Pricing Academy Article
SNIFFIE’S AUTOMATIC MARKET PRICE OPTIMIZER
Background on SAMPO
For some, the challenges in pricing can seem loosely grounded in reality and it might look like there isn’t just one thing that can be done to solve these problems. At Sniffie, we are a group of ambitious and hungry people, so of course we wanted to set out to solve these problems and this is how SAMPO (acronym for Sniffie’s Automatic Market Price Optimizer) was born. A challenging task, we admit, but at the same time we strongly believe this to be the future of price optimization.
Reinforcement Learning & SAMPO
After years of working with pricing there were a few key difficulties that we observed when it came to product pricing:
- In case of multiple products, there may not be enough time to price every product efficiently
- There isn’t enough knowledge on how pricing products affects the sales of other products
- There usually isn’t enough data on different prices and sales volumes to model the relationship between price and sales volumes
With the problems mentioned above in mind we started to rule out different solutions and after some brainstorming we landed at using reinforcement learning. In addition to solving our initial problems, we were now also able to react to the changing market conditions. Not long after the initial solution was formed, we asked for advice from one of our customers and we found out that the main problem was that their predicted sales volumes weren’t big enough for our AI to use in forecasting. We went back to work and started focusing on how we could use our solutions also for products that sell less. A couple of weeks later we were able to produce a solution that worked even when the store sold less products. Our solution was able to optimize the price levels for even these low volume products sold infrequently.
Market Cannibalization & SAMPO
The most difficult part of developing SAMPO was finding out a solution for market cannibalization. Our goal was to develop a solution that would cannibalize only on the competitors products, not on our own. This was done because most of the time, consumers would buy an alternative if the initial product was not available.
The 2nd Purpose of SAMPO
While SAMPO was optimizing prices, we also realized that the information which we used to observe SAMPOs actions was also useful to learn about the market and products. We used this observation to create a secondary purpose for SAMPO; to produce relevant market data. We further processed the data to include price history, sales history, price elasticity of demand, the profit elasticity of demand, and monitored price data from the web.
What Does SAMPO Do?
SAMPO (an acronym for Sniffie’s Automatic Market Price Optimizer, and also a Finnish male name which is why SAMPO is referred to as “he” in our text) is a reinforcement algorithm. By telling SAMPO the outcome of the chosen price, SAMPO changes his opinion on that given price. The outcome can be viewed as pure profit or as profit per customer. Profit per customer is especially important in situations where the float of customers is not evenly distributed. If for example the number of customers is double during christmas time, also the amount of sold products can be roughly estimated to double. To separate the effect of price from the effect of the number of potential customers; profit per customer can be used.
Due to the variability in product pricing, SAMPO has been designed to be flexible. You are able to modify both the amount of feasible prices as well as the interval of feasible prices and even the rounding of the feasible prices can be modified to make sure they are always in line with your company’s strategy. The sensitivity of SAMPO can also be modified. If SAMPO is set to be very sensitive, he will react to small changes which can be useful in situations like COVID-19. On the other hand, SAMPO can also be set to not be so sensitive and only be interested in long term changes and trends. This feature was specifically designed to give the user a chance to use SAMPO in different ways depending on the users needs.
Use of Strategies With SAMPO
When it comes to strategy, SAMPO can be set to pursue an aggressive tactic and start exploiting a good price point as soon as it is encountered. The opposite option would be to have SAMPO continue to try price points nearby in order to find an even more profitable price, something that is also possible to choose in SAMPO’s settings. None of the options in SAMPO are dichotomous (on-off), meaning that the level of sensitivity and aggressiveness can be set to the desired level.
As we have mentioned before, one of the biggest challenges in pricing is market cannibalization and this was something we wanted to tackle with SAMPO.
SAMPO can now be used to price multiple products and this feature also accounts for market cannibalization. When pricing multiple products, SAMPO is optimizing the profit of the whole product-category or bundle. Due to this, the prices will always be increased or decreased together by a similar factor.
