B2B agencies often measure a client funnel with more labels than agreements. One team calls a form completion a lead, another calls the same record an inquiry, and sales uses a different rule for acceptance. A report can then show a precise rate while no one can explain what moved, which records belong in the denominator, or what decision the number should change.
This template is for an agency and client team that need a shared measurement contract. It defines a stage, the evidence required to enter and leave it, the owner of the definition, and the next action when the evidence is weak. It does not create an attribution truth, a benchmark conversion rate, or a promise of pipeline.
Start with the decision boundary
Write the decision before choosing events or formulas:
> Given the evidence for [scope and period], should we continue, change, investigate, or stop [specific funnel motion]?
Name the motion: a service page, paid route, content path, partner source, or campaign family. State the audience, geography if material, lifecycle scope, source systems, and data that is late or excluded. “Measure the B2B funnel” is too broad. “Decide whether the agency’s integration-content route receives another bounded test after reviewing accepted inquiries” is measurable and reversible.
The decision owner should be able to change the next action. If a report has no accountable decision maker, it is an observation log, not a decision template.
The funnel measurement contract
Use one row per stage. Do not allow a stage name to carry meaning that exists only in someone’s memory.
| Field | Completion guidance | Evidence required | |—|—|—| | Stage name | Use a short label and a stable ID | Versioned definition | | Business question | State what this stage helps decide | Named decision owner | | Entry criteria | What must be true before a record enters? | Source and field or review rule | | Exit criteria | What makes the next stage appropriate? | Observable evidence and timestamp | | Owner | Who can accept, reject, or correct the record? | Role and escalation route | | Denominator | Which records are eligible for the rate? | Inclusion and exclusion rule | | Measurement source | Where is the observation captured? | System, extraction date, and access boundary | | Latency | What delay can occur before the stage is reliable? | Expected lag and reconciliation rule | | Alternative explanation | What else could produce the same signal? | At least one competing cause | | Action | What will change if the signal is weak or strong? | Reversible next step and stop rule |
The contract is complete only when another analyst can reproduce the row without asking which spreadsheet or person is authoritative.
Define stages by decisions, not platform events
A useful agency funnel may include stages such as:
- Reachable context: the intended account or audience could encounter the message.
- Meaningful interaction: the defined content or route interaction occurred.
- Inquiry: a response met the agreed minimum information rule.
- Accepted inquiry: the receiving team reviewed it and accepted the agreed fit boundary.
- Commercial conversation: an accountable owner confirmed a next conversation or evaluation step.
These are example labels, not a universal funnel. A client may need fewer stages or a different order. The important distinction is between an event and a business state. GA4 Events documentation describes an event as a measured interaction or occurrence. It does not define the client’s sales acceptance rule or make an event equal accepted demand or revenue.
Write the stage contract in plain language. For example:
“text Accepted inquiry = a response reviewed by the named owner that meets the agreed account, problem, and permission conditions. A form event alone does not enter this stage. “
Denominators and lag
Most agency disputes are denominator disputes. For each rate, state:
- the eligible population;
- the inclusion and exclusion date;
- duplicate and test-record handling;
- whether the rate is cohort-based or period-based;
- which records are still open or late;
- whether the numerator and denominator come from the same source and definition.
If sales acceptance can arrive after the reporting period, label the result provisional rather than hiding the lag. An incomplete cohort may be useful for an operational decision, but it should not be presented as a final rate.
The GOV.UK guidance on measuring success is a useful process reference because it starts with a question and combines evidence types. It does not provide a B2B-agency funnel formula or benchmark.
Evidence levels for a funnel decision
Use a confidence note beside each metric:
| Level | What is established | What remains unknown | |—|—|—| | Observed | An event, record, or review action was captured | Whether it represents fit or intent | | Defined | The stage and denominator are documented | Whether data is complete and timely | | Reconciled | Two owners agree on sample and definitions | Whether the pattern generalises | | Decision-ready | Evidence is sufficient for the bounded action | Commercial impact beyond the decision |
Do not turn observed into decision-ready by adding decimal places. The NIST Information Quality Standards provide a useful lens for utility, integrity, objectivity, and correction history; they do not certify an agency’s data.
