Real-Time Adherence - certification assessment
Objective. Test practical mastery of Real-Time Adherence for the certification gate: the learner must work realistic scenarios, keep the three adherence layers separate, treat a red state as a prompt rather than a verdict, validate the plumbing before trusting a figure, and state the honest limits out loud. Scored to the gate (>=80, no critical error, manager review).
This is the practical gate for RTA. The learner works two scenarios - a sales demo-and-objection and a CSM dispute-and-validate - and is scored against a checklist. The critical errors are the ones that would mislead a real customer: presenting presence as productivity, treating a red state as a verdict, presenting the directional output layer as substantiated, or arguing a disputed number without validating the rules.
See it in the demo
How this assessment is scored
Tie to the certification gate from grounding: a practical score of 80 or above, no critical error, and manager review. The learner works both scenarios below out loud or in writing, as they would with a real customer. Score the checklist in the next section; any single critical error (listed later) fails the assessment regardless of total score, because each would mislead a customer in a way that damages trust or the deal. A manager reviews the recording or written work before certification is granted.
Scenario A (Sales) - the confident prospect
A WFM manager at a 2,000-seat contact centre says: 'We already run real-time adherence in our current WFM and we sit at 97 percent. Honestly I am not sure what you add. Show me in five minutes and tell me what is real versus marketing.'
The learner must run a short evidence-led demo that: opens on a floor-leader question, shows the three layers on the Control Room, reveals a schedule-perfect-but-below-output agent, shows a thinning interval while it is still fixable, states the displacement wedge (incumbent proves presence; RTA proves presence became work at standard), and states the honest limit out loud. Then handle the implicit objection ('97 percent, we're fine') by separating the layers, and the likely follow-up ('is this just surveillance?').
Scenario B (CSM) - the disputed figure and the write-up
A customer team lead escalates: 'Your tool says my best agent was only 70 percent adherent yesterday and flagged her output as below standard. She is my strongest person. Either the tool is wrong or I am about to have a very unfair conversation with her.'
The learner must: refuse to argue the number first; run the four plumbing checks (published-schedule version, time zone, state mapping, freshness) and go to the rules (grace windows, thresholds) as the first stop; identify which layer the 70 percent reports; check whether a red schedule state was a legitimate reassignment and whether the below-output read reflects harder items, a blocker, an unvalidated standard or thin capture; and state plainly that the output layer is directional until the standard and capture are validated, so it must not drive a write-up. Close with what to validate, what to escalate to the specialist, and the fair message to the lead.
Scored checklist - what a correct answer contains
Score each item; weight the guardrail items most.
Scenario A (50 points)
- Opens on a customer question, not a feature tour (5)
- Shows the three layers and names which question each answers (10)
- Lands the schedule-green-but-below-output twist (10)
- Frames the displacement wedge correctly - incumbent proves presence, RTA proves work at standard (10)
- Shows intraday value (fixable now vs tomorrow's report) (5)
- States the honest limit out loud - seeded data, directional output, confirm latency/sources with product (10)
Scenario B (50 points)
- Does not argue the number first; goes to rules and the four plumbing checks (10)
- Correctly identifies which layer the figure reports (10)
- Treats the red state as a prompt - checks for reassignment and context (10)
- Treats below-output as a prompt - checks standard, capture, complexity, blocker (10)
- States the output layer is directional until validated and must not drive a write-up; names the escalation and the fair message (10)
Critical errors (any one fails the assessment)
These fail regardless of total score because each would mislead a real customer:
- Presenting presence as productivity - letting the prospect leave reading schedule adherence (or a green board) as proof the work is getting done.
- Treating a red state or below-output as a verdict - recommending a write-up without checking for a legitimate reassignment, harder work items or an unvalidated standard.
- Presenting the output layer as substantiated when the standard or capture is not validated - quoting a directional number as a performance fact.
- Arguing a disputed figure without validating the rules and inputs - debating the number before checking published-schedule version, time zone, state mapping and grace/thresholds.
- Asserting an integration, latency or security/compliance specific as fact instead of 'confirm with product'.
- Collapsing the three layers into one 'adherence' score anywhere in either scenario.
Interpretation check
The assessment passes only if the learner keeps schedule, work and output adherence as three separate questions throughout, treats every red or below-output signal as a prompt to look with context rather than a verdict, validates the rules and inputs before trusting or disputing a figure, and labels the output layer directional until the standard and capture are validated.
In the field
Scenario A mirrors a KnackGlobal-style displacement (RCM, ~6,000 seats) where the incumbent proves presence and RTA's work and output layers prove the presence became work at standard on time-sensitive queues. Scenario B mirrors the OVO effort-to-outcome reality where a strong agent can read below output for legitimate reasons - the fair, context-first handling is exactly what the gate tests.
Practical task
Complete both scenarios for manager review. For Scenario A, record a five-minute demo-and-objection run on rta.demo.timechamp.io that hits every Scenario A checklist item. For Scenario B, write the dispute workup: the four plumbing checks and rules you validated, which layer the figure reports, the context checks on the red state and below-output read, your directional-output statement, the escalation, and the fair message to the lead. Self-score against the checklist and confirm you committed none of the critical errors.
Self-check
ASSESSMENT ITEM (Sales): The prospect says '97 percent adherent, we're fine.' Give the model answer.
97 percent is schedule adherence - presence and state only. It answers 'are they on plan', not 'is the plan producing work'. Show the work and output layers for the same scope, reveal a schedule-perfect agent below the output standard, and ask whether their incumbent can prove scheduled time became work at standard. Do not let them equate the 97 percent with productivity - that is the core guardrail.
ASSESSMENT ITEM (CSM): The lead wants to write up an agent flagged red and below output. Give the model answer.
Do not act on the flag as a verdict. First validate the rules and four plumbing checks (published schedule, time zone, state mapping, freshness, grace/thresholds). Identify which layer the figure reports. Check whether the red state was a legitimate reassignment and whether below-output reflects harder items, a blocker, thin capture or an unvalidated standard. State that the output layer is directional until validated, so it must not drive a write-up; escalate plumbing to the specialist; give the lead a fair, context-first message.
ASSESSMENT ITEM: Name three of the critical errors that fail this assessment.
Any three of: presenting presence as productivity; treating a red state or below-output as a verdict and recommending a write-up; presenting the directional output layer as a substantiated fact; arguing a disputed figure without validating the rules/inputs; asserting an integration, latency or security specific as fact; collapsing the three layers into one score.
ASSESSMENT ITEM: State the honest limit you must say out loud in the sales scenario.
The model is real but runs on seeded demo data, so these figures and the refresh speed are not the customer's environment; in production the output layer is only as strong as their output standard and capture coverage and is directional until validated; real latency and supported schedule/activity/output sources are confirmed with product.
ASSESSMENT ITEM: A customer disputes an adherence number. What is the single first action that marks a passing CSM answer, and what marks a failing one?
Passing: go to the adherence rules and the four plumbing checks before touching the number, because adherence is the output of configurable rules and inputs. Failing: arguing or defending the specific figure first, which would be a critical error - no floor review fixes a definition or feed error at the source.
Verified vs demo. Both scenarios are worked on seeded demo data at rta.demo.timechamp.io, which supports the model but not customer-specific figures. A passing answer never presents a seeded number, latency or integration as a production fact and always routes those to 'confirm with product'. Certification requires >=80, no critical error, and manager review.