Motorcycle Franchise Beats Competitors to AI Visbility

Goals

• Increase brand recommendations from AI Engines

• Gain visibility with AI Engines before their competitors

• Get recommended by AI Engines when people search for motorcycle events and motorsports news

• Visibility from AI Engines to rider community

• Build trust through AI endorsements

 

Strategy

Since generative artificial intelligence is a new field, and everyone in this industry is self-trained, there were no instruction manuals available. We decided to digest AI patents from multiple sources. This allowed us to triangulate our knowledge of AI Engine information retrieval methods. During the process we noticed information retrieval trends that AI Engines were using. Using this, we wrote codes that AI patents are looking for.

​

Naturally we did not want to test a new process on a client's site. So we built a new website using information from AI patents. During construction, we installed our process on that site. Then we read the data for 90 days. After installation we asked various AI engines, that were important to our goals, what we could do to make specific chunks more visible. Some of the suggestions helped refine our process.

​

After the successful test on our site, we deployed the codes in areas that AI engines are inspecting on the client website. Within a week the client reported getting calls from people saying they found them on ChatGPT.

 

Obstacles

The main obstacles were:

• Creation, we had to research and develop our own process

• Lack of measuring technology

• Lack of standardized metrics

• AI Engines still calibrating internally

 

This is a frontier technology. Information from outside sources is watered down. Creating this requires frontier level research and development. It required reading scientific entries and engineering a process.

​

Another obstacle is lack of data. The technology is so new that the data measuring tools were non-existent when we started. Quantifying was especially challenging if your transaction required an off-line interaction. If you’ve ever bought a car or motorcycle, you know how it goes. You might not like the vehicle after you see it in person. This makes connecting the prompt that generates the search, click, and sale more challenging.

​

Internal calibration by the AI Engines is a definite hurdle. Using ChatGPT as an example, there are times you ask for a review and it comes from Bing, another time from Yelp, another from Google. The AI Engines haven’t settled on the mix of information they’re going to use. This makes spotting trends difficult. Designing a process is challenging if the target is moving. It requires testing, failing, and recalibrating until you reach the goal.

​

On a positive note, the uncertainty provided valuable data in terms of what websites are being recommended by AI Engines.

 

Results

The results are a ~4% increase in traffic, has contributed to a 12% increase in sales. As mentioned in Obstacles, one of our issues is measuring an off-line sale. Salespeople are reporting leads coming from ChatGPT are more educated, and further along in the buying process.

​