Back to all articlesAudience segmentation

Choose the right recent-activity window for an audience

Alactic Editorial10 min read

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

A contact who requested product details last month may still be evaluating a purchase. A contact who viewed a short-lived offer last month may no longer have a relevant need. Putting both into a “recent interest” audience because the same date filter is convenient gives unlike evidence the same shelf life.

Choose an audience recency window by asking how long a qualifying action remains useful evidence for the audience’s current purpose. Then compare shorter and longer windows using the same events, exclusions, and evaluation date. Inspect the contacts whose membership changes—not just the audience totals—and record why the extra history should count.

The goal is not the largest audience or the freshest possible list. It is a defensible membership rule: this person belongs because this evidence is still relevant. This article focuses on that rule, not when to send a message, how to attribute a result, or how frequently to follow up.

1. Define the current need before choosing the lookback

Start with the decision the audience will support. “People with recent activity” describes a filter, but it does not explain why those people belong together.

A more useful audience purpose names a current need and the evidence that can support it. In a hypothetical equipment business, the purpose might be: “Identify contacts who may still be comparing equipment options and could benefit from selection assistance.” That is different from identifying people interested in a specific promotion or people with an unresolved service request.

Write a short audience contract before choosing dates:

  • Purpose: What current need does membership represent?
  • Qualifying evidence: Which actions reasonably indicate that need?
  • Expiry reason: What makes that evidence less useful over time?
  • Overriding evidence: Which later facts make the original action irrelevant?

These choices constrain the window. A temporary offer can lose relevance when the offer ends. A detailed comparison request may remain relevant while an evaluation continues. A completed purchase can invalidate an earlier pre-purchase signal even when that signal is very recent.

Recency is a limit on evidence, not a substitute for interpreting it.

2. Decide which activity can start or renew membership

A window needs an event definition. Otherwise, “active within the last month” may combine meaningful interest with administrative changes, unrelated enquiries, or passive interactions.

For each qualifying event, specify what happened, which product or need it concerns, and which timestamp represents the action. Use the time the activity occurred rather than an import or record-update time where reliable occurrence data is available. Importing old history should not make old interest appear new.

Also define what renews membership. If a contact requests a comparison guide and later asks a substantive question about the same product, the later question may renew the evidence. An unrelated profile edit should not. Repeated weak events should not silently become strong evidence merely because they are frequent.

For the hypothetical comparison below, the qualifying events are an explicit equipment-comparison request or a substantive question about equipment selection. General page views, administrative updates, and unrelated support activity do not qualify. The most recent qualifying event determines activity age.

That definition intentionally leaves some interest uncounted. The trade-off is clearer interpretation: every included contact has evidence tied to the audience’s purpose. If a weaker signal is useful, evaluate it separately rather than quietly adding it to the same rule.

3. Filled hypothetical comparison: 7, 30, and 90 days

Fictional worked example: All contacts, counts, ages, findings, and business circumstances in this example are hypothetical. They illustrate a recommended comparison method, not customer results or product capabilities.

A fictional equipment supplier wants an audience for selection assistance. Its working assumption is that an explicit comparison enquiry can remain useful during a multiweek evaluation, but older enquiries need additional evidence. The team compares three windows at a single frozen evaluation time.

The following filled comparison is the practical decision record:

  • Audience purpose: Contacts with plausible current equipment-selection needs.
  • Qualifying event: An explicit comparison request or substantive equipment-selection question.
  • Membership rule: Include when the latest qualifying event is no more than the selected number of elapsed days old, subject to exclusions.
  • Boundary convention: An event exactly at the age limit is included; anything older is excluded. Use one documented time zone and evaluation timestamp.
  • Shared exclusions: Completed purchase for the same need, explicit statement that the project ended, duplicate records, and applicable contact restrictions.
  • Comparison method: Apply all three windows to the same frozen records after the same exclusions.

Hypothetical 7-day window

  • Membership: 24 contacts.
  • Evidence retained: Only qualifying activity from the most recent week.
  • Observed review finding: The sample is fresh, but it excludes contacts whose documented equipment comparisons are still open after a week.
  • Operational trade-off: Easy to explain as very recent interest, but too narrow for this audience’s selection-assistance purpose.

Hypothetical 30-day window

  • Membership: 61 contacts, including all 24 from the shorter window and 37 additional contacts.
  • Evidence retained: Qualifying activity up to a month old.
  • Observed review finding: Of the 37 additional contacts, 26 have a dated note supporting a continuing comparison, six have no later context, and five have ambiguous notes requiring review.
  • Operational trade-off: Retains documented multiweek evaluations, while creating a manageable set of uncertain records to inspect.

Hypothetical 90-day window

  • Membership: 104 contacts, including all 61 from the middle window and 43 additional contacts.
  • Evidence retained: Qualifying activity up to three months old.
  • Observed review finding: Of those 43 additional contacts, eight have supporting evidence of a continuing evaluation; 35 have no later evidence confirming that the old enquiry still reflects a current need.
  • Operational trade-off: Captures some long evaluations, but treats substantial unconfirmed history as current interest.

Hypothetical decision: Choose the 30-day window for the general selection-assistance audience. It preserves documented multiweek interest that the 7-day window loses, without extending the same assumption to every enquiry in the 90-day window. Review explicitly documented long-running evaluations separately instead of widening the general rule to accommodate them.

The 35 unconfirmed records in the longest window are not proven irrelevant. The problem is insufficient evidence for treating them as current members of this particular audience.

