Find where campaign engagement stops becoming useful outcomes
Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.
A campaign can generate hundreds of replies without producing many completed objectives. The way to locate the gap is to count the same recipient cohort through four distinct stages: delivered, unique response, useful response and completed objective. Calculate each transition separately, then calculate completion from the original delivery base.
The essential distinction is between activity and progression. Ten messages from one recipient are conversation activity, not ten respondents. A polite acknowledgement is a response, but may not advance the campaign objective. An expression of interest is not proof that the objective was completed.
The worksheet below shows how to make those distinctions explicit. It measures one campaign without changing your sales pipeline, follow-up process or onboarding stages.
1. Define one objective that leaves observable evidence
Start with a sentence that says what the campaign is intended to produce and what evidence will count. Avoid objectives such as “increase engagement” or “generate interest”: both can rise while the actual business objective remains unchanged.
For the fictional worked example throughout this article, the campaign invites eligible contacts to book a product demonstration. Its objective is a confirmed booking, not attendance, a sales opportunity or a purchase.
Hypothetical objective: Count one completed objective per recipient when a confirmed demonstration booking is recorded after a useful campaign response and within the defined observation window.
This choice has consequences. A booking request is useful, but it is not yet a completed objective. A confirmed booking counts even if the demonstration takes place later. Attendance would require a different chain and a different observation window.
Choose the evidence before inspecting the results. Otherwise, a disappointing booking count can quietly become a report about “positive conversations.” Record the evidence location as well: for example, a booking record linked to the recipient. A free-text note saying “seems keen” is not equivalent evidence.
2. Fix the cohort, counting units and observation windows
A campaign cohort should be a fixed set of eligible recipients, not everyone who happened to appear in an inbox during the reporting period. Decide membership before measuring progression, and retain the original denominator when results arrive.
Use separate units for separate questions:
- Message events describe sending and conversation volume. One recipient can generate several events.
- Unique recipients describe how many people or recipient identities reached each campaign stage.
- Objective records provide completion evidence. Several records may still represent only one completed recipient in this chain.
Define the recipient key your records can support. If the key is a contact ID, report unique contact IDs rather than claiming you have perfectly deduplicated people. Resolve known duplicates before freezing the cohort, and document unresolved identity limitations.
In the hypothetical worksheet, each recipient can count once at each stage. Several replies or two bookings do not increase that recipient's stage count. Raw message volume remains a separate diagnostic.
The example uses a response window ending seven days after each recipient's first campaign delivery and a completion window ending fourteen days after that delivery. These are hypothetical operating choices, not recommended benchmarks. Select windows that fit the objective, then apply them consistently. Do not compare a fully observed cohort with a newer one that has had less time to complete.
3. Write eligibility rules for every transition
A funnel is only interpretable when each stage has an entry rule. For a strictly nested chain, every recipient at a later stage must also appear in the preceding stages.
- Delivered: At least one designated campaign message has recorded delivery evidence. Preparation or sending alone does not satisfy this rule.
- Unique response: A delivered recipient sends at least one attributable inbound response within the response window. Count the recipient once, including negative responses and opt-outs.
- Useful response: A respondent provides evidence that advances the specified objective. In this example, that means explicitly requesting a demonstration booking or providing scheduling information for one. An acknowledgement or general question alone does not qualify.
- Completed objective: A useful respondent has a linked confirmed booking recorded after the qualifying response and before the completion window closes.
All respondents are eligible for useful-response assessment. All useful respondents are eligible for completion measurement in this example because recipients with an existing booking were excluded before launch.
That last condition matters. If your campaign reaches people who have already completed the objective, either exclude them before launch or report them separately. Removing them only after seeing their responses would change the denominator retrospectively.
Also decide how to handle routes that bypass replies. Someone may receive a message and book directly without responding. That is a valid observed outcome, but it does not belong inside this reply-mediated chain. Report it separately and deduplicate recipients when calculating an all-route completion rate.
4. Fill a cohort worksheet before making a dashboard
Fictional worked example: Every count, time window and result below is hypothetical. The worksheet illustrates a measurement design, not customer performance.
CAMPAIGN COHORT WORKSHEET — HYPOTHETICAL Objective: confirmed demonstration booking Cohort key: campaign ID + unique contact ID Eligible cohort: 1,000 recipients Baseline exclusion: recipients with an existing booking Delivery anchor: first recorded campaign delivery per recipient Response window: delivery through day 7 Completion window: qualifying response through day 14 after delivery Report status: every recipient's observation window has closed STAGE RECIPIENTS ENTRY EVIDENCE Eligible cohort 1,000 Frozen eligibility list Delivered 900 Recorded delivery Unique response 360 Attributable inbound response Useful response 144 Booking request/scheduling details Completed objective 54 Linked confirmed booking SEPARATE ACTIVITY COUNT Inbound message events 620 Not the respondent denominator CONDITIONAL TRANSITION RATES Delivered / eligible 900 / 1,000 = 90.0% Response / delivered 360 / 900 = 40.0% Useful / response 144 / 360 = 40.0% Completed / useful 54 / 144 = 37.5% END-TO-END RATES Useful / delivered 144 / 900 = 16.0% Completed / delivered 54 / 900 = 6.0% Completed / eligible 54 / 1,000 = 5.4% Completed / response 54 / 360 = 15.0% RECONCILIATION Eligible, no recorded delivery 100 Delivered, no response 540 Responded, not classified as useful 216 Useful response, no verified completion 90 Verified completed objective 54 Total 1,000
The reconciliation is a practical error check. These mutually exclusive final positions should add back to the frozen cohort. If they do not, look for duplicate recipient records, missing classifications or inconsistent campaign attribution.
