Of all the elements of the marketing mix (product, price, place and promotion) the easiest and quickest to change is usually price. It is also easy for customers and sales to claim that a product or service is too expensive, and this is fed back to the organisation. As a result, when sales fall short of expectations, price is often in the firing line, especially if turnover goals are separated from profit responsibility.
It is no surprise then that price is often the focus of market research. However, price is only one aspect of the value for money equation. Considering price without also considering the value your product brings is only half the picture. You get a glimpse of the larger picture when also considering the significance of your product price, competitors, alternatives within your range and the rest of the marketing mix. The outcome of a full pricing study might be short-term tactical price changes as well as strategic recommendations for product development or repositioning. Or these strategic aspects may just be off the table.
Price changes are complicated by psychology. There are many rules of thumb to tweak prices, such as ‘just below a significant number’ or ‘differences have to be at least 10% to be noticed’. Clearly these can be incompatible, and they break down with higher prices, so management judgement (and potential testing) is the final arbiter. There are also cognitive biases to consider. The most common associated with pricing research is Anchoring. This occurs when exposed to any number, prior to the next stimulus. Subsequent numbers are (estimated/quoted/wanted/given?) closer to the anchor, when compared to samples without the anchor. Anchoring also occurs when comparing against competing products. If you have a product range, you can potentially take advantage of behavioural economics: people find it easier to choose between two similar products and a third (e.g. two Colas and another brand) than just two similar. Mentally it seems this is just easier to do. All these tactical techniques are commonly used.
The above assumes that people are actively deciding, and trade off based on price; if not then it may be counterproductive to draw their attention to it, so a good pricing study should first establish awareness and accuracy. But, assuming respondents are aware and informed about price then we can draw up a few rules of thumb:
- If you want a price increase to be kept below the radar, aim for around 10%. Having, determined this approximate value, see if you can nudge it higher/lower to keep below a psychological price point. Also take advantage of alternative (competitor) prices, if they are already higher, to ‘close the gap’ such that any direct comparison with you puts you in a favourable light. There might be further anchoring effects that you can explore if already premium-priced. Direct comparisons in monetary terms on pack should be avoided; key features, endorsements etc. can take their place. Try to time the price increase with product or service enhancements.
- If you want price to signal differences within a product range, then do the opposite of this (i.e. price differences greater than 10%).
As an example assume that you have a product costing £5 and that research suggests could be priced much higher. Not even considering competition, we would suggest a three-year planned increase of £5.49, £5.99 … and here we hesitate slightly as a psychological threshold is crossed, of £6.49. Why limit this to around 10%? Mainly so that the research predictions can be field trialled and verified. Other products below this should then be set at a level to reinforce intra-brand differentiation, so start at an approximate 10% difference, then nudge up/down to take advantage of perceptions and psychological price points.
Specific research techniques can be used to home in on an optimum price, each with strengths and weaknesses. Each should also consider cost; however, this is unavailable for a large number of studies.
Gabor-Granger was developed in the 1960s by Clive Granger and André Gabor. Best practice for this is to start at a random point on the pricing scale and depending on the initial response ascend the pricing scale until you establish ‘no buy’ or descend the pricing scale until you establish ‘buy’. It is done this way to minimise anchoring within the product. Responses are averaged across all respondents to get % demand at each price point and % demand multiplied by the price point to derive relative revenue, obtaining the classic price demand ‘curves’. Its disadvantage is that it measures only one product at a time and quickly becomes repetitive. Respondents can soon learn that as soon as they reverse their original position the questioning stops (or at least moves on). By design it assumes that lower prices are associated with higher demand (usually a good assumption) and leads to classic volume/revenue curves.
The same information presented as an ‘efficient frontier’.
Where there is a possibility that price might be used as a proxy for quality or there are no existing references, then van Westendorp variants can be considered. This was introduced in 1976 by Dutch economist Peter van Westendorp. Four questions are asked to establish four price points; ‘too cheap to consider, bargain, ‘getting expensive’ and ‘too expensive to consider’ (to minimise anchoring either the bargain or ‘getting expensive’ should lead). These are used to generate four curves, the intersections of which define an area denoted the range of acceptable pricing. The underlying assumption is that respondents use price as a proxy for utility, and output usually identifies realistic ranges. It also usually highlights psychological price points (although our example below, based on smoothed curves, does not incorporate this aspect).
For the ‘Price Sensitivity Measure’ variant, two further questions are included. These establish the probability of purchase at bargain and getting expensive. This creates two hill-like curves, one identifies the price that will maximise trial (for promotions/share goals) and the second the price that maximises revenue. This variant is much easier to interpret. This combines all the questions to create the volume curve; the revenue curve is derived from this and price.
Brand Price Trade Off should be briefly mentioned if only as a warning! It was developed in the 1970s to try to include competing products and prices in the choice. In practice it heavily overestimates elasticity/price sensitivity. Rather than use this, do nothing! To recognise it, if proposed as part of a solution you would expect to see several products presented, priced at their lowest level. The respondent selects a product and the price is increased. The process is repeated and continues until the respondent picks another product, reaches the maximum price or picks ‘none of the above’.
A much better and much more flexible approach to multiple products, brands and prices is to use Choice Based Conjoint (aka Discrete Choice). Sometimes conjoint with only two attributes is also referred to as brand-price trade off; this is not subject to the same issues discussed previously. Conjoint is the most widely used pricing technique but is not recommended for impulse buying or complex decision-making processes. There are many variations, within the conjoint family, however the output, in the form of a simulator, is immediately useful.
A simulator can answer both tactical questions (what will likely happen to Product A, if we reduce price by 10%) and strategic questions (what happens if competitors match the decrease, or we also introduce Product B priced 10% more). To create a conjoint experiment, product attributes and levels are summarised in a grid, which is used to create an experimentally designed mix of choices. Respondents then indicate their preference, usually between 3-4 product options, over several questions. More complex variants such as Menu Based Conjoint may be suitable if respondents can build their own product from priced components.
In market tests (aka a/b testing), a ‘real life’ experimental design, may also be an option if customers can be effectively isolated from each other. This is increasingly difficult given modern communications and price transparency.
If you have near finalised products, then concept tests using matched samples and/or Gabor-Granger may suffice. Customers are split into matched samples, and different prices are used for the same concept. Take-up is then compared between the samples. Larger sample sizes are needed but this is good for when the product attributes can’t be changed, or the objective is to understand repeat purchase.
There is no one pricing technique for all occasions. Those we recommend are based on the objectives, timescales and information available.
Finally, I skipped over costs before. If marginal costs per unit are available then they should be incorporated into the analysis, as profit is important. However, in our experience, under research timescales they seem difficult (but not always impossible) to source. The goal though should be to maximise on profit first and only if unavailable turn to revenue. Maximising on volume share is easy (for all techniques except van Westendorp): just lower prices.
0 Comments