Analytics Dashboard : Metrics Guide & FAQ

Last updated: November 11, 2026

What this guide covers

This guide explains how every metric on the AiPrise Analytics Dashboard is calculated, and why the numbers you see there can differ from what you see in Case Management. It is written for support teams and can be shared with customers.

Data freshness. The dashboard shows a daily snapshot, refreshed once per day at 00:00 UTC (the "Production data as of" stamp in the top-right corner shows the snapshot time). Case Management is live. This single difference explains most day-to-day discrepancies — see the FAQ below.

Filters. Every number on the dashboard reflects the filters currently applied:

Filter

What it does

Type

Combined, KYC only, or KYB only

Count by

The counting unit: Sessions (verification runs) or Profiles (distinct entities). Switching changes every count on the dashboard

Associated with

Business Profiles, User Profiles, and/or Independent sessions

Date

The date range; sessions are attributed to dates in UTC

Countries

Limits to sessions from selected countries

Templates

Limits to sessions run through selected verification templates

If a number looks wrong, the first thing to check is which filters are active — two people comparing numbers with different filters will not match.

Core concepts: what the dashboard counts

The dashboard counts in the unit chosen in the Count by control: Sessions by default, or Profiles (distinct entities). The definitions in this guide are written for Sessions; with Profiles selected, read counts as profiles. Understanding the three layers of the data model is essential for interpreting any number — and for understanding discrepancies with Case Management.

Term

Definition

Verification type

KYC (an individual) or KYB (a business). A customer may use one or both.

Check

A single test inside a verification: address check, document check, ID check, AML screening, PEP screening, business license, etc. Checks never appear in headline counts — they explain why a session was declined, not how many verifications ran.

Session (case)

One verification run. A session bundles several checks. "Case" in Case Management and "session" in Analytics refer to the same object.

Profile

Groups multiple sessions that belong to the same entity. Example: one company runs proof-of-address, business-license, and business-details verifications — that is 3 sessions under 1 profile. Profiles exist for both KYC and KYB, and the profile feature is enabled per customer.

The key implication: the dashboard counts sessions by default; switching the Count by control to Profiles counts distinct entities instead. A business that went through 3 verification runs contributes 3 to "Total sessions," even though it is 1 profile in Case Management. This is the single most common source of "the numbers don't match" questions.

Overview tab — metric definitions

All Overview metrics respect the active filters (type, date range, countries, templates, associated-with). The calculations below are written for Count by = Sessions; with Count by = Profiles, the same metrics count distinct profiles instead.

Metric

How it is calculated

Notes

Total sessions

Count of all verification sessions created in the selected date range

Includes sessions still in progress or abandoned

Completed sessions

Count of sessions that finished the verification flow

Shown as "X of Y total sessions." The gap between total and completed is sessions that were started but never finished

Approval rate

(Approved sessions ÷ completed sessions) × 100

The denominator is completed sessions, not total. A neutral metric — higher is not automatically better in a risk context

Decline rate

(Declined sessions ÷ completed sessions) × 100

Same denominator as approval rate

In review

(Sessions awaiting a decision ÷ completed sessions) × 100

A session can be completed by the applicant but still awaiting a decision (human review or external input)

Outcome mix

Approved + In Review + Declined + Unknown as shares of completed sessions

The four shares sum to 100%. "Unknown" is a small residual where no outcome was recorded, generally sessions with invalid input.

Daily volume by outcome

Count of sessions per day, stacked by outcome

Sessions are bucketed by their creation date in UTC. Colors are consistent everywhere: green = approved, amber = in review, red = declined

Country distribution

Share of sessions per country, shown separately for KYC and KYB

"Unknown" means the session had no country recorded (e.g., abandoned before that step)

Volume per document type

Count of documents collected, by document type, split KYC / KYB

Counts documents, not sessions. One session can collect several documents, so these numbers will not sum to the session totals

Worked example (round numbers): 2,000 total sessions in the date range, of which 1,500 completed (500 were started but never finished).