4. Inspect the contacts whose membership changes

Totals show the scale of a choice. Individual records explain whether that choice makes sense. The following fictional contacts use the same hypothetical rules and evaluation time as the comparison above.

  • Maya — qualifying question 4 days ago: Included in the 7-day, 30-day, and 90-day windows. Her membership does not help distinguish the options because every rule includes her.
  • Jonah — comparison request 18 days ago: Excluded at 7 days; included at 30 and 90 days. A dated note says he is still comparing equipment configurations. He illustrates the useful interest lost by the shortest window.
  • Leena — selection enquiry 29 days ago: Excluded at 7 days; included at 30 and 90 days. There is no later context. Her record shows uncertainty near the chosen boundary, not proof that the middle window is wrong.
  • Omar — comparison request 54 days ago: Included only at 90 days. His record contains a documented extended evaluation. He supports a separate long-running-evaluation rule, not necessarily a longer window for everyone.
  • Priya — selection question 83 days ago: Included only at 90 days. No later evidence supports a current need. Her old enquiry alone is a weak basis for present membership.
  • Ellis — comparison request 3 days ago, followed by a completed purchase: Excluded from every candidate audience. The later outcome overrides the recent interest signal.

When reviewing your own comparison, concentrate on the bands added by each longer window. Ask what would justify including those contacts now. A larger total is not evidence that a broader rule is better; a smaller total is not evidence that a narrower rule is more accurate.

5. Choose using evidence quality and the cost of being wrong

Two errors matter: including someone whose interest is no longer relevant, and excluding someone whose need continues. The appropriate balance depends on the audience’s purpose.

If membership creates a manual review queue, stale inclusions consume review capacity and can obscure better-supported cases. If membership helps identify unresolved evaluations, a very short window can hide contacts who still need assistance. Neither concern justifies making assumptions about every contact in an age band.

Use the comparison to answer three questions:

  1. What useful evidence does the shorter window discard? Look for documented continuing needs, not simply older activity.
  2. What unsupported assumption does the longer window add? Identify where historical interest is being treated as current without corroboration.
  3. Can a separate rule handle exceptions more honestly? A recorded ongoing evaluation may justify membership under a distinct rule with its own review condition.

Do not judge a window solely by outcomes among included contacts. That leaves excluded contacts invisible. Review records outside the proposed boundary as well, so you can detect relevant needs the rule would miss.

6. Document why evidence expires at the chosen point

A defensible window needs more than a label such as “recent leads.” Record the reasoning beside the rule so another operator can understand both the choice and its limits.

Filled hypothetical decision note:

Use a 30-day activity window for the equipment-selection audience. Explicit comparison requests and substantive selection questions qualify. The latest qualifying event renews membership; unrelated activity does not. Completed purchases and closed projects override earlier interest.

The 7-day comparison excluded documented continuing evaluations. The 90-day comparison added many enquiries without later evidence of current need. Thirty days is our provisional general rule because it retains supported multiweek interest while limiting unsupported carryover. This is an operational boundary, not a claim that interest expires on a precise day.

Documented extended evaluations will be reviewed under a separate rule. The audience owner will reassess the boundary after the next record review, including records just outside it.

Keep the event definition, boundary convention, exclusions, owner, and review trigger with that note. If definitions change, preserve the earlier version so a membership shift can be explained rather than mistaken for a change in customer interest.

7. Repair misleading windows without widening everything

When membership looks wrong, first check whether the window is actually the cause.

  • Old records suddenly look recent: Check whether import or update timestamps replaced event occurrence times. Correct the timestamp interpretation and rerun the comparison.
  • Purchased contacts remain included: Fix the overriding-outcome rule. Shortening the lookback merely hides some cases and leaves the underlying error intact.
  • Unrelated activity keeps renewing membership: Narrow the qualifying-event definition and verify the activity concerns the same need.
  • Contacts flicker at the boundary: Check evaluation time, time zone, and inclusive versus exclusive age limits. Explain that legitimate expiry can also change membership without any new activity.
  • Long evaluations disappear too early: Add a separately justified ongoing-evaluation rule rather than extending every historical enquiry.
  • Event dates are missing: Place those records in an unresolved-data review group. Do not silently treat missing age as fresh activity.

Audience membership should also remain distinct from permission to contact someone. Evidence of interest does not itself establish that a particular communication is permitted. Apply the relevant permission and opt-out checks separately.

8. Make the next review measurable

Before changing a live audience, save a comparison snapshot. Record the candidate window, total membership, incremental contacts added, confirmed continuing needs, contradicted needs, and unknown cases. Keep unknowns separate: missing context should not be counted as either confirmed interest or confirmed disinterest.

For a manageable review, examine every changed record. For a larger audience, use a documented sample across the added age bands and inspect records just beyond the proposed cutoff. Record how the sample was chosen so an easy-to-access subset does not become the entire basis for the decision.

Track observable rule outcomes: how many reviewed members still have supporting evidence, how many are contradicted by later facts, and how many excluded contacts have documented continuing needs. Compare like-for-like snapshots using unchanged event definitions and exclusions.

Your next action is to write the audience purpose, freeze an evaluation time, and produce the three candidate memberships. Review the differences, select a provisional window, and save the reason alongside the rule. The result should be a boundary another operator can explain—and revise when the evidence no longer supports it.

Audience segmentationActivity recencyCRM operationsAudience 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