Choosing a ChatGPT Website Chatbot

Choosing a ChatGPT Website Chatbot

When Aria, Tage, and me came out with our ChatGPT website chatbot product, we were literally the first company on the planet to provide a professional ChatGPT-based website chatbot. Since then, we've seen thousands of companies and freelancers trying to sell their website chatbots based upon ChatGPT.

This "diversity" obviously makes it confusing for companies to decide which chatbot to chose. I have therefor decided to measure one of the largest ChatGPT-based website chatbots against our technology to illustrate the difference. However, first we need to talk about what's at stake here.


When Walmart increased page load time of 0.1 seconds, they lost 1% of their revenue. This implies that if one website requires 2 additional seconds to load, the owner of that site will lose 20% of its revenue. According to Google 10% of your users will click the back button on their browser unless your website is at least somewhat visible in 1 second. At AINIRO we've spent an insane amount of cognitive energy on optimizing initial page loads for our chatbots. Something you can verify by running this page through PageSpeed Insights. In fact, our chatbots never load more than 48 kB of data, in less than a handful of HTTP requests, before you actually click the button - At which point it will lazy initialise and load the rest of its requirements. The importance of this cannot possibly be over estimated for your website.

Any technology that kills 20% of your online revenue, should not be interesting for you for obvious reasons

One of our primary unique selling points, is that our ChatGPT website chatbots are supposed to increase engagement. If half your users are clicking the back button in the browser because your website takes too long time to load, that is a very, very, very big problem.

The most important feature when looking for a chatbot, is not what it does, it is what it does not do!

Our largest competitor

Below is a screenshot of one of our largest competitors. I will keep them anonymous because I don't like to out people. However, look at the bandwidth consumption.

CustomGPT bandwidth consumption

Their demo page includes a couple of YouTube videoes embedded, and of course like everything originating from Google, embedding a YouTube video on a page significantly adds to the amount of garbage your page needs to download. Reloading YouTube on an empty browser cache freakin' hurts. If you don't agree with me here, you're not spending much time at YouTube. Google Engineers seems to be brain damaged when it comes to "following their own best practices" for reasons unknown. My theory is that Google engineers simply don't care.

The average Google Engineer probably doesn't give a shit!

However, the same vendor did not include their own chatbot on their own landing page. Ignoring the YouTube videoes, and Google Analytics - This should be a key takeaway to understand their chatbots performance. I assume they considered that embedding their own chatbot on their own page would simply have too many negative consequences for their Google ranking and conversions, so they chose not to embed it on their own landing page. These are just my assumptions though, and I have no idea why they chose not to include their chatbot on all their pages.

It could be argued they're selling something they don't want to use themselves ...

For the record, we don't have a sample chatbot page - Simply because we don't need one. Our chatbot is fast enough to be embedded on every single page in our website - Something you can see in the bottom/right corner of this page. So every single page in our website is an example chatbot web page ... 😇

Our chatbot's performance

Below is a screenshot of our chatbot's bandwidth consumption. Notice, we include our chatbot on every single page in our website.

Our chatbot's performance

This implies that our competitor's example chatbot consumes 23 as much bandwidth. Even removing their YouTube videoes, Google Analytics, Google Tag Manager, and other things unrelated to their chatbot, I suspect they would still easily be consuming at least 10x as much bandwidth, probably more like 50x. Yet again, let me repeat my key take away with this article.

The most important feature when looking for a chatbot, is not what it does, it is what it does not do!

And of course, our ChatGPT website chatbot does not download 23 megabytes of data. Which means according to Walmart's findings in these regards, it does not reduce your sales by 100% - Technically more than 100%, since each 0.1 additional second, reduces conversions by 1%, and on a phone on a 3G network or something, we're probably talking about 25 seconds to simply load the page, before the page becomes responsive. Implying every single one of your customers would simply click the back button before they get to see your beautiful chatbot.

I have written extensively about SEO and ChatGPT website chatbots before. However, the way you would notice this is by seeing a significant drop in "Engagement time" in your Google Analytics account. If your engagement time drops in half after embedding your brand new chatbot, this means half your users are clicking the back button because they're simply tired of waiting for your page to load.

How to measure your chatbot

Before you chose a ChatGPT website chatbot, it is crucial that you measure your candidate bots performance. This is the by far most important metric to measure before you chose a chatbot. The way to do this, is to go to your chatbot vendor's example page, click F12, chose the network tab, and click the "Disable cache". This allows you to see how new visitors are experiencing a similar chatbot on your webpage. If the total amount of bandwidth is more than 5MB, my piece of advise is to simply click the back button and find somebody else.

Disable cache

In addition to the above test, you can copy the URL for their demo page, and use PageSpeed Insights to see how their demo page performs. If their demo page performs badly, it's highly unlikely you'll get their chatbot to perform on your page.

Another way to measure this, is to turn off wifi on your phone, empty your browser cache, and visit your own website over a 3G connection if you can. If it takes more than 4 seconds before you see your page, you might want to re-evaluate your chatbot vendor - And/or re-evaluate what other things you're including on your web page ...


I'm obviously biased, if not for any other reasons than that we just so happen to have the fastest ChatGPT website chatbot in the industry - However, there's a reaons why we spent so much energy on optimizing our chatbot technology, while everybody else were obsessed with adding 44 new images, 124 additional fonts, and 1,157 CSS files, downloading some additional 25 megabytes of data, seemingly creating a "better product" than we did - While we focused on what's important ...

Removing as much garbage as possible ... 😉

A Zen circle is perfect because it illustrates a void. In the middle of a Zen Circle, there is nothing. This void is what keeps the ring around it balanced. Focus on less, because as we all know, less is more 😁

Thomas Hansen

Thomas Hansen I am the CTO of AINIRO.IO AS and the CEO of AINIRO.IO, Ltd. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 19. May 2023

Free 7-Day Trial

Create your own Free 7-Day Trial Custom ChatGPT Chatbot
Or contact us if you want us to discuss your particular needs

Call to Action icon to create a ChatGPT demo website chatbot