Wouldn’t it be great to actually know what your customers are thinking? What they like, the common reasons why they reach out for support, and more? That’s exactly what Einstein Conversation Mining helps you do!
By tapping into customer conversations – whether in chats, emails, or calls – this AI-driven tool pulls out valuable insights from each interaction, giving you a clearer picture of what actually matters to your customers.
What Is Einstein Conversation Mining?
Einstein Conversation Mining takes advantage of advanced natural language processing (NLP) and machine learning, then uses them to spot common themes, track customer sentiment, and uncover trends across every customer touchpoint.
This functionality isn’t just tech for tech’s sake – it’s also a powerful tool for service and sales teams alike. It helps teams see what customers care about, flag potential issues early on, and make data-driven decisions that can elevate customer satisfaction.
Features and Benefits of ECM
Einstein Conversation Mining (ECM) offers a range of powerful features designed to help teams dive deeper into customer conversations, allowing businesses to turn these interactions into strategic insights. Here’s how each feature can benefit your team:
Automatic Call Transcriptions
ECM can automatically convert spoken conversations with customers into text. By doing these call transcriptions, there is no longer a need for manual note-taking during support. These transcripts are automatically analyzed by ECM, ensuring no detail gets missed.
Sentiment Analysis
This feature can automatically detect customer emotions throughout interactions and identify whether they’re positive, negative, or neutral. With sentiment analysis, teams can quickly identify if a customer is satisfied or frustrated, helping address potential issues or even spot upsell opportunities.
Topic Identification
Topic identification looks at and highlights the main topics discussed during customer calls. Doing this lets sales and service teams quickly pinpoint areas of interest or concern so they can respond effectively to customer needs and prioritize high-impact areas.
Actionable Insights
This feature uses AI to generate recommendations based on conversation data, providing your team with clear guidance on the best next steps to take with customers. This helps your business’ support remain proactive and make customer interactions more personalized.
Trend Analysis
Trend analysis detects patterns across multiple customer conversations, revealing not just recurring issues, but also successful strategies. With this information, your team can identify what’s working well and maintain it, or spot potential pain points and improve processes accordingly.
Conversation Summarization
This feature automatically generates clear and concise summaries of customer calls so there’s no need to manually sift through full transcripts. These summaries make it easier for teams to quickly review key points and move forward with relevant information.
Customizable Dashboards
Customizable dashboards allow users to tailor their reports and metrics to specific needs, ensuring that insights are relevant to their roles and objectives. With this flexibility, teams can track the specific metrics that matter most and take action based on tailored insights.
How Does Einstein Conversation Mining Work?
Imagine a customer calling a service center to reschedule an appointment. The conversation might look like this:
Agent: “Hi! How can I help you today?”
Customer: “Hi, I need to reschedule my appointment for next week.”
Agent: “Sure! I can book a new slot. How about Tuesday at 2 PM?”
Customer: “Actually, do you have anything later?”
Agent: “Yes, 3:30 PM is available. Would that work?”
Customer: “Yes, that’s perfect!”
It’s a pretty simple interaction, but here’s how ECM uses the features mentioned above and extracts insights from it:
- Data Collection and Storage: ECM automatically gathers data from channels like calls and messages. In this case, the ECM generates a call transcript and is stored in Salesforce, ready for analysis.
- Natural Language Processing (NLP): Using NLP, ECM detects key aspects of the conversation like tone and sentiment. In this example, it identifies “appointment reschedule” as the main “contact reason”. ECM organizes your data into these contact reasons across thousands of records. In this case, it classifies the rescheduling request as a common support inquiry using:
- Pattern Recognition and Machine Learning: As ECM processes thousands of interactions like this, it learns to detect patterns. Over time, it might recognize “appointment rescheduling” as a frequently recurring topic, flagging it for potential automation to save time.
- Sentiment Analysis: This interaction does not have any negative words, and so ECM picks up on the customer’s “positive tone” once their needs are met (for example, the response with “Yes, that’s perfect!”). Recognizing tone and sentiment helps teams track satisfaction levels, which is very important for proactive service adjustments.
- Conversation Turn Analysis: ECM logs each exchange, or “conversation turn.” In this interaction, we count four exchanges or conversation turns before the issue is resolved. ECM also uses this metric to help teams identify simple requests (most likely those with the fewest conversation turns) that could be automated for efficiency.
- Actionable Insights from Reports: From this conversation and others like it, ECM provides insights once you generate a report, like “appointment rescheduling” being a high-frequency contact reason, for example. This helps teams consider automation options for similar inquiries, freeing up time for agents to focus on more complex issues.
