Bundles: measuring the value created after cannibalisation.
Key distinctions
Three conditions before calculating.
Heterogeneous valuations
A bundle is particularly useful when customers value the components differently. An average preference does not reveal whether the valuations offset each other at the individual level.
Explicit alternative
Compare pure bundle, separate sale and mixed bundle. Removing the separate components can force adoption but lose the customers who value only a part of the offer.
Valued transfer
Every bundle buyer is attributed to what they would have chosen without the bundle. The margin given up on that alternative is deducted as an opportunity cost.
Method
Assessing a bundle with a counterfactual scenario
- 01Measure the separate choices
Estimate purchase of A, purchase of B, purchase of both and no purchase without the bundle; use behaviour, an experiment or a choice model rather than a single stated intention.
- 02Define the architecture
Test bundle price, retention of the components, eligibility, usage limits and visibility; the architectures must be compared on the same population.
- 03Compute the transfers
Bundle contribution − contribution of the abandoned alternative − specific cost; separate new customers, buyers of A, of B and of both.
- 04Test the robustness
Vary new adoption, cannibalisation of dual buyers, cost to serve and usage rate; publish the threshold that cancels the gain.
The contribution bridge between separate sale and mixed bundle
The adoption gain is set against the embedded discount, the cannibalisation and the cost of complexity in order to isolate the value genuinely created.
A €150 bundle that creates only €1,000 after complexity
Without a bundle, 300 prospects buy A with €50 of contribution and 160 buy B with €40. With a mixed bundle, 240 choose the bundle at €70 of contribution, 120 keep A and 40 keep B. The launch costs €2,000.
Reference: 300 × 50 + 160 × 40 = €21,400. Mixed bundle: 240 × 70 + 120 × 50 + 40 × 40 = €24,400. After €2,000 of own cost, the net gain is €1,000.
The launch is only marginally positive. It must remain testable and reversible; a fall of 15 new purchases is almost enough to cancel the gain.
Acceptance conditions
What must be true to act.
- 01
Separate sale, pure bundle and mixed bundle are compared.
- 02
Every adoption is tied to a counterfactual alternative.
- 03
Discount, cost of entitlement, support and complexity are included.
- 04
The threshold of new buyers needed for profitability is published.
Limits
What this analysis does not prove.
- Stated-preference models sometimes overestimate the interest in little-used components.
- Valuations can change after learning or a change of use.
- Digital bundles with a low marginal cost do not generalise to physical goods.
References
Works cited.
- Adams, W. J. & Yellen, J. L. (1976), “Commodity Bundling and the Burden of Monopoly”, Quarterly Journal of Economics.
- Stremersch, S. & Tellis, G. J. (2002), “Strategic Bundling of Products and Prices”, Journal of Marketing.
- Bakos, Y. & Brynjolfsson, E. (1999), “Bundling Information Goods”, Management Science.