Hypothetical worked example
The example is illustrative and does not describe a client or benchmark.
| Field | Example completion | |—|—| | Decision in scope | Decide whether to repeat a technical-content route for operations leaders | | Stage under review | Accepted inquiry | | Entry criteria | Response is reviewed by the named agency-client owner and meets account, problem, and permission rules | | Denominator | All inquiries received through the selected route during the cohort window, excluding tests and duplicates | | Change observed | Several inquiries reference the technical problem; acceptance timing is delayed | | Confidence | Medium: the response sample is reviewed, but late dispositions are not complete | | Alternative explanation | The topic attracts researchers outside the account boundary; routing lag changes apparent acceptance | | Next action | Reconcile a bounded sample, tighten the audience exclusion, and review again before expansion | | Stop rule | Hold if owner, definition, or permission remains unresolved |
The record does not claim that the route generated pipeline. It shows what can be checked next and who can make the decision.
Build the review around the template
Run the review in six passes:
- Scope: confirm motion, audience, period, and excluded sources.
- Definitions: read each stage contract and flag changed versions.
- Data: check source, timestamp, duplicate, test, and permission handling.
- Cohort: reconcile numerator, denominator, open records, and lag.
- Interpretation: compare at least one alternative explanation.
- Decision: choose continue, change, investigate, or stop with an owner.
Keep the agency and client roles explicit. The agency may own instrumentation or analysis; the client may own sales acceptance or regulated-data decisions. Neither role should silently inherit the other’s authority.
When to connect advertising and analytics data
If the agency connects advertising and analytics systems, document which source is used for which question. Google’s guidance on creating conversions from Analytics events in Google Ads distinguishes measurement across website or app activity from Google Ads-only conversion measurement. The implementation may be useful, but the choice does not prove incremental impact or make two systems’ definitions identical.
Record consent and data-use boundaries before joining sources. Never use an account label, event parameter, or inferred identity as a substitute for an agreed qualification rule.
Common failures and fixes
| Failure | Why it fails | Correction | |—|—|—| | One “lead” stage for every client | Fit and acceptance differ by engagement | Version a client-specific stage contract | | Rate without denominator | The number cannot be reproduced | Add eligible population and exclusions | | Conversion event called revenue | Event and commercial state are different | Separate observation, acceptance, and outcome | | Late dispositions hidden | Open cohorts look final | Show lag and provisional status | | Agency owns every definition | The client cannot accept the handoff | Assign decision rights to the owner of the state | | More tracking solves a semantic gap | More events do not clarify meaning | Resolve the decision and definition first |
Advertising language also needs a proof boundary. For a U.S.-specific prompt, the FTC advertising and marketing reference can orient a review; the agency and client still need the rules and evidence applicable to their market.
Copy-ready decision template
“text Review date: Client / motion: Decision owner: Decision in scope: Audience and exclusions: Reporting period and cohort rule: Stage name and version: Business question: Entry criteria: Exit criteria: Owner and escalation route: Denominator and exclusions: Measurement source and extraction date: Latency / provisional status: Observed change: Alternative explanation: Data-quality or permission gap: Evidence level: Next reversible action: Continue / change / investigate / stop rule: Next review trigger: “
The template is useful when the client can challenge a definition, reproduce the denominator, see the lag, and understand exactly what action follows. That is more valuable than a funnel report that looks precise but cannot support a decision.
Sources and limits
This template uses GA4 Events, Google Ads conversion import guidance, GOV.UK measuring success, NIST Information Quality Standards, and FTC Advertising and Marketing as process references. It does not establish a funnel benchmark, attribution truth, revenue result, consent decision, or legal conclusion.
How did this article land?
Choose one reaction. You can change it anytime.
