Investigate metrics and discrepancies
Objective. Before interpreting any metric, check definition, scope, denominator, timezone, exclusions, freshness and source - so you can reconcile or honestly explain a mismatch and write an escalation packet that a specialist can act on.
The single most common CSM fire drill is 'your number does not match ours'. This lesson gives you a fixed investigation order that resolves most discrepancies without engineering, teaches you to tell a real difference from a definitional one, and has you write an escalation packet precise enough that a specialist does not have to redo your work.
See it in the demo
The customer problem
A customer's analyst pulls Time Champ and their own report side by side and the totals differ. Trust is on the line immediately - if the CSM shrugs or guesses, the customer concludes the data cannot be relied on and the deployment stalls. Most of these mismatches are not bugs; they are two different, each-internally-correct, definitions of the same word. 'Utilisation' with a different denominator, 'output' over a different period or timezone, counts that include or exclude different exclusions. Business impact: unresolved discrepancies become the reason a sponsor refuses to put Time Champ numbers in a board deck, which quietly kills the expansion and renewal case. The skill that prevents this is a disciplined investigation, not faster reassurance.
How Time Champ addresses it - the investigation order
Run the checklist in this fixed order before you interpret or concede anything. (1) Definition: what exactly does each side mean by the metric - is 'utilisation' time-on-work over available hours, or over paid hours? (2) Scope: which teams, clients, processes are included on each side - an unmapped queue (C02) silently drops work. (3) Denominator: the most frequent culprit - attendance %, utilisation % and output attainment have different denominators and answer different questions. (4) Timezone: shifts crossing midnight land in different days across timezones. (5) Exclusions: breaks, leave, non-productive labels, rework - are they in or out on each side? (6) Freshness: is one side today and the other through yesterday? (7) Source: are both reading the same system of record? Use Time and Activity attendance/dashboards and Work Output transactions to locate each parameter; RCM Insights lets you inspect the evidence and scope behind an answer.
What to say / do - reconcile, then write the packet
Talk track to open: 'Before we decide whose number is right, let us line up the definitions - I expect we are measuring two slightly different things, and I want to show you exactly where they diverge.' Then walk the checklist out loud with the analyst. In most cases you will find a definitional difference and reconcile on the spot - document which definition the customer will standardise on going forward. When you cannot reconcile - the numbers still differ after definition, scope, denominator, timezone, exclusions, freshness and source all match - that is a genuine discrepancy and you escalate. Write an escalation packet a specialist can act on without re-interviewing the customer: the two figures, the exact definitions and scope of each, the checklist results (what matched, what did not), screenshots with the parameters visible, the time window and timezone, and your hypothesis. A vague 'customer says numbers are wrong' ticket wastes a day of specialist time.
Reading metrics with context
Half of discrepancy work is teaching the customer that two numbers can both be right because they answer different questions. Attendance % tells you who was present; utilisation % tells you how present time was used; output attainment tells you whether the work got done - a team can be high on one and low on another, and that is information, not error. Carry the product's own interpretation guards: compare two teams only after aligning work type, complexity, period, tenure and quality; recoverable value, billed value, collected cash and capacity value must stay separate; an answer in RCM Insights without suitable data coverage or evidence should not become a customer commitment. Your value here is not making numbers agree - it is making the customer confident they know what each number means.
Objections and honest limits
'Just tell me which number is correct' - resist; often both are correct for their own definition, and the honest answer is to agree a single standard definition going forward. 'This is taking too long, the tool is clearly broken' - the checklist is faster than it feels and usually resolves in minutes once denominators are compared; if it does not, you now have a precise escalation rather than a vague complaint. Honest limits: do not assert a root cause you have not verified; if after the full checklist you cannot explain it, say so and escalate with evidence rather than inventing an explanation. Where the discrepancy may stem from a capture or integration detail you cannot see, write 'confirm with product/specialist' - guessing at engineering internals erodes trust more than saying you will find out.
Interpretation check
A mismatch between two numbers is a definition difference until proven otherwise. Run definition, scope, denominator, timezone, exclusions, freshness and source before concluding the data is wrong.
In the field
If KnackGlobal's finance team sees a cost-per-claim that differs from Time Champ's, the usual cause is denominator and scope - which claims count as 'clean', which payers are in scope - not a tool error; resolving it by agreeing one definition is what lets the number survive in a board deck during the ProHance fight.
Practical task
Take a seeded mismatch (e.g. customer utilisation differs from Time Champ, or output totals differ). Work the seven-check order, document each check's result, and either reconcile it (stating the agreed standard definition) or write a full escalation packet a specialist could act on without contacting the customer.
Self-check
List the seven checks in order and name the one that most often resolves a mismatch.
Definition, scope, denominator, timezone, exclusions, freshness, source. The denominator is the most frequent resolver - attendance %, utilisation % and output attainment use different denominators and answer different questions.
A customer's daily output is consistently lower than yours by the same proportion each day. What two checks jump the queue?
Scope (an unmapped process or client silently dropping a fixed share of work) and timezone/period (a consistent day-boundary offset). A steady proportional gap points to a structural scope or boundary difference, not random error.
After the full checklist, the numbers still differ. What do you produce and what must it contain?
An escalation packet: both figures, each side's exact definition and scope, the checklist results (matched vs not), parameter-visible screenshots, the window and timezone, and your hypothesis - enough for a specialist to act without re-interviewing the customer.
Why is 'just tell me which number is right' often the wrong question?
Both numbers are frequently correct for their own definition. The useful outcome is agreeing a single standard definition and scope going forward, not declaring a winner between two valid measures.
Verified vs demo. The parameters and inspection surfaces (denominators, scope, freshness, RCM Insights evidence) are demo-shown. Whether a specific discrepancy's root cause lies in capture/integration internals is not CSM-verifiable - escalate those with evidence and 'confirm with product'.