  • Approved: 750 → approval rate = 750 ÷ 1,500 = 50%

  • In review: 450 → in-review = 450 ÷ 1,500 = 30%

  • Declined: 300 → decline rate = 300 ÷ 1,500 = 20%

The three outcomes add up to 1,500 completed sessions and 100% of the outcome mix. Note none of the rates use the 2,000 total — dividing by total sessions instead would give 37.5% / 22.5% / 15%, which is the most common way customers arrive at "different" numbers.

Trends tab — metric definitions

The Trends tab shows movement over time at a selectable interval (Daily, Weekly, Monthly). All series are counted in the unit set by the Count by control — Sessions by default, or Profiles.

Metric

How it is calculated

Notes

Volume trend

Count of sessions per interval, as separate KYC and KYB lines

Sessions are bucketed by creation date in UTC. A daily interval makes single large batches look like spikes — switch to Weekly or Monthly for slope

Approval rate trend

Approved ÷ completed sessions × 100, per interval, per type

On low-volume days this is very jumpy: 1 approval out of 1 completed session = 100%. Read daily rate lines together with the volume chart

Decline rate trend

Declined ÷ completed sessions × 100, per interval, per type

Same caution as approval rate — a 100% decline day may be a single session

Reading rate trends correctly: a rate computed on an interval uses only that interval's sessions. The Overview approval rate (whole date range) is a weighted average, so it will generally not equal the average of the daily points on the Trends chart — this is expected, not a bug.

Risk tab — metric definitions

Metric

How it is calculated

Notes

Monitoring coverage

Sessions under ongoing monitoring ÷ completed sessions with a known result × 100

The denominator excludes sessions whose result is unknown (e.g., 104 of 1,346 = 7.73%). Measures how much of the verified population is enrolled in continuous monitoring

Session failed rate

Failed sessions ÷ total sessions × 100

Note the different denominator: total sessions, not completed (e.g., 369 of 2,070 = 17.83%). "Failed" means the session could not complete — technical failure or abandonment — which is different from "declined" (a completed session with a negative decision)

Top declines by country

Count of declined sessions per country, top 10, tagged KYC or KYB

Absolute counts, not rates. A country with high volume will naturally show more declines — always read alongside that country's share in Country distribution before concluding it is riskier

Top warnings

Count of warning signals raised, top 10 by signal type, tagged KYC or KYB

Warnings are soft signals that did not block a decision (e.g., AML Match, Birth Date Mismatch, Selfie Video Missing). A session can raise several warnings and still be approved, so warning counts exceed session counts and do not equal declines

Top warnings by country

Count of warning signals per country, top 10

Same reading rule: absolute counts — interpret against country volume

Declined vs failed vs warned — three different things:

  • Declined — the session completed and the decision was negative.

  • Failed — the session never reached a decision (abandoned, technical error).

  • Warning — a check raised a signal, but the session could still end approved.

Why Analytics and Case Management can show different numbers

Both surfaces read the same underlying verification data, but they answer different questions at different moments. When a customer says "the dashboard doesn't match my case list," it is almost always one of these seven causes — in rough order of frequency:

  1. Snapshot vs live. Analytics refreshes once a day at 00:00 UTC; Case Management is real time. Anything created or decided after the snapshot (sessions run today, a case an analyst approved an hour ago) is in Case Management but not yet in Analytics. It appears after the next refresh.

  2. Sessions vs profiles (cases vs entities). Analytics counts sessions. In Case Management, a profile-enabled customer often looks at profiles — entities. One business that ran 3 verifications is 3 sessions in Analytics but 1 profile in their case view. Neither number is wrong; they count different units.

  3. Filters. The dashboard numbers reflect the active date range, type (KYC/KYB), countries, templates, and associated-with selections. A Case Management list usually has its own, different filters. Align both filter sets before comparing.

