FAQ

What data sources can I send to ChatGPT via Coupler.io?

Any source available in Coupler.io can be sent to ChatGPT — including HubSpot, Shopify, Google Analytics, Airtable, PostgreSQL, and many more. You can even combine multiple sources in a single data flow using Coupler.io's Join or Append transformations before pushing to ChatGPT.

Do I need to know how to code to use this integration?

Not at all. The whole point of the ChatGPT destination is to let you explore your business data using natural language. Once your data flow is set up and running, you just open the custom GPT and start asking questions in plain language.

What ChatGPT plan do I need?

You need at least ChatGPT Plus to use custom GPTs. The free tier of ChatGPT does not support custom GPTs. OpenAI recommends using the latest available model for the best results when working with the Coupler.io GPT.

Coupler.io ChatGPT is available on all plans. However, the Free plan might not be enough for thorough data analysis.

How often does the data in ChatGPT refresh?

Data in ChatGPT reflects whatever was sent during your most recent data flow run. You can refresh it at any time by running your data flow manually. Once you've completed a successful manual run, you can also set up automatic scheduling so your data stays current without any manual effort.

Start a new chat to fetch fresh data automatically or ask ChatGPT to do so if you want to continue the ongoing conversation.

What's the difference between this and just uploading a CSV to ChatGPT?

Uploading a CSV is a one-time snapshot — every time your data changes, you'd need to export and re-upload manually. Coupler.io keeps your data pipeline live, so you can re-run the flow whenever you need fresh data pushed to ChatGPT. It also handles larger, more complex datasets from sources that don't export to CSV easily.

Can multiple people on my team use the same data flow in ChatGPT?

Each user must authenticate with their Coupler.io account when they open the custom GPT. If you want teammates to access the same datasets, they'll need appropriate access to your Coupler.io workspace and their own ChatGPT subscriptions.

ChatGPT gives incorrect or vague answers about my data — what can I do?

The quality of ChatGPT's answers depends heavily on the structure and clarity of your data. A few things that help:

• Rename columns to descriptive, human-readable names using Coupler.io's transformation step.

• Remove columns that aren't relevant to your analysis to reduce noise.

• Break very large datasets into focused data flows (e.g., one for sales data, one for marketing data) so ChatGPT has a cleaner context to work with.

Answers also depend on the prompt. Asking vague questions like "how's my PPC doing?" will produce poor results. You can brainstorm what you should ask with ChatGPT first and then provide a detailed prompt for data analysis.