Skip to content
AskVault Docs
Try Free →

The Analytics dashboard

Written by , Principal Architect, Networks and Tech, AskVault · Reviewed by Allison Chan
Last updated: · 5 min read

What the dashboard shows

Eight panels arranged in a single scrollable view. The KPI cards and query-volume chart are on every paid plan; the other six panels require Growth or above Growth+ :

  1. KPI cards. Total queries, conversations, satisfaction, average response time.
  2. Query volume over time. Area chart of daily query counts.
  3. Top questions. Most-asked questions with answer count. Growth+
  4. Unanswered queries. Questions where retrieval came back weak. Growth+
  5. Channel breakdown. Pie chart of conversation volume per channel. Growth+
  6. Satisfaction trend. Thumbs up/down ratio over time. Growth+
  7. Peak hours heatmap. Day-of-week + hour-of-day grid. Growth+
  8. Sources usage. Which documents drove the most retrievals. Growth+

One date-range picker at the top controls every panel. The whole dashboard exports as a single CSV on Business and above (see CSV export).

KPI cards

Four cards at the top show this period's headline metrics:

  • Total queries. Count of AI-generated responses across all channels.
  • Total conversations. Unique conversation threads. Multiple queries per conversation count once.
  • Satisfaction. Percentage of thumbs-up votes among responses that received feedback.
  • Average response time. Mean latency from customer message to bot reply.

Each card shows the value plus a percentage change vs the previous period (e.g., last 7 days vs the 7 days before). Green for improvement, red for regression.

Date range

The date-range picker at the top controls all panels:

  • Last 7 days. Useful for spotting this week's issues.
  • Last 30 days (default). Most common view.
  • Last 90 days. Longer-trend visibility.
  • Custom range. Pick any start and end date.

Up to 365 days of history in one request, on every plan that includes this panel.

Top questions

Lists the 25 most-asked questions in the date range. Each row:

  • Question text (de-duplicated across paraphrases).
  • Count. How many times asked.
  • Average satisfaction for responses to this question.
  • Top source the bot cited.

Useful for spotting:

  • High-volume questions to add to your FAQ. If "what's your refund policy?" comes up 200 times a month, it deserves a dedicated FAQ entry.
  • Underperforming responses. A high-volume question with low satisfaction signals the bot's answer isn't good. Investigate.

Click any row to see the conversations behind it.

Unanswered queries

Lists the 25 most common questions where the bot's confidence was low or it refused to answer. Each row:

  • Question text.
  • Count. How many times asked.
  • Top retrieval source the bot found (even though it wasn't confident enough).

This is the most actionable panel. Unanswered queries are knowledge gaps. Plug them by:

  1. Adding a Q&A pair with the canonical answer.
  2. Adding a snippet with the relevant info.
  3. Re-crawling content if your documentation has the answer but indexing missed it.

Review weekly. Reduce the unanswered list by 5 to 10 entries per week and the bot's coverage compounds.

To go further, the AI Insights panel clusters these failures into topics, diagnoses why each one failed (content missing, unclear, contradictory, or out of scope), and drafts a fix you can apply as a Q&A pair in one click.

Channel breakdown

A pie chart of conversation volume per channel. Helps you see which channels are getting traction.

For most B2B SaaS:

  • Widget dominates (60 to 80% of conversations).
  • WhatsApp for India and SEA-focused customers (10 to 30%).
  • Email Assistant for B2B support escalations (5 to 20%).
  • Slack, Telegram, Discord as supporting channels.

Track the breakdown over time to spot channel shifts (e.g., WhatsApp adoption growing) and budget your channel investments.

Satisfaction trend

A line chart of thumbs-up percentage over time. Daily granularity by default.

Three things to watch:

  • Baseline. Most teams settle at 75 to 90% satisfaction. Below 70% suggests systemic issues.
  • Spikes down. A bad day correlates with a content update gone wrong, a model provider issue, or a sudden volume spike.
  • Long-term drift. Slowly declining satisfaction signals content rot. Schedule re-crawls.

The chart excludes responses without feedback (most). Only thumb-voted responses count toward satisfaction.

Peak hours heatmap

A 7x24 grid showing conversation volume per day-of-week and hour-of-day. Useful for:

  • Staffing decisions. When is your team busiest? Match coverage.
  • SLA configuration. Set tighter SLAs during peak hours.
  • Campaign timing. Launch outbound at low-volume hours to avoid drowning the inbox.

Heatmap cells deepen in color with higher volume. Hover for exact count.

Sources usage

Lists the 25 most-cited documents in the date range. Each row:

  • Document name.
  • Citation count.
  • Average satisfaction for responses citing this document.

Useful for content investment decisions. The top-10 sources usually carry 50 to 70% of citations. Focus content updates there.

CSV export

For deeper analysis (BI tools, custom reports):

  • Export. Click Export on the dashboard toolbar. AskVault generates one CSV covering the current date range — an overview summary, daily query volume, your top 20 questions, and the channel breakdown — and downloads it.

Export available on Business and above. Business+

Real-time vs cached

Every panel queries live conversation data — there's no overnight batch step. Data is cached for 60 seconds per date range, so switching tabs and back within that window shows the same numbers; after 60 seconds, or on a fresh page load, the panels refetch.

For a live view of an individual conversation as it happens, use the Live Chat inbox directly.

Limits

  • Data retention. 365 days, all paid plans including Enterprise.
  • Date range per request. Up to 365 days.
  • Plan availability. KPI cards and query volume on every paid plan (Starter+). The other six panels (top questions, unanswered queries, channel breakdown, satisfaction trend, peak hours, sources usage) require Growth+. CSV export requires Business+.

Common pitfalls

Satisfaction looks low. Customers only vote on responses they care strongly about (positive or negative). 60 to 70% is fine; 30% is concerning.

Unanswered list grows over time. New questions arrive faster than you backfill content. Schedule weekly review and aim to plug 10 to 20 per week.

Channel breakdown skewed by one customer. A high-traffic enterprise customer dominates the chart. Filter by audience or workspace tag to see segment-level data.

Peak hours don't match your assumptions. Often the inbox is busier in evenings than during business hours, depending on your customer mix. Trust the data, not your assumptions.

FAQ

Can I see analytics per channel?

The Channel Breakdown panel shows the per-channel split for the selected date range. There isn't a dashboard-wide channel filter that applies to every other panel today.

Can I pick a custom date range?

Yes. The date-range picker at the top applies to every panel, including a custom start and end date up to 365 days. There isn't a dedicated side-by-side range-comparison view today.

Are agent metrics included?

Per-agent response times and resolution counts are available in the Help Desk's agent reports.

How accurate is the data?

Counts are exact — every panel queries live conversation data, not an overnight batch. Latency is real-time measured. Satisfaction is opt-in from customer feedback.

Was this page helpful?