Transcript Search & Export

What it's for

If handle time rises on an enquiry, or sentiment drops for a sub-enquiry, Transcript View shows you the interactions causing it. You told us the trends were only half the story - to act on a change in the data, teams sometimes need to see the conversations themselves. Transcript View puts that evidence one click from the chart.

Transcript View is a table of interactions. Each row shows the interaction date, Root Cause, Enquiry, Sub-Enquiry, channel, queue, agent, handle time, predicted resolution, predicted satisfaction, sentiment, a transcript preview and the source ID.

From the table you can:

• Search transcript text
• Sort and colour-scan for outliers
• Open any conversation in full
• Select interactions and send them to OI-Q for analysis
• Export the sample to CSV

How to use it

Opening Transcript View

• From a bar in the chart - hover over any bar in the Metric chart and click View Transcripts in the toolbar that appears. The table only shows interactions from that Root Cause, Enquiry, Sub-Enquiry or Lens layer, and it respects every filter on the page - date, model, team, agent, queue and so on.
• From the widget menu - this opens a random sample from everything currently on your dashboard. Use it when you want a representative look across your filtered data rather than one category.

Working with the table

• Records to show - choose 100, 250, 500, 1000, 2000 or 2500 rows.
• Sorting - click a column header to sort ascending or descending.
• Colour coding - metric values are banded green (top 25%), yellow (middle 50%) and red (bottom 25%). Bands are relative to the interactions in your current sample, and relative to the metric - for example, handle time is better when lower, but predicted satisfaction is better when higher.

Managing columns

1. Click Manage Columns in the top-right corner of Transcript View to open the panel.
2. Show or hide columns - every available column is listed with a toggle. Switch a toggle off to hide that column, or on to show it again. At least one column must stay visible.
3. Reorder columns - drag a column by the handle (⋮⋮) on its left to move it up or down the list. The order in the panel, top to bottom, is the order the columns appear in the table, left to right.
4. Click Apply to save your changes, or Cancel to close the panel without changing anything.

Your layout is saved to your user profile, so the table looks the same next time you open Transcript View.

Searching

Table search looks at transcript text only, not agent names, queues or other columns - use filters for those. Two counts appear as you search:

• How many interactions contain your term (e.g. 18 of 1,000 interactions)
• How many times it appears in total (e.g. 42 total matches), since one conversation can mention a word or phrase several times

In-transcript search works inside an open transcript - type a term and every match is highlighted in the conversation.

Reading a full transcript

1. Click the expand arrows on a transcript cell to open the full conversation.
2. It shows as a scrollable thread, with agent messages in blue and customer messages in white.
3. Timestamps appear in HH:MM AM/PM format, and metric values appear at the top.
4. A metadata panel at the bottom shows Interaction Date, Root Cause, Enquiry, Sub-Enquiry, Channel, Queue, Agent Name and Source ID. Scroll it sideways to see everything.
5. Collapse the view from the top-right corner to return to the table.

Selecting interactions

• Click a row to select it.
• Shift-click to select a range - the fastest way to build a large sample.
• When rows are selected, an action bar appears with two options:
  • Assess in OI-Q sends the selection for deeper analysis. You can send up to 1,000 interactions to OI-Q.
  • Clear selection deselects everything.

Exporting

Click Export to download the current sample as a CSV file. The export follows your Records to show setting, so raise it before exporting if you want more rows.

Use cases

Investigating an iPhone login issue for the digital banking team

Your digital banking product team suspects iPhone users are having trouble logging in. They've asked for examples of the worst interactions to take into their next sprint review.

1. Find the enquiry. In the classification filter bar above the charts, click Enquiries. Open the dropdown, type "Can't log into" to find Can't log into Internet Banking App, select it and click Apply. Set the date range to the period the team is concerned about.
2. Open Transcript View. Hover over the Can't log into Internet Banking App bar and click View Transcripts. The table is scoped to that enquiry and your current filters, with the most confidently classified interactions first.
3. Widen the sample. Set Records to show to 2500 so you're searching as many interactions as possible.
4. Search for "iPhone". The results count shows how many interactions mention it - for example, 452 of 2500 interactions. That count alone tells the product team how often the device comes up in login complaints.
5. Pick out the worst experiences. Scan the Predicted Satisfaction column for cells banded red (the bottom 25% of the sample). Click each red row until you've selected 25. Open one or two with the expand arrows to check that the search matches are about the login problem rather than an incidental mention.
6. Set the columns for export. Open column management and hide everything except Enquiry, Handle Time, Predicted Resolution, Predicted Satisfaction, Sentiment, and Transcript.
7. Export. Click Export to download the CSV and share it with the product team.

Want to see more? 

See the Insights #2 module in OI University.