Acquisition growth loops need a declared entry event

A growth loop is not a circle on a strategy slide. This article uses the term for an acquisition or referral loop: a sequence of events with a declared entry rule and a visible next-entry condition.

A measurable acquisition growth loop runs from an eligible entry unit through participant action, transfer, received value, next-entry eligibility, reward or cost, and an outcome window. In this article, the next eligible unit is a distinct recipient or other unit outside the declared starting population. If the arrow returns to the beginning but no new eligible unit is recorded, the diagram describes an aspiration rather than an auditable acquisition mechanism.

This distinction protects the useful idea in the loop. It also prevents a referral, a reward, or a second purchase from being counted as proof of compounding growth.

What is a growth loop?

A growth loop is a declared mechanism in which an outcome from one participant can create or expose an eligible entry for the next iteration. This article intentionally uses the narrower acquisition and referral class: the next entry is a distinct user, account, buyer, invited contact, partner, or content recipient outside the starting population. The entry event might be activation, a referral invitation, a shared asset, a purchase, or another observable event. Retention, usage, and expansion loops can reuse the same user or account and need a different identity and state rule; they are outside this article’s scope.

The loop needs at least six boundaries:

• the eligible entry population and unit identity;
• the participant action;
• the transfer event;
• the received or accepted value;
• the event that makes the distinct next unit eligible; and
• the cost and outcome window.

The growth-accounting article owns the bridge between new, retained, expanded, and lost revenue. This page owns the acquisition event sequence that might generate a new eligible unit before the revenue bridge is evaluated.

Why is a funnel not enough?

A funnel counts units at states. An acquisition loop claims a transfer from one iteration to the next. The funnel can show that 100 users are active and 20 referrals were recorded. It cannot tell you whether those 20 referrals reached a new person, whether that person became eligible, or whether the same unit was counted again.

The loop therefore requires an identity and time rule. A referral event is not automatically a new eligible unit. A share is not automatically a received impression. An activated user is not automatically a paying customer. The record must state which event crosses the boundary.

What does research say about referrals and relationship conditions?

Biyalogorsky et al. (2001) model customer referral management through reward programs. Their result is conditional on the economic and behavioural assumptions of the model. Rewards can be profitable under some conditions, but the analysis leaves room for free riding and reward misallocation. It is not a universal referral rate or viral coefficient.

Morgan and Hunt (1994) identify commitment and trust as key mediating variables in their relationship-marketing model. Their model was tested cross-sectionally with automobile tire retailers, so the setting and single-point design limit transfer to another market or time horizon. That gives a useful mechanism boundary: a relationship condition can make cooperation or referral more plausible, but it does not prove that every customer will refer or that a proposed loop will compound.

Kumar et al. (2010) distinguish customer engagement value beyond transactions through customer lifetime, referral, influencer, and knowledge components. A transaction-only view can therefore undervalue or overvalue a customer when those non-transactional contributions matter. The framework is a valuation lens, not a universal loop multiplier.

Villanueva et al. (2008) compare marketing-induced and word-of-mouth acquisition across speed, cost, and customer-equity paths in a bounded setting. Their web-hosting evidence reports different short-term and long-term value paths for the two acquisition sources. The setting and horizon stay attached to the result; it is not a universal paid-versus-word-of-mouth ratio.

Which fields belong in an acquisition growth-loop ledger?

The table is an acquisition and referral example. A retention, usage, or expansion loop would need a different identity and state contract rather than silently reusing this new-recipient ledger.

The example is deliberately unresolved at the outcome row. Eight recipients became eligible within the illustrative window, but that does not tell us whether they stayed, generated contribution, or would have arrived without the loop. That is the difference between recording a mechanism and proving its economics.

When does a referral become a new loop entry?

A referral becomes a new loop entry only when the recipient crosses the declared eligibility rule. For example, a team might define eligibility as a new account that receives an invitation, accepts a declared value event, and starts a paid plan within 30 days. Another team might use a qualified opportunity or a verified partner introduction. Both can be valid. They are not the same numerator.

