Oppositional Loyalty

What it is

The construct is useful because it is behavioural rather than attitudinal: it predicts something about what a member does with a competitor's launch, which a satisfaction score does not.

It also makes the firm's position worse than a loyalty figure suggests, because the same mechanism can run backwards.

What was measured. Longitudinal data on 7,506 members across four brand communities and two product categories, covering participation, membership duration and adoption behaviour. This is the best-powered behavioural study in this publication's brand-community file.

Three findings, and the third is the one worth the page.

1. Heavier participation and longer membership raised the likelihood of adopting the preferred brand's new products.
2. The same variables lowered the likelihood of adopting an opposing brand's, but only where the rival's product was not first to market. First mover beats membership.
3. Where memberships overlap, higher participation in one community may raise the likelihood of adopting a rival's products instead, a result the authors themselves describe as surprising and disconcerting.

The third finding is the one managers cannot act on and should know anyway, because firms usually do not know which other memberships their members hold. A community programme measured only on its own brand's adoption cannot see the reversal happening.

And the moderator in finding two is the honest version of the whole construct. A community that reliably refuses a rival's launch is worth money. A community that refuses it only when the rival is slower is worth a great deal less, because being slower is not something a competitor is obliged to be.

Reading level: this paper was read at the level of its publisher-deposited abstract, which is where the 7,506 figure, the moderator and the reversal are all stated. The full text was not read, so no model or coefficient is held.

In effect

In research: the rival account that has to be answered first

Andrew Ehrenberg, Gerald Goodhardt and Patrick Barwise's double jeopardy says a small brand has far fewer buyers than a large one in any given period and its buyers also buy it less often, so loyalty largely tracks size. The loyalty visible inside a community is therefore a candidate for the brand-size explanation before it is a candidate for the community explanation. Karen Nelson-Field, Erica Riebe and Byron Sharp supply the measurement that makes this concrete: two brands' fan bases skewed towards the heaviest buyers while the real buyer bases followed the ordinary spread.

In everyday life: the refusal that is legible

The person who will not own a particular phone, console or supermarket is describing an identity rather than a preference, and the refusal does work the purchase cannot, because it is visible to other members. A purchase is private and a refusal is a statement, which is why oppositional loyalty is the form community loyalty most often takes.

In commerce: the structural limit

Michael Morris's structural definition of a cult is a community that severely restricts the prestige signals its members receive, with plural membership as the ordinary protection. Thompson and Sinha reach the same fact from the measurement side: where memberships overlap, participation can raise rival adoption. So plural membership is what stops any one community owning a person, and it is also what stops a firm owning a customer. The same structure that protects the member limits the asset.

What it does not say

It does not say membership causes refusal. The design is longitudinal observational data on people who chose to join, so the selection objection applies in full: the members were disproportionately the heaviest buyers before anybody joined anything.

It says that among members, participation and duration track adoption in the directions above, under the two stated conditions, and the conditions are not decoration.

It does not quantify the effect for a reader. No model or coefficient from the paper is held here, because the full text was not read.

And it does not generalise beyond four communities in two product categories. One study, however well powered, is one study.

---