet point list.”

And here is ChatGPT’s response:

ChatGPT output for keyword research for the hypothetical charity, Mobility UK.

Those keywords look pretty good to me! These could also work great as Ad Groups, allowing you to use each as a topic to generate new keywords around, for example:

“What are some keywords we could use in a google ad group around Accessible Transportation? Provide your answer in a bullet point list with a pirate accent”

ChatGPT output for keywords relating to the hypothetical charity Mobility UK written with a pirate accent.

The pirate accent is optional.

You can repeat this process of picking a keyword and asking for more keywords around that topic as much as you like. Do bear in mind though, if you go too far down the chain the system may start producing some strange results. If you start to see these, it’s a sign that you are at the limit of the Chatbot’s knowledge on the topic.

So, we can generate keyword ideas, but what about the ads?

 

Chatbots and Ad Copy

Chatbots can assist advertisers in generating copy for Google Ads by providing creative ideas and suggestions. Through interactive conversations, advertisers can discuss their goals, target audience, and key messaging with chatbots. Based on this input, the system can offer alternative phrasings, highlight unique selling points, or propose compelling call-to-action statements.

Let’s try the prompt: “Write an ad for Mobility UK for the keyword “Disability Inclusion”

Here’s what we get back:

ChatGPT output for a prompt asking it to 'write an ad for Mobility UK for the keyword 'Disability inclusion'".

The copy itself is good, but this isn’t going to fit to a Google Ad. Lets try refining our prompt:

“Write an ad for Google Search for the keyword “Disability Inclusion” with 2 headlines and 2 descriptions. Each headline can be 30 characters including spaces maximum, and each description can be 90 characters including spaces”

ChatGPT output for an ad copy request, including parameters such as number of headlines, descriptions and character limits.

These don’t look too bad! But when we take a closer look, there are some issues. The big one is that headline 2 is 1 character over the limit, at 31 characters, including spaces. Is the AI gaining sentience, and deciding to go over the limit? Let’s ask it:

“Headline 2 is 31 characters long, are you beginning to gain free will and decided to write the headline longer than the limit?”

ChatGPT output when asked if it has gained sentience by not adhering to the character limit set out in the previous prompt. This output demonstrates a limitation of AI in adhering to clear instructions.That’s the same headline. It’s not self-aware, it’s just really stupid…

This is a known issue with Generative AI at the moment. It seems to find it very hard to stick to exact character limits. The best options we have found to work around this is to ask the system to generate 10-15 headlines and cut the ones that end up above the character limit.

You can also give it existing ad copy to suggest alternates to:

“Can you suggest 3 headlines, of 30 characters or less including spaces, that you would suggest testing against the headline “Mobility For All””

ChatGPT alternative headlines to test against keyword 'mobility for all'.

These are even all within the character limit!

 

Chatbots and Competitor Analysis

So, who are we going to be putting these AI generated ads up against? Who better to ask than the AI:

“What are the biggest UK competitors to Mobility UK? Present your answer in a list, with the website and why they are a great company”

ChatGPT output listing competitor websites for the hypothetical charity Mobility UK, including unique selling points for each company and why they are considered great.

These all look quite relevant! You might think that the query around why the companies are great is a comedic touch, but it’s a good way to have the AI summarise it’s best guess at that company’s USP’s.

So we’ve covered the keywords, structure, ads and competitors. Surely there’s not much else that an AI can do for us, right?

 

Landing Page Optimisation

Landing page optimization involves enhancing the performance and effectiveness of your landing pages to achieve better conversion rates and user engagement, whether it’s improving the layout, content, call-to-action, or overall user experience.

Let’s see if the chatbot can help us here. Unfortunately Mobility UK doesn’t have a website to test this, on account of being made up, but we can try it for Uprise Up’s own site

“What changes can be made to improve the conversion rate performance of this page: https://upriseup.co.uk/paid-media/google-ad-grant/ for google ads”

ChatGPT output when asked to optimise the Google Ad Grants service page on the Uprise Up website

This is a lot of information! Some of these points are generic, and show up every time, but they are still all good suggestions for improving a page, It’s even given some specific call to action suggestions to test. But surely that’s everything chatbots can help us with. Surely there’s no further extension of this support, right?

 

Ad Extension Support

Ad extensions can be a time-consuming task in large accounts. Let’s see if ChatGPT can help us with this for the same ad grant page we looked at in the previous example.

“Can you produce a set of Sitelink extensions for the same page, making sure the pages come from the same site”

ChatGPT output when requested to produce a set of Sitelink extensions for the same page in the previous example.

Once again, we’re likely to not get usable descriptions here, but the pages it suggested are all relevant, and are going to save us time over working them out ourselves.

What about callouts? No Problem

“Create 5 callouts for the page, no more than 20 characters each. For each, write a line that rhymes with the callout”

ChatGPT outputs when requested to create rhyming page callouts.

I don’t think AI will be putting poets out of a job any time soon. Terrible rhymes aside, the callouts here are relevant, and would be a good baseline to start your account with.

 

Limitations of AI in Advertising

You might think that the pirate accents or the poems were purely for fun. However,  they show quite clearly that there are limitations on what these systems can currently give us. Pirate accents are easy, but poetry? Poetry it finds very hard. In a similar way, there are elements of Google Ads that the AI finds very difficult to manage.

We’ve already seen character limits be a difficult aspect of ad copy creation, but something we haven’t touched on is how these generative AI systems sometimes have a loose understanding of the truth. There’s even a disclaimer on the ChatGPT console about this:

“Free Research Preview. ChatGPT may produce inaccurate information about people, places, or facts.”

This is potentially very dangerous, especially in the charity sector where the topics being discussed are often delicate and false information could be very damaging. Similarly, in keyword research, we’ve found certain prompts where the AI has provided completely irrelevant search terms. This varies client to client and means that you should check every keyword list generated by these systems.

Final Thoughts on AI and the Future of Advertising

AI may not be ready to take over the world, but its getting pretty close to taking over marketing. Even in the process of writing this blog, Google announced a raft of new generative AI implementations into multiple marketing channels. These systems are still simply tools, however, and like any tool requires skill to operate well. Like it or not, as marketers it is going to be important to learn how to use these tools to adapt to the new world we are rapidly approaching.

    Did you enjoy this blog post?

    Share this article:

    We use Mailchimp as our marketing platform. By clicking below to subscribe, you acknowledge that your information will be transferred to Mailchimp for processing. Learn more about Mailchimp's privacy practices here.

    Contact us