Back to all articlesCRM measurement

Decide Whether to Count People, Accounts or Opportunities

Alactic Editorial10 min read

Prepared with AI assistance by Alactic Editorial. Examples identified as hypothetical are illustrative, not customer results.

A sales report shows nine enquiries, six contacts, four businesses and four opportunities. Those totals can all be correct. The inconsistency begins when someone labels all four measures “leads”, or adds channel-level contact counts and expects them to match the overall total.

Choose the counting unit from the business question: count enquiries for inbound demand events, people for individual participation, accounts for business coverage and opportunities for distinct sales attempts. Then specify the reporting period, qualifying records and relationship rules. A distinct count removes repeated IDs within its scope; it does not make separate scopes safe to add.

The hypothetical worksheet below shows how to put those rules into practice. It addresses analytical counting grain, not duplicate merging, identity resolution or how to structure a sales pipeline.

1. Start with the question, not the available total

Before building a dashboard, finish this sentence: “We want to know how many what, meeting which conditions, during which period?” The missing noun determines the counting unit, also called the grain.

  • Enquiries: How many separate inbound requests did we receive? A person can make several requests.
  • People: How many individuals made at least one qualifying enquiry? Several individuals can represent the same business.
  • Accounts: How many businesses had at least one qualifying enquiry? One business can have several people and sales attempts.
  • Opportunities: How many distinct sales attempts were associated with qualifying enquiries? Several enquiries can support one opportunity, and one account can have several opportunities.

None is the universally correct number. An operations manager planning enquiry handling needs a different measure from a sales manager assessing business coverage.

A useful metric name preserves that distinction. “Accounts with an enquiry in the reporting month” is more informative than “active accounts”. “Opportunities linked to this month’s enquiries” is not the same measure as “opportunities created this month”. The wording should expose the selection rule rather than hide it.

2. Declare the relationships before counting

For the fictional worked example, assume each enquiry belongs to one person and one account. Each enquiry links to either one opportunity or none. An opportunity belongs to one account but can involve several people. A person can submit several enquiries and participate in more than one opportunity.

These are worksheet assumptions, not universal CRM rules. If your business supports enquiries involving several accounts or several opportunities, use a separate relationship list and count each entity by its own ID. Do not silently copy the enquiry into multiple rows and then treat those rows as separate enquiries.

Also decide what an enquiry represents. In this example, it is a separately logged inbound request, not every message in a conversation. A follow-up message does not create another enquiry unless it meets the team's agreed definition of a new request.

The reporting scope is a hypothetical month called Period M. All nine enquiries below occurred within that period. The example uses relationships as recorded at the reporting cut-off. It does not claim that linked opportunities were created or closed during Period M.

That final distinction matters: an enquiry date selects the enquiry cohort. It does not automatically select an opportunity-creation cohort.

3. Use this filled hypothetical counting worksheet

Fictional worked example: all identifiers, relationships and totals in this worksheet are hypothetical. Web and WhatsApp are simply recorded enquiry channels; the example makes no claim about a product integration.

Entity register

Accounts
A1 = North Workshop
A2 = Cedar Supply
A3 = Harbor Services
A4 = Elm Studio

People and their account in this example
P1 = Alex, A1
P2 = Blair, A1
P3 = Casey, A2
P4 = Drew, A2
P5 = Ellis, A3
P6 = Finley, A4

Opportunities and their account
O1 = A1, equipment purchase
O2 = A1, maintenance contract
O3 = A2, initial order
O4 = A2, expansion order

Enquiry ledger: one row per enquiry

Enquiry | Person | Account | Opportunity | Channel
E1      | P1     | A1      | O1          | Web
E2      | P1     | A1      | O1          | WhatsApp
E3      | P2     | A1      | O1          | Web
E4      | P1     | A1      | O2          | WhatsApp
E5      | P3     | A2      | O3          | Web
E6      | P4     | A2      | O3          | WhatsApp
E7      | P3     | A2      | O4          | Web
E8      | P5     | A3      | None        | WhatsApp
E9      | P6     | A4      | None        | Web

Reconciled worksheet totals

Measure                         | Counting rule                  | Result
Enquiries                       | Distinct enquiry IDs           | 9
People making enquiries         | Distinct person IDs            | 6
Accounts making enquiries       | Distinct account IDs           | 4
Linked opportunities            | Distinct non-empty opportunity | 4
Enquiries linked to opportunity | Enquiry IDs with a link        | 7
Enquiries without opportunity   | Enquiry IDs without a link     | 2
People linked to opportunity    | People on linked enquiries     | 4
Accounts linked to opportunity  | Accounts on linked enquiries   | 2

The enquiry reconciliation is seven linked plus two unlinked, giving nine. The people reconciliation is four with a linked opportunity plus two without one, giving six. The account reconciliation is two with a linked opportunity plus two without one, giving four.

Those last two additions work here because the groups are mutually exclusive in this particular dataset. In a different dataset, a person or account might have both linked and unlinked enquiries. Then the two groups would overlap, and adding their distinct counts would overstate the total.

4. Write aggregation rules beside each measure

A worksheet becomes reusable when it includes a short metric contract. Record the entity ID, qualifying conditions, date field, relationship cut-off, missing-value treatment and permitted breakdowns.

For the hypothetical example, a completed contract for account coverage would read:

Metric: Accounts with enquiries in Period M
Unit: Account
Key: Account ID
Population: Accounts on qualifying Period M enquiries
Date selection: Enquiry received date
Relationship basis: Links recorded at reporting cut-off
Calculation: Distinct count of non-empty account IDs
Missing account: Exclude from account count; report separately
Channel breakdown: Account can appear in several channels
Grand total: Recalculate distinct accounts across all channels
Do not: Sum channel-level distinct account counts

Missing relationships need explicit treatment. A blank opportunity link contributes no opportunity to the distinct count, but its enquiry still counts. A blank account link should not become a fictional shared account called “Unknown”. Report the number of enquiries missing account links separately so readers can see what the account measure excludes.

