Data processing best practice with gpt-3.5-turbo

Hi everyone,
i’m developing a chatbot using gpt-3.5-turbo. I want the chat-bot to answers questions about ours products technical manuals. It’s terrific! The first tests are very impressive, so we decided to proceed with this project. I’ve pre-processed the data following this example

Like shown in the example i’ve inserted all the text in a pandas DataFrame and, for every paragraph, i’ve put a title and an heading (in the example are then used as index). The question is: is it really necessary to use a title or a heading, is it not enough to use a general key (numeric or alphanumeric) as index? I ask because in my use case not all paragraph can have an heading, and the title in most cases is always the same for it is the title of the manual itself.

Thank’s in advance
Kind regards