- Training and Enhancing Bots: Since ECM reports can be accessed while building a bot in Einstein Bot Builder, the insights gathered from ECM can be used to enhance bot dialogs and make their conversation handling a lot smoother.
ECM’s approach lets service teams understand customer needs better and helps spot areas for improvement by suggesting to automate frequent requests, improving overall efficiency and satisfaction.
Setting Up Einstein Conversation Mining
ECM is available on Salesforce Performance, Unlimited, and Developer Editions. To activate it, navigate to Setup and search “Einstein Conversation Mining”. The result can be found under Service Cloud Einstein. Simply click on it and toggle the switch.
To generate reports on conversation data, enable the Connect to Data Cloud feature. This links your conversational data with Data Cloud, making it possible to create Einstein Conversation Mining reports. To ensure the setup is complete, check for a “Connected” status.
Reporting and Visualization With ECM
One of the most valuable aspects of Einstein Conversation Mining is its amazing reporting capabilities. However, it is important to take note that ECM requires at least 2,500 records with an identifiable contact reason before you can generate a report with it. Because it excludes conversations without a clear and identifiable contact reason (like incomplete conversations or those with no actual business use case), this count of 2,500 applies even after filters.
Start creating by clicking the New Report button from the same window where we enabled ECM earlier.
Give your report a name and select the channel where you want to collect customer conversations from. You can pick Chat, Email-to-Case, Web-to-Case, or Enhanced Conversations.
Next, define a specific date range and channel type where you’d like to focus your analysis on.
Aside from the date range, you can also include segmentations that help filter out the conversation data even more. With this, you can focus more on only the specific categories or metrics you’re interested in.
Hit ‘Build Report’ on the bottom right to generate your report! While the report can take up to 24 hours to build, the insights gathered from it can be used to enhance and train your Einstein bots’ capability to handle support conversations specific to your business.
The report details can also be accessed in Service Intelligence, the app for Service teams built on Data Cloud. Since Service Intelligence provides prebuilt dashboards, they enable a visual and analytical view of ECM insights with the help of AI. This is great for people who prefer concise visuals when absorbing data!
By using these reports, teams can respond to emerging issues more rapidly, deploy resources more strategically, and even influence broader organizational strategies based on customer feedback and patterns. From the insights gathered on these reports, you can also update and train your Einstein Bots!
Considerations and Best Practices
ECM is great and all, but before enabling Einstein Conversation Mining, here are a few things to consider:
Language Limitations
Currently, Einstein Conversation Mining supports only English. If your org has multilingual customer bases, this may limit ECM’s effectiveness across all service interactions.
Email-to-Case Reporting
If you’re using Email-to-Case, keep in mind that only the initial email in an email thread will be analyzed. Subsequent responses are not included, which may impact reporting completeness for cases involving lengthy exchanges.
Web-to-Case and Sandbox Reporting
Web-to-Case data is not accessible within the Service Intelligence dashboard, and Enhanced Conversations reports are unavailable in sandbox environments.
Exploring Contact Reason Data
Within each report, you can explore “contact reasons” in two ways: by clicking the summary to see related details within a topic or by selecting “View Details” in Top Recommendations.
Additionally, a downloadable list of contact reasons is available within the Topics section, with up to 500 excerpts per contact reason available in CSV format.
Einstein Conversation Mining is not to be confused with Einstein Conversation Insights (ECI). While they may seem (and sound!) similar, ECI is more like a conversational intelligence software that focuses more on each individual conversation. It can immediately extract key information and reliable next steps from a call without having to take notes or listen to hours of recordings. It’s geared more towards Sales teams.
ECM on the other hand, analyzes conversations in bulk to identify trends and patterns – which is why a single ECM report can take up to 24 hours to build! It’s geared more towards Service teams.
Summary
These days, providing customer service that’s unique to your business is crucial. Teams looking to gain a deeper understanding of their customers’ sentiment and feedback can greatly benefit from Einstein Conversation Mining, which has the potential to revolutionize the customer service industry.
Using AI-driven insights from conversations across multiple channels, ECM enables teams to identify trends and high-frequency contact reasons that can be immediately addressed after report analyses.
Despite restrictions around language support and other factors, it’s great that ECM gives teams the ability to proactively deal with issues, improve everyday activities, and engage in strategic planning. Embracing AI and the value of ECM’s insights can be transformative – helping teams work smarter, not harder.