The ledger should therefore keep these events apart:

• invitation or referral created;
• delivery or receipt confirmed;
• first value or acceptance observed;
• payment, qualification, or activation recorded;
• exclusion, duplicate, or pre-existing relationship identified; and
• downstream outcome observed at the declared horizon.

For this acquisition loop, a referral becomes a next entry only when the recipient is a distinct, deduplicated unit outside the starting population and crosses the declared rule. A retention or usage loop would instead need a different identity relation and is not measured by this article’s formula. This is why a referral rate is not a loop-conversion rate. The former may count actions per participant. The latter counts eligible entrants that reach the next defined acquisition-loop state under a named window.

How should reward and value be treated?

A reward is an incentive and a cost. The cost may include the reward itself, delivery, sales support, fraud review, service effort, or another declared resource. If the record only says “rewarded,” the economic boundary is incomplete.

The same applies to value. Paid acquisition, referral, influencer, knowledge, retention, and customer lifetime contributions can all matter to customer economics. They should remain separate until the decision names why aggregation is appropriate.

The partner-led growth article owns partner-motion selection, and the customer-selection article owns the allocation of scarce commercial capacity. A growth loop can generate an input to either decision. It does not decide the allocation on its own.

How can acquisition-loop conversion be calculated without inventing compounding?

Declare the denominator first. A bounded descriptive convention is:

loop conversion = entrants with at least one qualifying next entry within the window / eligible loop entrants

Each eligible entrant is counted at most once in this numerator, so the conversion is bounded from 0 to 100 percent. The next entry must be a distinct, deduplicated unit under the same unit, period, and exclusion rules. When one entrant can create multiple qualifying units, report a separate output-yield metric:

next-entry yield = deduplicated qualifying next-entry units / eligible loop entrants

Yield can exceed 100 percent because it counts output units rather than entrants with at least one output. A team may also report transfer rate, acceptance rate, time to next entry, reward cost per eligible entry, and later contribution. These are different metrics.

In the synthetic ledger, suppose eight of 100 starting users each produce at least one qualifying next entry and eight distinct recipients qualify. The illustrative loop conversion is 8%, and the yield is also 8%, or 0.08 qualifying units per entrant, because this example has one output per converting entrant. If those same eight entrants produce 20 distinct recipients, conversion remains 8% while yield becomes 20%, or 0.20 units per entrant. Neither value is a benchmark or proof that the next iteration will reproduce the output, because the next population, cost, and outcome have not been observed.

What does a growth-loop diagram not prove?

A diagram does not prove causal growth, profitability, payback, retention, or self-sustaining demand. The referral model may depend on assumptions that do not hold in the new setting. Trust may be absent. The recipient may already be exposed through another channel. The reward may attract low-value or duplicate units. The apparent next entry may be a measurement artifact.

To claim causal growth, specify the intervention, eligible population, counterfactual or comparison, outcome, attribution rule, and horizon. To claim payback, add contribution, cost, cash timing, and a recovery rule. A loop is a hypothesis about a mechanism until those tests are done.

How should a team run a loop test?

Choose one loop and one period. Version the entry rule before the test. Then record:

1. who was eligible at entry;
2. what action was taken and by whom;
3. what transfer event was observed;
4. which distinct recipient crossed the next-entry rule;
5. what reward and channel cost were incurred;
6. what exclusions or duplicates were removed; and
7. which contribution, retention, or customer-value outcome remains open.

Review the ledger after the outcome window closes. If the loop cannot identify the next distinct eligible unit, keep it as a campaign or referral activity, not as a measured growth loop. If it can identify the unit but not the outcome, keep the mechanism and mark the economics unresolved.

The defensible conclusion is narrow: a growth loop is a declared event sequence, not a circular arrow. Its credibility comes from a visible entry population, transfer event, next-entry rule, cost, and outcome horizon. Measure the mechanism first, then test whether it creates value relative to a credible comparison.