I can’t be of much help without knowing how that end of the process works.
Well, as I mentioned in previous messages, many things can be done if you have skills (or budget) or you’re comfortable with an incomplete or difficult-to-sustain and maintain approach. Here are your options.
Modify the Intake at the Source
This probably requires that you determine why the encoding issues are occurring in the first place and modify that process to eliminate them.
Add a Sanitization Process at Google Sheets
This requires Google Apps Script to transform every record as it arrives and before Integromat retrieves it. While Apps Script has excellent libraries for encoding and decoding HTML entities, it does require code and process automation to use them in this scenario.
Add a Sanitization Process at Integromat
This approach likely requires the use of special Integromat HTML or text functions that can be weaved into the automation process. There’s a good chance transformation into plain text should make the outcomes more predictable, but you may lose some formatting that is actually needed when the content arrives in Airtable.
Add a Sanitization Process at Airtable
This could be done with formulas looking for every instance of the “odd” texts and either eliminating them altogether or transforming them into their decided equivalents. Or, you could do something similar in a Script Block. Both of these options come with a lot of issues such as knowing in advance all of the “odd” situations the source content may include. This is not easily predictable and Airtable doesn’t provide the ability to perform a global HTML entity transformation (i.e., Script Blocks do not allow use of the DOM thus ruling out a universal decoding process).
Search and Replace Manually
This is about as ugly as it gets and likely to lead to lots of missed entities. Since all entities start with an ampersand (&), it might make discovery a little easier, but even so, this has got to be the worst possible pathway.