It’s textbook to make a feature behave the same in different environments, but what if that’s wrong?

When introducing Translation features across chat messaging, knowledge articles, and email, it made sense for product execs to ask why each experience was different, and push for them to be the same. In this case, however, user needs, capabilities, and pricing expectations meant that different experiences would be best for the user.

I worked with design and other product managers to build our case and convincingly present rationale to do the right thing.

How did each separate experience work?

  • Chat conversations were priced in an all you can eat manner, so automatic translation made sense. Don’t make agents think, just show what they need to see to be effective.

  • Emails, and their related threads, can be much longer, and were priced by the character (including spaces!) so translating only what is necessary is the most economical option. Initiating translation of translation of specific content gets the user what they need, when they need it.

  • Knowledge articles brings in an issue of scale. They need to be available to internal and external viewers in multiple languages. Add to that, users can choose whether translations are generated by AI, or sent to a human translator on a per-language basis.

If each experience was treated the same way it could be catastrophic.

This information convinced execs that our directions were the right ones.

  • Live translation in email: Insane volume, insane cost.

  • On-demand email translation in chat:
    Counter to the immediate nature needed in live conversations; forcing more overhead than necessary.

  • Either experience in Knowledge: Forced repetitive tasks with timeboxed ability to work, untenable with the scale required for multi-language KBs.

  • Any combination of these concessions would add liability to Salesforce, additional work for product and engineering, and would erode trust with customers once their bills showed up.