Auto-Tagging, Salesforce, CSV Exports, and More
Curated release notes for February 2019
Since we first launched with alpha partners in October 2018, we’ve continued to invest across the product to help you understand and act on the substance of your customer conversations — everywhere they happen. This includes new connectors to unify your data; enhancements to the interface to help you clean up, enrich, and report on your data; and new destination to help you export, broadcast, and sync data to another location.
This year we publicly released the Frame hub with a major breakthrough in Tag Management, enabling customers to build a new source of truth in a risk-free sandbox environment. Here are a few more areas where we improved.
Performance
One of Frame’s core values is to make interactive data exploration exciting — so it’s important to keep things snappy. But as we opened Frame to the public, we saw a surge in large accounts signing up, which exposed bottlenecks in our real-time search, reporting, and neural tagging infrastructure.
The good news is that the performance enhancements didn’t just make things more usable for big customers — they made Frame faster for everyone.
New Features
More CSV Export Options
One of Frame’s goals is to put an end to long waits or painful manual work in filtering and exporting data . For example, you shouldn’t need to write an API query just to export a simple transcript from a handful of tickets.
In Frame, it’s already possible to export any arbitrary list of your conversations from other systems, including full transcript with comments, simply by crafting a search. Check the box next to the conversations you want — or export them all.
Now, you can also export:
- Any individual visualization within a dashboard
- The full list of tags, including # of applications and when it was last used
The option to “expand {tags/contacts}” enables pivot-table-based work in Excel/Sheets, which depends on one tag per column.
Salesforce Service Cloud Connector
Now you can bring your chats and emails from Salesforce into Frame. Search, clean up, and report on your conversations with the assistance NLP-driven enrichment.
This connector is currently in beta. If you’d like to be among the first to use it, contact us or sign in and connect Frame to your Salesforce account.
Auto-Tagging Now Available on All Accounts
You can enable auto-tagging on any tag, whether it’s coming from a source system(s) or one you created in Frame.
Here’s how it works:
- Frame continually analyzes all your tags. For any tag with at least 20 applications, it generates a neural model representing the conversations to which it has been applied. (~50 applications is better, but 20 is a good start.)
- Using this neural model, Frame suggests conversations that should have the tag applied. Using the review/edit flow, you can take Frame’s advice or ignore it — one conversation at a time.
- When the neural model seems accurate enough, you can turn on auto-tagging. Doing so asks Frame to evaluate conversations as each message arrives, and apply the tag for you when it seems appropriate.
- Neural models update nightly, or on-demand if you’re using the bulk editor to make changes to quickly train a new tag.
Frame’s customer success team works closely with customer team members to (a) identify good use cases for auto-tagging and (b) bulk edit the history of conversations to further tune the model.
A note on the word “tag”
Some teams prefer to think of tags as strictly operational in nature — e.g., urgency, priority, routing and escalation — while using the term “insights” or “categories” for analytics purposes. That’s OK!
Whatever you call it, at some layer, your data are being organized with labels.
We prioritize human/AI collaboration over lean-back automation, and thus expose these labels as a “tag” that users can add, remove, group, rename, and more. We also make tags and tag groups first-class citizens in the search and reporting UIs, enabling rapid iteration and report generation.
View by Conversations Started
The Overview dashboard can already display volume of activity based on the number of exchanges, messages, or active time.
Now, it displays volume based only on exchanges that started in the search period as well. This helps Frame align with the metrics offered in source systems like Intercom and Zendesk.
Design and Share Your Dashboards
The Tags dashboard displays trends and co-occurrence data on the current search. To tailor which tags are displayed, use the interactive legend to add/remove tags and choose the color for each series. Every change updates the URL, making it easy to bookmark and share exactly what you see now.