The conditional rates should also reconcile mathematically: 40% responding × 40% useful among respondents × 37.5% completing among useful respondents equals 6% completing among delivered recipients. Use unrounded values in your calculations.
Notice what the activity count cannot tell you. The hypothetical 620 inbound messages do not mean 620 people responded. Dividing 54 completions by 620 message events would mix units and would not measure recipient conversion.
5. Read the loss at each transition, not just the final percentage
The worksheet shows where progression stops, but it does not establish why. Each transition points to a different investigation.
Delivery to response
In the hypothetical cohort, 540 delivered recipients did not respond within the window. That is the largest absolute drop. It makes response generation worth investigating, but it does not prove that the wording was weak or that recipients saw and rejected the message. Delivery evidence and attention are different things.
Response to useful response
Of 360 respondents, only 144 produced a qualifying useful response. The campaign generated conversation more often than objective-relevant intent.
A hypothetical review of the other 216 respondents assigns one mutually exclusive primary reason per recipient: 36 opt-outs, 108 unrelated or acknowledgement-only responses, and 72 clarification-only conversations. These categories explain the composition of the gap. They should not be combined into a vague “engaged” segment.
Useful response to completion
Of 144 useful respondents, 90 have no verified completion. This is the transition to inspect when managers say, “People ask to book, but bookings do not appear.” The measurement task is to verify the outcome records and classify what is observable—not to redesign the entire follow-up operation.
Do not automatically prioritise the lowest conditional rate. Consider the number of recipients affected, the business value of the objective and the reliability of the evidence. The largest gap by volume and the least reliable transition may deserve different investigations.
6. Separate incomplete outcomes from missing evidence
“No verified completion” is not always the same as “did not complete.” It can include an explicit refusal, an unfinished request or a booking that exists but has not been linked to the campaign recipient.
For the hypothetical 90 useful respondents without verified completion, a review finds:
- 45 requests still awaiting confirmation when the window closed;
- 27 recipients who explicitly declined to proceed;
- 18 recipients whose outcome cannot be established from the available records.
The reported completion rate remains 54 out of 144, or 37.5%, because only 54 completions are verified. Do not remove the 18 unknown cases from the denominator to make the rate look better. Instead, show that outcome evidence is unresolved for 18 of 144 useful respondents, or 12.5%.
If evidence later confirms that a booking occurred within the original window, revise the cohort with a dated correction. If a booking occurred after the window, retain the original result and report it as a late completion. This preserves comparability without discarding useful information.
Use the same discipline for classification uncertainty. An ambiguous reply should remain unclassified until reviewed rather than being forced into “useful” or “not useful.” Show classification coverage beside the rate when unresolved cases remain.
7. Prevent attribution and eligibility changes from hiding the gap
Several reporting choices can make a weak transition disappear without improving outcomes.
- Counting every reply: A long conversation inflates engagement. Keep message events separate from unique respondents.
- Using a moving cohort: Adding new recipients to an older campaign mixes observation windows. Freeze membership and version corrections.
- Changing usefulness after review: Broadening the rule to include every friendly reply changes the meaning of the metric. Record the definition and its version.
- Counting every later booking: A booking outside the window does not satisfy the original completion rule. Report late outcomes separately.
- Ignoring overlapping campaigns: One response may be claimed by several campaigns. Apply a documented attribution rule and flag unresolved cases rather than silently crediting all of them.
- Dropping ineligible cases halfway through: If a useful respondent is later found unable to book, retain that fact in the chain. Report the eligibility issue separately unless correcting a demonstrable baseline data error.
A strict nested chain is easy to audit, but it does not capture every route to an outcome. That is its trade-off. Keep a separate all-route completion measure if direct bookings matter, and avoid presenting reply-mediated completion as the campaign's total impact.
These rates describe observed progression under your attribution rules. They do not, by themselves, prove that the campaign caused the bookings. That would require a separate evaluation design.
8. Run one bounded measurement review next
For the next completed campaign, build this worksheet before requesting a new dashboard. Export only the fields needed to establish cohort membership, recipient identity, delivery timing, response timing, useful-response evidence and completion evidence. Limit access to people who need those records for the review.
- Write the objective, stage rules and observation windows on one page.
- Freeze the eligible cohort and identify which recipients have fully elapsed windows.
- Calculate unique stage counts, conditional rates and end-to-end rates.
- Reconcile the final positions to the original cohort.
- Review unresolved classifications and missing outcome evidence.
- Select one transition for a deeper investigation, with a named owner and review date.
The measurable output is a reconciled cohort, a visible count of unknown outcomes and one transition selected for investigation. In the hypothetical example, the team could first reconcile the 18 unknown outcomes before interpreting the useful-response-to-completion gap.
That produces a more useful management question than “Why are conversions low?”: “Of the recipients who explicitly requested a booking, how many have verified completion evidence, and what accounts for the rest?”
Put the next conversation to work.
Explore Alactic Communications for WhatsApp conversations, campaigns and AI-assisted workflows. Choose the product and plan that fit your team.
Explore Communications