ChatGPT Optimization: An Investigation - Part 1

As HLM is always striving to stay abreast of industry trends and what's next in order to provide our clients with proactive recommendations, we've performed our own investigation into ChatGPT and the shift in consumer behavior towards using this AI tool as a search engine, instead of searching for businesses and answers to questions on Google. Here's part one of how we're doing this. 

What We're Doing

We receive calls and emails every single day from companies looking for help solving their business & marketing problems. We always ask how they heard about us.

As you can guess, the vast majority say "Google," (with a close second being "word-of-mouth," to include referrals from our official partners), but a growing answer is this:

That sounds an awful lot like a typical Google search, doesn't it? 

We like to take a somewhat pragmatic approach to things here at HLM, instead of getting excessively mired in debates around what the "experts" say, or trying to predict the next "big thing."

Instead, we follow the data and the feedback we receive from our work with clients.

So we launched our own grassroots effort to determine how ChatGPT is answering this question:

If our investigation uncovers any trends or insights that allow us to reliably determine how ChatGPT is answering that question, then it's possible we'd be able to make recommendations to clients on ways to ensure they're showing up when ChatGPT is being asked questions relevant to their business.

Likewise, if we're unable to decipher any clear patterns, then we've at least learned that ChatGPT might not be something businesses should be overly concerned with trying to optimize for, and should instead spend time and resources focusing on more tried-and-true methods of marketing.

How We're Doing This

Like all good things in life - we begin with a spreadsheet. We really like spreadsheets. 

We started by identifying the top 50 U.S. markets that are relevant to our clients. We then asked ChatGPT this question for each market, for several different types of industries (we'll use restoration companies as our example): 

After asking ChatGPT the question above, we'd be provided with a list of what ChatGPT considers to be the best restoration companies in that market, similar to the screenshot below:

You can see from the results above that we're provided with 4 options, each of which cite the businesses website itself as a source of information. The fact that it cites the businesses' website means that what has always been important remains important - every business needs a robust website optimized for SEO, with significant amounts of good quality, fresh content.

But we wanted to know what other sources ChatGPT is using to compile this list, so, in cases like the above one, we then ask ChatGPT this question:

Sometimes ChatGPT would respond with a list of third-party websites such as the Better Business Bureau, Yelp, Angie's List, etc., but more often than not, ChatGPT would cite the business websites once again. In those cases, we then asked this:

Occasionally we'd then receive a response like the image below, which again speaks about content on a businesses website, their reviews, and their testimonials. This furthers our belief that what has always been important remains important - robust website content and social proof such as reviews and testimonials.

We're Not Giving Up!

However, as HLM doesn't give up that easily, we continue to push ChatGPT with the same question until it gives us a little more insight into where else it's deriving info from, and eventually we're presented with info like this image below:

ChatGPT finally gives us some third-party sources that it's used to provide us a list of top restoration companies in Birmingham, AL. You can see from it's fourth entry though that it isn't perfect - it just provided us with a restoration company that it hadn't referenced at all up to this point.

So What Does This All Mean?

Remember, the title of this post is, 'ChatGPT Optimization: An Investigation - Part 1' - meaning that this is just step 0.5 out of many more steps to come in our investigation. 

But already, some takeaways are clear:

• We'll have to gather data from all 50 Markets once every 2-3 weeks to record the responses, and see if any clear patterns emerge. It's possible that in 3 months we've discovered that the Better Business Bureau or Expertise.com emerge as - on average - significant sources of information for ChatGPT. If that's the case, we'd recommend optimizing those listings as much as possible.
• ChatGPT still heavily relies on information found on business websites, just like Google does. This includes reviews and testimonials, as signals to ChatGPT that these businesses standd out from others and are a safe bet to recommend.
• There's a lot of variability in ChatGPT's responses. This tells us that we have plenty of work left to do, asking ChatGPT tons of additional questions over time, to see if patterns in it's responses emerge.

What If This Is All A Waste Of Time?

We know it's completely possible that, after months of investigation, we glean no insights into the inner workings of ChatGPT. We might not see any patterns. We might not have any new recommendations to give clients on how they can leverage ChatGPT and AI in general to grow their businesses.

That in itself would be worth our time, and would be a finding that provides value to our clients.

How's that, you ask? Because we can at least then tell clients, with significant confidence, that they can stop worrying about ChatGPT, and instead focus on the tried-and-true methods of growing a business and building a brand...

Be on the lookout for Part 2 of our ChatGPT Investigation soon! In the meantime, if you need help growing your business right now, give us a call at (888) 717-4249 to speak to someone within minutes, or book a call directly with a consultant on our contact page.