SAMPO Is Not Only a Price Optimizer
SAMPO is not only a price optimizer he also produces valuable market data and he is designed to use his knowledge to help Sniffie’s customers better understand the market and make better pricing decisions in the future.
When SAMPO gets to work, he starts by pricing for a period of time and tries to estimate your mean sales volumes. This initial phase can of course be accelerated using accurate sales data that SAMPO can use for analysis, instead of first gathering the data needed to start pricing optimally. After getting a grip of the initial mean sales data, SAMPO starts to explore different price points based on prior knowledge and respecting the safeguards you have set. While using SAMPO does not guarantee bigger profits for a single product, the overall probability of a person beating SAMPO’s choices in pricing is diminished with multiple products and a longer timeframe.
How Can Anyone Working With Pricing Benefit From SAMPO?
We developed SAMPO to help pricing professionals become even better at their jobs and earn more profits. That being said, SAMPO’s benefits can be divided into two categories:
1. Increased profitability
2. Increased knowledge.
The primary benefit of using SAMPO
The primary benefit of using SAMPO is increased profitability. The objective when pricing products is to find the optimal price and SAMPO is designed to help with this objective. We wanted to leave the opportunity for the user to stay in control as much as they want so we also configured SAMPO so that you can choose which products you want SAMPO to price and which you want to keep pricing manually; there is no “all or nothing” scenario here. This way you can outsource the pricing of products that sell on less frequent occasions or products that are outside your comfort zone. If you have, on top of the types of products mentioned above, strategic products that are sold for different purposes than to optimize profits you can use e.g. Sniffie’s rule-based pricing for your strategic products. The key is to know the purpose of the product and then you can choose how you wish to price the product; manually, with rule-based pricing strategies or using SAMPO.
While SAMPO is optimizing product pricing he is also learning more about your products and the market. SAMPO will create useful insights of the market and products including price history, sales history, the price elasticity of demand and profit elasticity of demand. You are able to view elasticities, not only as point elasticities, but also as continuous functions making it a very useful tool for analytics and decision making. The continuity is also very important since the elasticities are dynamic in natured and therefore not constant over wider price ranges.
All the information SAMPO generates will take you even closer to making completely informed pricing decisions, something which we all know, is the key to profitable pricing. If you can choose between having more information to base your decision on and not knowing very much, it’s quite clear which option you are going to go for. This mentality has also been present in everything we do when developing SAMPO, we wanted to know as much as possible and put together our brightest minds in creating SAMPO, all so that you can reap the benefits. You can even add and remove products from SAMPO at your own will; let SAMPO price and gather market data for a while and then switch over to manual pricing, if you want, armed with the knowledge SAMPO has mined for you. This is especially useful when wanting to investigate a new product on the market or when you are unsure of how the pricing for a certain product or category should be done and you have no data to go on in the beginning.
SAMPO’s controlled automation
We know that it can feel a bit scary to give over all control to a computer so we didn’t want to force anyone into that situation. With SAMPO’s controlled automation you can choose to remain in full control and SAMPO will ask for your confirmation before changing any prices. To make sure you don’t miss out on valuable time confirming the price changes, we even developed a mobile app that lets you view price suggestions and confirm price changes right on your smart device, wherever you are.
If you are willing to go all in on AI pricing and pricing automation in general, you can just set up safeguards and thresholds under which SAMPO can operate and it does everything automatically all the time, without you lifting a finger! It’s all about your comfort level in pricing automation, and we at Sniffie want to be able to cater to your needs and desires in anything related to pricing or pricing automation.
Read more about what SAMPO has to offer
Check out our latest blogs
“We don’t have enough historical transaction data.” “We don’t have big enough sales volumes.” “We don’t want to be the cheapest.” “We can’t change prices
Behind the retail & e-commerce buzzword: Actionable definition and 4 practical examples of dynamic pricing
Dynamic pricing is one of those buzzwords that gets thrown around so much that it barely means anything anymore. That’s why we decided to ask