Simulate market share shifts before launch by using modern choice models and disciplined scenario design. Learn how to quantify portfolio lift versus internal cannibalisation and select the right estimator.
Which approach is standard for simulating pre‑launch share shifts among competing items?
unaided awareness surveys
A/B tests on live price cuts only
discrete‑choice models such as choice‑based conjoint (CBC)
simple linear regression on sales only
The MNL model’s IIA property can distort cannibalisation estimates because it ______.
requires only one attribute per product
forces all prices to be identical
forbids including competitor items
assumes a new near‑duplicate option steals proportionally from all options
Which model structure relaxes strict IIA by grouping similar items to better capture within‑group substitution?
nested logit
ARIMA time series
ordinary least squares
Poisson regression
To quantify cannibalisation in simulation, analysts typically ______.
increase the market size assumption arbitrarily
compare portfolio shares with and without the new SKU under the same conditions
remove all competitor items first
sum utilities across respondents without simulating
Accurate cannibalisation forecasting in CBC depends on including ______.
only your hero SKU
open‑ended comments without profiles
your own SKUs and key competitors with realistic attribute and price ranges
only the cheapest competitor
Which estimator is commonly used to obtain individual‑level part‑worths for robust portfolio simulations?
k‑means clustering
principal components analysis
naive Bayes classifier
hierarchical Bayes (HB) for CBC
A red flag for harmful cannibalisation is ______.
new item gains offset by a negative net incremental share for the total portfolio
feature awareness increases
competitors lose share
media ROAS improves
A sound validation step when using choice models for forecasting is to ______.
drop respondents with uncommon preferences
test out‑of‑sample predictions with holdout tasks or periods
ignore standard errors in simulations
fit the most complex model to minimize in‑sample error only
Retailers can curb cannibalisation risk in assortments by ______.
eliminating private label universally
pricing all items identically
curating a mix that targets distinct shopper segments rather than duplicative options
maximizing SKU count regardless of overlap
Omitting a major rival from choice tasks usually ______.
only affects price sensitivity but not shares
automatically improves forecast accuracy
has no effect if sample is large
overstates substitution into your own items and misreads cannibalisation
Starter
Review CBC basics and run clean base vs. new simulations.
Solid
Strong grasp—stress‑test models against holdouts and rivals.
Expert!
Excellent—your simulations optimize mix for portfolio growth.
