Product Cannibalization Research: Methods & Example
Product cannibalization research measures what a launch might replace. Compare range choices, contribution and the evidence needed to judge portfolio growth.
Product cannibalization research checks whether a new offer replaces sales of products you already sell. Compare the existing range with the proposed range, then calculate what the change means for total orders and contribution. The worked example below shows why a successful new item may still leave a launch below its financial target.
Define which products and buying period belong in the comparison. A switch between your own items is different from an extra order for the business.
Example: does a new bottle add contribution?
A switch from a competitor can add a sale to your brand while leaving category sales unchanged. A switch between your own products can improve or reduce contribution, depending on their economics. A purchase brought forward can make the launch month look stronger while weakening the next one.
Our concept-testing guide explains why product appeal and incrementality are separate questions. Here, take the next step: define the portfolio, time window and commercial threshold before deciding what the research must return.
Morrow is a hypothetical drinkware brand deciding whether to add an insulated bottle beside its single-wall bottle. Both hold 750 mL and use steel. The decision is due before the first production order. Morrow sets a six-month contribution window. The existing bottle is $28 including delivery; the insulated one is $38 including delivery.
Assume variable costs of $12 for the existing bottle and $22 for the new one. Each order contains one bottle. Prices, other variable costs and product mix stay unchanged. Here, contribution is receipts minus those specified variable costs, before other fixed costs.
Commercial input | Worked value |
|---|---|
Existing bottle contribution | $28 − $12 = $16 per order |
New bottle contribution | $38 − $22 = $16 per order |
Additional launch cost | $12,000 |
Additional portfolio orders needed to cover it | $12,000 ÷ $16 = 750 |
The higher selling price does not create higher contribution per order under these assumptions. Morrow needs 750 additional portfolio orders to cover the launch cost. A new-bottle sales target alone cannot tell the product lead whether that requirement is met.
Displaced contribution changes the launch result
Suppose the new bottle sells 1,000 orders in that six-month period, and 900 replace orders that would otherwise have gone to Morrow's existing bottle. That leaves 100 additional portfolio orders.
The new bottle contributes 1,000 × $16 = $16,000. The 900 displaced existing-bottle orders would have contributed $14,400. The difference is $1,600 before the additional $12,000 launch cost, leaving a $10,400 shortfall against covering that cost in the period.
These are assumptions, not collected responses or a sales forecast. Order records show purchases. They cannot show that 900 customers would have bought the other bottle if the launch had not happened. Estimate that alternative outcome with a credible comparison.
Cannibalization is also not automatically a reason to reject a product. A replacement could protect customers who would otherwise leave, improve a lower-margin range or support another objective. Morrow would need evidence for that objective and its value. In this scenario, a higher new-item order count is insufficient because the two items contribute the same amount.
The range comparison includes realistic outside choices
Begin the research with two conditions: the existing Morrow offer, and that same offer with the insulated bottle added. Include realistic competing bottles, keeping their terms and presentation consistent. Let buyers keep their current bottle, postpone or buy nothing when those are genuine alternatives.
Randomly assign eligible buyers to see one range condition first. Giving everyone the old range and then the new range changes their information and exposes the launch explicitly. That sequence can explore trade-offs, but it does not recreate an unaffected comparison group.
Define the buying occasion before showing the offers. Someone replacing a damaged bottle faces a different decision from someone who already owns a working one. Separate current Morrow customers, competitor customers and people without a relevant planned purchase when those groups could change the decision. Do not invent their market proportions from the number of responses generated.
Ask about choice, comprehension of insulation and the objection behind rejection. A later direct comparison can help explain why someone prefers one item. Keep those explanations distinct from an observed individual switch or a causal estimate of displacement.
Zappi's PepsiCo case describes later-stage research considering incrementality across category, portfolio, brand and occasion, with inputs for a separate forecasting partner. That is a useful distinction: research can prepare the portfolio decision without being the final demand forecast.
Connect portfolio choices with the commercial decision
Agree the report before starting. Ask for one row per range, buyer group and available offer, including outside choices. Show the exact version, response source, valid response count and choice share. Label human and simulated responses separately.
Then request a portfolio summary that combines Morrow's two items within each condition. Show the difference in simulated portfolio choice share, the groups behind it, uncertainty and the reasons buyers give. A shift toward the insulated item alongside little portfolio movement deserves a different recommendation from increased portfolio consideration among competitor buyers.
Report where switching evidence comes from: stated usual purchases, follow-up explanations or observed transactions. Keep unknown sources unknown. Separate choice shares for two range conditions cannot reveal every buyer’s unobserved alternative.
Finance owns the cost inputs and sensitivity calculation. Product owns the deadline and launch decision. Keep the six-month window consistent and check later demand if the launch may bring purchases forward. Six months is a planning choice, not proof that it covers the buying cycle. The 750-order target applies to that window.
The pilot measures orders across the portfolio
Sapien can help choose a range candidate, investigate the audience differences and identify what the pilot must resolve. Its interactive report lets your team explore responses and ask follow-up questions. Generated choice shares remain simulated evidence. They are not additional paid orders.
For the commercial comparison, plan a feasible controlled exposure to the old-only and old-plus-new ranges. NIST's experimental-design guidance explains random assignment to experimental units. Choose an assignment and analysis that suit the channel, and account for repeat exposure or other complications before starting.
Track total portfolio orders and contribution across both bottles, with availability, returns, discounts and launch costs. Check the subsequent buying period too. The launch decision should use the portfolio result, including uncertainty, rather than celebrate the new item's sales in isolation.
Our shopper-research guide adds the shelf, availability and competing-choice context when the product will sell through retail. Bring the same realism into the range you ask people to evaluate.
Book a Sapien introduction with the current range, proposed product and contribution threshold. We can help turn them into a study that explains what the launch might replace, which buyers to investigate and what evidence you need before committing.
