This is the second part of the How to use data to master the 4ps of marketing series.
Modern technology has made promotion and distribution a frictionless experience for marketers across the world. Through Google Adwords and Facebook Ads one can reach their target audience with a limited advertising budget. Whilst marketplaces like Amazon and Ebay do the heavy lifting of attracting buyers, managing transactions and — in the case of the former — fulfil orders for independent merchants.
Although it can be argued that this somehow democratises consumer attention and purchase consideration to anyone with internet connection.
It can equally be argued that this creates noise in some markets, with sazzy online marketers competing on price, subsequently shifting the market value away quality and customer service to “where can I get this cheaper?”.
As highlighted on the initial article, I subscribe to the idea that the customer is the judge in the marketplace. But this doesn’t mean the customer is always right.
A cheaper product might be music to the ears of shrewd consumers, but history has shown that markets competed on price are not sustainable. At some point something has to give along the supply chain, eventually delivering poor quality and reduces creativity and innovation in the marketplace.
The old retail advice of simply doubling or quadrupling your costings is not sufficient in this complex battle of finding pricing and market fit.
So how do you compete on pricing without undermining quality?
You guessed it — data. Businesses already operating in a market place have the unfair advantage of transactions data to build simple but effective models which do not require data scientists.
To best showcase this, I will use my wife’s online business as an example.
In 2015, my wife reached a crossroad many creative entrepreneurs face — ‘how can I stop bankrolling my business and it begin to pay for itself?’
At this point we had tested everything from low margins, matching more expensive competitor, wholesale and flash sales, as we tried to find our position in the homeware market.
Outcome of these experiment were not surprising:
- Flash sales increased sales — but lowered profitability
- Matching expensive competitors meant 0 sales — the silver lining here at least our competitors did not have it figured out. (coincidently none of the sellers positioned here are in business today)
- Wholesales increased sales — but squeezed our profit margins with the retailers taking the greater share of retail price.
The hypothesis after this result was: ‘we had the right product, but depending on the place of distribution the prices differed massively. Online market it sold when discounted, in retail it sold at a higher price point’
The decision was to invest in finding customers who are willing to pay the retail price, through our online channels and limit the number of sales to once a year.
But the challenge still remained. How do we price effectively and measure success?
A measure of success would be an increase in cart to detail, buy to detail rates, revenue and Average basket value.
Cart to detail rate: Product adds to basket divided by product views
Buy to detail rate: Product purchases to basket divided by product views
Our Pricing Model
We took two and a half years worth of sales data, and created a simple model which calculated the average price and median net prices per product. Then worked out the standard deviation (essentially the difference one way or the other from the average/mean).
We then created the lowest price and highest prices for each product using average and median. For example the above screenshot for a fictitious Product 1, the Lowest price is around £6 and the highest price of £18.
This gave us the boundaries to play with when picking a price on a product by product basis, and also helps inform the cost of manufacturing. If product 1 cost £15 to produce and market, then we would consider discontinuing it as there is only a £3 profit margin at best and potential £9 loss when discounted.
Now, it would be untrue to suggest that this result was only driven by the price optimisation, because we had to invest in marketing, UX, professional photography as well as a more efficient supply chain. But at the core of this investment was having prices, which we were confident worked for the market and the business.
So in summary, experimentation with pricing does not require complex algorithm. A simple model using some statistic functions on Excel can provide a healthy boundary for experimentation and iteration.