  4. Date attribution and timezone. Analytics buckets sessions by creation date in UTC. A session created at 11 pm local time can land on a different calendar day than the user expects, and "this month" in a local-time case view is not the same window as "this month" in UTC.

  5. Status changed after the snapshot. A case that was in-review at snapshot time but approved this morning still counts as in-review in Analytics until the next refresh. Overturned or re-opened decisions shift counts between buckets the same way.

  6. Different denominators. Rates on the dashboard use specific denominators (completed sessions for approval/decline/in-review; total sessions for failed rate). Manually dividing counts from a case list by a different base will give a different percentage.

  7. Sessions that never become visible cases. Abandoned or failed sessions are in Analytics' "Total sessions" but may never surface in a review-oriented case queue, so Total sessions is usually higher than any case-list count.

Rule of thumb for support: differences that disappear after the next daily refresh are cause 1 or 5. Persistent differences are almost always 2 (unit of counting) or 3 (filters).

FAQs

  • Q: I approved a case an hour ago. Why hasn't the dashboard changed?

    Analytics is a daily snapshot, refreshed at 00:00 UTC (see the "Production data as of" stamp top-right). Your decision will be reflected in the next refresh. Case Management always shows the live state.

  • Q: The dashboard says 2,070 sessions, but I only see ~1,800 entities in Case Management. Which is right?

    Both. Analytics counts sessions (verification runs); your case view may be counting profiles (entities). One entity can have several sessions — e.g., a business that ran proof-of-address, business-license, and business-details checks is 3 sessions but 1 profile. To compare like for like, switch the dashboard's Count by control to Profiles — it will then count distinct entities.

  • Q: Why don't Approved + In Review + Declined add up to Total sessions?

    Those three outcomes are shares of completed sessions, not total. The difference between total and completed sessions is runs that were started but never finished. A small "Unknown" share also exists where no outcome was recorded.

  • Q: Approval rate looks different depending on where I calculate it. Why? Check the denominator. The dashboard divides approved by completed sessions. If you divide approved by total sessions from a case export, you will get a lower number. Neither is wrong — they answer different questions.

  • Q: Why does the daily approval rate chart swing between 0% and 100%?

    Low volume. A day with one completed session is either 100% or 0%. Read the rate chart together with the volume chart, or switch the interval to Weekly or Monthly.

  • Q: What's the difference between "Declined" and "Session failed"?

    Declined = the verification completed and the decision was negative. Failed = the session never reached a decision (the applicant dropped off, or a technical error). They are counted separately and against different denominators.

  • Q: Why do warnings outnumber declines by so much?

    Warnings are soft signals that did not block the decision. One session can raise several warnings (an AML match and a birth-date mismatch) and still be approved. Warnings explain where friction and risk signals appear, not how many sessions were refused.

  • Q: Italy shows the most declines — is Italy our riskiest market?

    Not necessarily. "Top declines by country" shows absolute counts. A country with more volume produces more declines at the same decline rate. Compare the decline count against that country's share of total volume before concluding.

  • Q: Why does a session show "Unknown" country or outcome?

    The session ended before that data point was captured — typically an abandoned or failed run. These appear in totals but are excluded from some rate denominators (e.g., monitoring coverage excludes unknown-result sessions).

  • Q: The counts changed for a past date — can historical numbers move?

    Yes, slightly. If a case is overturned, re-opened, or receives a late decision, it moves between outcome buckets at the next refresh. The session stays attributed to its original creation date, so a past date's outcome mix can shift.

  • Q: My "this month" numbers don't match my own report for this month.

    Analytics buckets by creation date in UTC. A report built in your local timezone draws the month boundary at a different moment, so sessions near midnight land in different months. Also confirm the same date range, type, country, and template filters are applied.

  • Q: Does the dashboard count checks?

    No. Headline counts are sessions. Checks (ID check, AML screening, address check) sit inside a session and surface only as explanations — decline reasons and warnings. Document counts in "Volume per document type" are similarly per-document, not per-session.