You might be better off leveraging something like internal campaign tracking.
Similar to how UTMs or CIDs work, you can create your own internal campaign (far less complex than external campaigns (usually only a single param like itm=something or icid=something - but you can make this more complex as you grow), and set this into an eVar with a Visit level attribution.
So on your hero banner, instead of linking to https://www.domain.com/shop/ you link the banner to https://www.domain.com/shop/?icid=hero-banner
Now, when the person lands on the shop page, the internal icid campaign is set into your eVar (eVarx = "hero-banner")... for the rest of the visit, "hero-banner" will be applied to all pages and actions, including your purchases... if you correlate eVarX to your Orders metric, you will see how many of those purchases came from "hero-banner" (or other internal campaigns) and how many are unspecified.
If you have multiple internal campaigns in your site, the above will show the last touched, but you can always apply an attribution model (such as participation, or first touch, etc) to see how other internal campaigns were used within the journey.
However, that doesn't help the data you have already collected.. for that, I would create a segment looking at VISIT level, and look for visits that clicked on the banner. You can use this segment in conjunction with your purchases... you may want to apply sequential sequencing, to make sure within the visit, the banner was clicked before the purchase (just in case someone made a purchase, then clicked on the banner... )
