Auto-optimize a metric we want to minimize
- Description - For Adobe Target personalization and recommendation, there are use cases where we want to minimize KPIs like page load time, errors, agent contacts. It would be great to 1) set a reduction goal in Adobe Target, or 2) (less than ideal) be able to set negative outcomes in mbox attribute or prop/evars to enable optimizing towards zero
- Why is this feature important to you - We'd like to make more extensive use of Adobe Target Premium feature in digital self-service and show value reducing costs.
- How would you like the feature to work - You would select an Adobe Analytics metric to optimize, let say agent-assisted chats. In Adobe Analytics, the upward trend is set as "Bad (red)". In Adobe Target, we select select Agent-Assisted Chat as the primary metric and second drop down would appear: "Drive this metric: Up / down" with down preselected based on the analytics setting. ML models would then work to minimize the number of chats an agent need to handle by recommending the most helpful solutions.
- Current Behaviour -In Adobe Target, we select Agent-Assisted Chat as the primary metric. : "Drive this metric: Up / down" with down preselected based on the analytics setting. Since the ML models only maximize, the ML models would then work to increase the number of chats an agent would handle to take by recommending the least helpful solutions. OR we would need to find a proxy metric (highest rated support articles) that we would want to maximize that we believe causes less agent chats and then A/B test that relationship, but we may find articles that drives to chat preform high in article rating, but lower in issue resolution, time to resolution and overall customer satisfaction.
