Practical guide
AI fluency for demand generation managers
Reconcile fictional campaign events without confusing registrations, unique people and qualified follow-up decisions.
A demand generation manager can use AI to organize campaign records while preserving the difference between events, people and follow-up eligibility. This original exercise reconciles a fictional webinar list. It does not claim actual conversion results or authorize outreach, and it treats qualification as a defined business process rather than an inference from a registration alone.
Define the counting unit
A campaign event is not always a new person or a qualified opportunity.
The fictional export has five rows: person A registers, person B registers, A registers again after correcting a typo, person C attends through an existing invitation, and B downloads a handout. There are five events but three distinct people in this supplied packet.
Ask AI to classify event types and preserve stable person identifiers. Do not infer that every row is a new lead or that downloading a handout establishes a buying intention. The exercise provides no basis for those conclusions.
Reconcile counts before writing the report
A useful summary explains how raw events become the reported totals.
Count registration events, unique registrants and unique attendees separately, using only the information supplied. The packet explicitly records attendance for C but does not state whether A or B attended. Keep their attendance unknown rather than treating registration as attendance.
Document how the duplicate registration from A is handled. Preserve the corrected event history where relevant to the exercise rather than deleting evidence without explanation. AI can help identify likely duplicates, but the explicit identifiers provide the basis for this example's reconciliation.
Keep follow-up criteria separate from engagement
Interest signals do not establish every condition for a next action.
The exercise owner says follow-up requires an approved contact route and a relevant stated request. The packet contains no such request for A or B. Do not write send-ready messages or treat the event count as permission to contact them. Identify the missing process evidence instead.
This is an illustrative operating boundary, not a legal rule about outreach. Real campaigns need the organization's applicable permissions and policies. The manager's task here is to avoid silently converting participation data into authorization or qualification.
Write a bounded campaign note
The report should state what happened and what remains unknown.
A useful note distinguishes the five recorded events, three people and incomplete attendance information. It can request a verified attendance export and the approved follow-up criteria. It should not report a conversion rate when the necessary numerator, denominator or outcome definition is absent.
Ask AI to draft a headline, then check whether it overstates pipeline impact. Preserve uncertainty in the main summary rather than burying it in an appendix. A precise-looking percentage built from the wrong event units is less informative than a clear statement of the missing data.
Test the reconciliation with a later event
A new record should update the relevant measure without changing unrelated history.
Add a fictional confirmed attendance event for B. Update unique attendees to include B and C, while keeping registration history distinct. Do not increase unique people merely because B produced another event. Recheck any measure whose denominator depends on the updated attendance set.
Deliver the event map, counting rules and unresolved follow-up questions. Review identity handling and interpretation within this packet. The customer operations guide examines support-event measures; this resource concerns campaign participation, deduplication and the boundary between engagement evidence and a follow-up decision.
Sources and scope
The references below are background reading, not evidence that this fictional exercise has been validated or endorsed.
The case details, calculations and suggested review questions are original instructional material. Use them to discuss observable work, not to infer customer outcomes, professional credentials or performance in every setting. Before adapting the exercise, confirm the relevant facts, approved tools, data permissions and decision owners. If you change the case, revisit the expected answers and checks as well. These examples describe practice tasks, not a promise that a particular product includes the fictional features.
Sources: [1] [2]
Sources
- 1.AI RMF Core · NIST
- 2.AI foundation skills for work benchmark · Skills England