Measuring Imperfect Information,
or why customers pay more for less

In a perfect world, at least to economists, all decisions are rational and made with perfect
information. In this economists’ paradise, buyers will choose the best products and the
market will reward sellers who make the best products with higher sales.
This assumption may make the economics simpler but as Joseph E. Stiglitz noted in his
2001 Nobel prize lectures, unrealistic assumptions about available information lead to
poor economic models (

Understanding how much information a customer has when making a decision is critical in
understanding why good value products fail and expensive or poor ones succeed.

Case Study: Hebel Power Flooring


93% of flooring for a second storey home addition in Australia is particle board, a
relatively cheap, low quality material and the default choice for many builders.
Hebel Australia produce a premium flooring panels product called “Hebel Power Floor”.
Unlike particle board, the Power Floor panels don’t squeak or transfer sound, they don’t
sag or bounce and have a solid concrete like feel. Unlike other wooden floors it is also fire
resistant. Costing around 5% of the total construction price more than the other
alternatives, Hebel believe it is a truly superior product that should sell well.
Unfortunately customers didn’t appear to agree with less than 1% of the market choosing it
for flooring.

Modelling Choice

SurveyEngine were engaged to develop a models of choice to understand how customers
were purchasing flooring for home additions. Data was collected and modelled from an
online experiment where potential customers were presented hypothetical flooring
alternatives with different prices and features.
It is important to note, that unlike the real world, the experiment provided full information
on all the products and respondents had freedom to choose as if all products were
available. The models deliberately assumed perfect information.

The “Over Confident” Model

The choice model predicted a huge preference for the Power floor and that at current
pricing, the market share should be 67%. This figure however was wildly off the true
market share of 1%.

These models were of the type described by Stiglitz, accurate, but with the assumption that
the entire market had perfect information about, and access to, the superior Hebel product.
On it’s own, such a model would be regarded as poor predictor of a new product. However
understanding that the models were produced with the assumption of perfect information
meant that an valuable insight about consumer information could be made.

If actual demand was 1% percent but true demand (with perfect information) was 67%
percent, the discrepancy could be explained as an information gap.

This information gap included all impediments to purchase and inefficiencies in the
marketing and distribution. This could include of imperfect information from lack of
awareness, absent POS material, poor branding, lack of product availability, non-existent
to poor education of builders on the product and also stock availability and inertia other
products sales.

Main Model Insights

The Choice Model revealed four important facts.

1) confirmation that the product was, as suspected, superior to the alternatives.
2) that the pricing was if anything, too low.
3) valuation of each feature revealed that thermal and acoustic properties and fire rating
were key sales triggers.

But most importantly of all was

4) poor marketing and distribution of the product that was the main reason for poor sales,
not the product or pricing.


Following the study, senior company representatives confirmed anecdotally through their
own investigations that the product was indeed poorly marketed by their distributors. They
embarked on an education program,sharing the insights above, showing how distributors
could acheive more sales, higher margins and add their own value added services such as
installation warranties.
Since implementing the channel marketing strategy, sales of Hebel Power Floor have
increased dramatically. Importantly, this has been achieved without reducing the prduct