ChatGPT+ Avian GA4 Plugin
ChatGPT+ Avian GA4 Plugin
I am going to introduce you to ChatGPT+ via the plugin – Avian.
Overview of Video Dialogue:
Create an account with Avian. They do have a free trial on ChatGPT+ which is the paid version.
Go to the bottom left-hand corner and the 3 dots and go to settings. Make sure you click on Beta features, then enable the plugins and code interpreters. Click on the plugins and then the down arrow and go to the plugin store. Search for the Avian plugin and install it. There is also free plugin called Daig.ram which is also worth downloading.
You can see that we’ve got property name, because I’ve connected several different GA4 properties within the Avian app. Indicate the time period, the metrics, break down by dimensions, table rows and table columns to tell it what type of table you want and what the format is. We’ve copied that as it’s better to do it outside of ChatGPT just in case you type it in, and delete it by mistake.
It can take a few minutes, for the Avian plugin to retrieve, and process the data. It does have some limits, for GA4. It’s what they called 8,000 tokens, which is, I believe, around 6,000 words. When you’re querying page paths, URLs, page titles, that can be used up fairly quickly. Try and avoid those sorts of queries or do it in a series of small queries, say a month or a week at a time, depending how many page paths you have.
There’s no filter on the prompts because they don’t have filters, which is I think is a major limitation. That’s something they need to add in as otherwise you can’t refine your query.
It’s come up with the data, over the months, showing it by device category. Now, what I would do if I had filters is set desktop and mobile only, but I haven’t. We’ve got, Smart TV showing up by 2 users by looks of things or 4 users in total.
It’s asking if I want to visualise the chart. Here you have to carefully specify what to visualise. Say, I want to create a time-series chart using the user conversion rate data only, otherwise it will try and combine the two sets of data and it won’t be very useful.
Let’s put that in to prevent adding to chart the users alongside the conversion rate, which is very different metrics. It will take a few seconds to process. It’s using the Daigr.am plugin now so it will be interesting to see how that visualises the data.
Sometimes it’s faster to do the tables yourself because you know exactly what you want. Sometimes I upload data from GA4 in a spreadsheet, and that works well. Supposedly this plugin is designed for people who don’t know how to use GA4, but I don’t see how that will work. You still need to know things like the event and parameter names. You would need to specify exactly what you want and the type of analysis or chart. I think you are still going to need people who know what they’re talking about to put in the prompts.
They have got, some data now, desktop and mobile primarily. It took a while because it’s not as fast as they sometimes make out but it did get there.
I’ve explained how to connect to ChatGPT+ using a plugin named Avian. To begin, users need an Avian account and can access the plugin on the paid version of ChatGPT+. After enabling Beta features in the settings, you can install the Avian plugin by going through specific steps. There’s also mention of another free plugin called Daig.ram.
For using the Avian plugin, we suggest a specific prompt format that includes property names, time periods, metrics, dimensions, table rows, and table columns. This prompt is better formatted outside of ChatGPT to prevent mistakes, and then it’s pasted into the chat to retrieve data.
The Avian plugin has some limitations like token limits and lack of filtering options. We suggest avoiding queries that consume tokens quickly or breaking down queries into smaller periods. The absence of filters also makes refining queries challenging.
The Avian plugin fetches data, and we mention visualising the data using the Daig.ram plugin, which takes time to process. We find that manual data manipulation sometimes works better and questions the plugin’s suitability for those unfamiliar with Google Analytics (GA4) concepts.
Ultimately, We demonstrate how to access data and create charts using these plugins but also acknowledge limitations in speed and functionality.