With AI dominating discussions worldwide, it’s high time we explore its role in the PPC realm. Let’s dive into how artificial intelligence is reshaping the world of digital advertising.

Table of contents:

1. Definitions

2. Incentives for using AI

3. PPC AI tools

4. The future of PPC

Let’s start with the definitions

1. Artificial Intelligence (AI)

   is the theory and development of computer systems able to perform tasks normally requiring human intelligence.

2. Machine Learning (ML)

   enables computers to learn from data without being explicitly programmed to do so. ML is an integral part of AI – and AI is largely built on machine learning.

3. Generative AI (GenAI)

   is a type of AI that can generate new content (text, image, audio) based on existing content.

4. Large Language Models (LLM)

   LLMs are a subset of Deep Learning, it is a type of AI program that can recognize and generate text, among other tasks.

So how does ChatGPT fit into all this?

ChatGPT is a Conversational AI using Large Language Models (LLMs) to generate human-like texts.

Now that we are familiar with the different subtypes of AI, let’s take a closer look at how these AI-powered tools can help a PPC (Pay-Per-Click) professional’s daily job.

Questions that might pop up:

• What type of job should be delegated to an AI tool? Should it be tasks that we could do on our own? Or should they be tasks where we are missing certain competencies? Like creating regression analysis to predict future trends and metrics.
• What’s the risk of delegating tasks that we cannot validate properly?
• Should we trust these AI-driven tools at all?

• If Google Insights Tab shows whether we are tracking current search trends with our campaigns or falling behind in seasonality, what does it matter?

What is ChatGPT good for?

Just knowing what it’s not good for is a good start.

A few things to keep in mind:

1. ChatGPT is an LLM model after all, not a PPC Expert.

   Know what you ask for: If you’re missing the context of your questions, how can you validate the answer given by ChatGPT? (See also the GIGO concept)

2. Fact-checking and validating the output is a must.

   You can’t take the output as gospel, so don’t skimp on the verification.

3. Don’t ask about last night

   ChatGPT was trained with data only up until September 2021, so it won’t be able to help you with the latest trends (e.g., new campaign or ad types).

A good prompt takes time

It doesn’t hurt to know what you want the end result to be.

To realize the potential of ChatGPT, it is important to always provide the right context. You should start with a system prompt, describing the area of expertise you are working in and the question you are asking for help with. In more specialized areas of expertise (like PPC), it is important that we also train the model, by using web browser plugins to let GPT read documents and articles that are relevant to us.

Since January 2024, the Chat GPT Store has also been available, where you can use custom versions of ChatGPT, which users have even trained in specific fields.

What can ChatGPT do for PPC?

• Keyword Research: gives you a kick-start.
• Expanding keywords based on an existing root list.
• Writing ad texts (headlines, description lines).
• Audience research: Provides suggestions for potential target audiences based on a brief.
• Creating missing (complex) Excel functions.
• Basic coding: writing JavaScript code for a Google Ads script.

AI-powered tools in PPC

What AI-based solutions and tools are already built into advertising platforms and what do we use them for? Actually, we’ve been already using quite a few AI-powered solutions in PPC – the one is called value-based smart bidding powered by predictive analytics.

In the old days, how did we know how much a paid click was worth for us?

We used keyword-level bidding, meaning we assigned max bids to every keyword, then looked at the actual performance on many different levels, such as GEO, browser language, device, and OS, time, search term, and adjusting the bids manually after analyzing all these user signals.

And now with the Value-based Smart Bidding, we just need to give some data to the system: budget, target KPIs (revenue, transaction, lead, etc.), and let the AI do the job for us. Predictive Analytics looks at the historical data, analyzes all these user signals, and makes predictions about the potential outcome automatically and in real-time. It “only” needs a lot of data to make these predictions accurate enough.

What are the trends?

Whichever of the big ad platforms you look at – Meta, Google, LinkedIn – they are going in the same direction: target broadly, provide 1st party data, and give a bigger budget so that the system has enough data to learn. In short – give up (some) control as an advertiser and leave/delegate more and more work to AI.

• Simplified account structure

• Broad(er) targeting

• Automatic Creative Creation

Where does this lead?

More data but less control requires more trust from advertisers and PPC professionals.

My advice is don’t trust the AI blindly, test and back up your decisions with data.

What shall be the focus of our work in 2 years?

Still, to this day, we spend too much time managing campaigns, when to run what, how much to spend, and then reporting on these campaigns, leaving little time to analyze campaign data in any meaningful way, to gain insights from it – and especially to take two steps back to see a bigger picture and being able to work on strategy.

What’s gonna happen with AI?

What is likely to happen is that more and more AI-driven tools will be quietly integrated into the ad platforms – dynamic creative assets and automatically applied audiences will no longer be a recommendation but a default setting and that cannot be turned off. Also, the hype will be replaced by peaceful collaboration – where our job as experts remains to not lose sight: AI serves us and not the other way around.

The future is that the hype will subside, and AI will work quietly in more and more areas, as it has been doing so far. The question is not what this future will be like but what we, as experts, can do in the meantime to master as many applications as possible and make them a part of our regular practice.