Also distinguish zero from missing. “No linked opportunity at cut-off” can be an intentional state. A missing person ID may instead prevent a reliable people count. Neither should disappear from the underlying enquiry reconciliation.

Use the same discipline for time. Monthly enquiry totals can be added across non-overlapping months if each enquiry belongs to exactly one month. Monthly distinct people usually cannot: someone who enquired in both months must count only once in the combined-period people total.

5. Show exactly where distinct counts stop being additive

Channel reporting makes the problem visible. Applying the worksheet's rules produces these hypothetical results:

Channel  | Enquiries | People | Accounts | Linked opportunities
Web      | 5         | 4      | 3        | 3
WhatsApp | 4         | 3      | 3        | 3
Overall  | 9         | 6      | 4        | 4

The enquiry column adds correctly because each enquiry has exactly one recorded channel. The other columns overlap.

  • People: Web includes P1, P2, P3 and P6. WhatsApp includes P1, P4 and P5. Adding four and three gives seven appearances, not seven different people. P1 appears in both; the union contains six people.
  • Accounts: Web includes A1, A2 and A4. WhatsApp includes A1, A2 and A3. Three plus three gives six channel memberships. There are only four distinct accounts overall.
  • Opportunities: Web links to O1, O3 and O4. WhatsApp links to O1, O2 and O3. O1 and O3 appear in both channels, leaving four distinct opportunities overall.

For two groups, the reconciliation is the first distinct count plus the second distinct count minus the distinct intersection. With more groups, use the full union of IDs rather than improvised pairwise subtraction, which can mishandle entities appearing in several groups.

The same failure occurs when adding counts across owners, products, campaigns or opportunity statuses if entities can belong to several groups. A dashboard total should recompute the distinct set for its full scope, not add the visible rows.

If management needs additive channel credit, define a separate allocation measure, such as assigning each person to their earliest qualifying enquiry channel in the period. That makes categories exclusive, but it answers “assigned channel” rather than “channels used”. Keep both the label and the trade-off visible.

6. Prevent relationship joins from multiplying totals

Even correct source records can produce incorrect totals when reporting joins change the row grain. In the worksheet, O1 appears on three enquiry rows. Counting opportunity-bearing rows therefore counts three links for O1, not three opportunities.

The safer design is to build each measure at its intended grain before combining summaries. Count enquiries from the enquiry ledger. Count opportunities from the distinct linked opportunity IDs. If opportunity-level values are needed, select each eligible opportunity once before summing those values.

People-per-opportunity counts reveal a related trap:

Opportunity | Distinct people on its enquiries
O1          | 2: P1, P2
O2          | 1: P1
O3          | 2: P3, P4
O4          | 1: P3
Sum         | 6 person-opportunity memberships
Overall     | 4 distinct participating people

Six is valid if the intended measure is person-opportunity memberships. It is invalid if labelled “people”. P1 and P3 each participate in two opportunities.

As a practical check, compare distinct enquiry IDs before and after a join intended only to enrich enquiries. If the row count increases while distinct enquiry IDs stay unchanged, inspect the relationship multiplicity before using row counts or summing repeated values. The increase may be legitimate, but it changes what one row means.

7. Keep ratios on a declared unit and population

A ratio is not self-explanatory merely because both inputs are correct. In the hypothetical worksheet, four linked opportunities divided by nine enquiries is about 0.44 opportunities per enquiry. It is not automatically a conversion rate: repeat enquiries can link to the same opportunity, and one person can support several opportunities.

More interpretable measures include:

  • Enquiry linkage share: seven of nine enquiries have an opportunity link at cut-off.
  • Person-level opportunity participation: four of six enquiring people appear on at least one opportunity-linked enquiry.
  • Account-level opportunity participation: two of four enquiring accounts appear on at least one opportunity-linked enquiry.

These measures describe recorded relationships, not proof that an enquiry caused an opportunity. They also do not establish progression over time. A later cut-off could change the linkage share without any new enquiries arriving.

For a progression measure, define a starting cohort, an outcome and an observation window. Keep the denominator fixed and count qualifying members of that same cohort in the numerator. If reporting current relationship coverage instead, say so explicitly.

8. Repair an inconsistent report with a small reconciliation

Do not begin by forcing every dashboard to show the same number. First establish whether the dashboards answer the same question.

  1. Choose one disputed total. Export the IDs behind it, along with relevant dates, filters and links. Work within authorised access.
  2. Label the actual grain. Determine whether each row represents an enquiry, a person, an opportunity or a relationship between entities.
  3. Rebuild the distinct sets. Compare IDs rather than only comparing headline totals. Separate missing links from excluded records.
  4. Test one overlapping breakdown. Check whether channel or owner subtotals contain the same IDs. Recompute the overall union.
  5. Publish the corrected definition. Rename misleading measures and explain any changed totals. Preserve the old definition for comparison rather than silently rewriting history.

For the next reporting cycle, use the fictional ledger as a test fixture or create an equivalent authorised sample. The expected results here are nine enquiries, six people, four accounts and four linked opportunities. The channel enquiry counts should add to nine; channel distinct people, accounts and opportunities should not be summed.

The measurable next action is to give one disputed dashboard measure a complete metric contract and a reconciled ID list before the next review. Success is not identical totals everywhere. It is being able to explain every difference through a counting unit, a filter, a relationship or an explicit overlap.

CRM reportingCounting rulesSales measurementData quality

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