Tips:
•
When prompting, give the AI the goal as well (e.g., “We want to re-engage for purchase” or “Goal: improve satisfaction”).
The next best action might differ based on goal (e.g., upsell vs. service).
•
Include any known constraints or preferences in the profile (e.g., “customer prefers communication via text over email”).
The AI will then tailor the action to those channels.
•
Verify feasibility of the AI’s suggestion. Sometimes it might propose something you can’t do (like a discount outside your
policy). Treat it as a creative brainstorm – you can usually adjust the idea to fit your constraints.
•
If you have a lot of customers, you can loop this process: generate next actions for each segment or type of customer first,
or use automation with an internal LLM to scale it. Just make sure you have checks so the suggestions remain sensible.
•
Use next-best-action insights to inform broader strategy. If AI frequently suggests, say, “offer a loyalty reward” for many
lapsed high-value shoppers, it’s a signal that your loyalty program could be emphasized. In this way, individual
suggestions roll up into macro trends that your team can act on.
GenAI Quick-Win Playbook for Personalization
15
Powered by FlippingBook