Pricing has been a major player in the rush to aftermarket digitization. Between the huge amounts of data involved and the manpower required to sift through it all, automation seems like the perfect fit.

Author Nick Saraev

Photo: Freepik

However, if you rush into automation without the proper foundation, you’ll end up with more challenges and potential revenue loss. Daryan Parab, the Costing and Pricing Manager at A220, walked us through the process he used to bring automation and advanced computing to two spare parts pricing teams. By following his method, you can ensure your team is prepared for the complexities of pricing spare parts in a modern world. 

The Need for Automation 

The average manufacturer might have upwards of 50,000 individual SKUs and parts numbers to deal with. This vast amount of data is incredibly difficult to manage, especially when you consider the amount of slow-moving parts that won’t be referenced often. This is merely the tip of the iceberg when it comes to why spare parts providers need automation. 

The challenges that companies face are unique to the spare parts and aftermarket worlds, including 

  • Fast Response Times – Automation allows you to drastically reduce response time, lowering the risk of lost sales
  • High Numbers of SKUs – There are, on average, 20x more SKUs in aftermarket than in manufacturing 
  • High Effort for Low Reward – Spare parts often bring in low revenue numbers compared to the effort that goes into dealing with them, but this is transformed by automation 

When dealing with parts that aren’t ordered often, the results of a mistake are menial. This allows you space to experiment with automation and create a system that works. 

The Scale of Pricing Automation 

The focus of this digitization is to create prices without the need for human intervention. The customer may be the first human to see the price for certain items. This would be possible for items being priced for the first time, any one-off quotes that are requested, and any pricing updates that need to happen to stay competitive. 

Automation of this scale goes beyond simply inputting the cost of a part and adding a markup for price. However, it can lead to problems if you attempt to use it for more complex parts. Catalog prices, published prices, fast movers, and any pricing that has a high-risk factor should still be done by professionals. 

For example, if you have a one-off sale and a customer requests a quote, you can quickly generate it and not risk leaving vast amounts of money on the table. 

Designing Pricing Models and Algorithms 

A traditional pricing model will begin with inputting the cost to build an item. That number will then be marked up based on several factors such as 

  • Usage Criticality 
  • Material Groups 
  • Competition 
  • Life Cycle 

The result of these calculations is a sales price that is accurate and ready to be shared for complex parts and high-volume sellers. However, this system takes time and effort from team members. 

Automated pricing can take data from your system and bring multiple inputs including 

  • Price History
  • Cost History 
  • Inventory Value 
  • Market Price 
  • Equivalence/Similarity 

It then adjusts for risk factors and the usual mark-ups, creating a sales price without involving human input. The calculations are still relatively simple, not requiring any AI or machine learning. 

Taking the algorithm to the next step involves more advanced computing. The “input” will be an amalgamation of 

  • Price and Cost History 
  • Demand History
  • Customer Data
  • Equivalence/Similarity 
  • Market Data 
  • PN Patterns 
  • Higher/Sub Assemblies 
  • BoMs
  • Dimensions, Weight, Etc. 
  • Other Organized Data 

With all this information coming in, machine learning can apply regression, generalized linear models, clustering, and more to find patterns, and suggest prices that fit. 

Making the Jump to Advanced Computing 

Building up the infrastructure and frameworks to allow for advanced pricing automation takes time. You can not go from having no automation at all to having a fully automated, intelligent system. The key is building it up slowly.

Each step will take time, money, and effort. The system requires you to bring in more people to handle data and train the models, as well as build up the infrastructure for data management and tools to deal with advanced computing. 

Phase One: No Automation

At the start of your journey you will have basic tools and structures to individually price each part. This usually requires quite a bit of manpower for every pricing request, especially if it is for a part that isn’t sold often. 

  • Tools – Excel, Google Sheets, and other basic tools for pricing 
  • Data Management – Local or cloud server to store data from previous years
  • Organisational Structure – Pricing analysts and regional focal points that report directly to the finance head 

Phase Two: Automation

Bringing in tools specifically to automate the basic pricing process elevates the complexity of your structure, but already dramatically decreases the need for individual pricing analysts working on each price. 

  • Tools – Dedicated BI tools that interact with your basic pricing interface
  • Data Management – Data warehouse 
  • Organisational Structure – Pricing managers to focus on strategy, pricing analysts to focus on calculations and reporting, and pricing head to oversee

Phase Three: Advanced Computing

At this phase you will be generating prices based on a wide range of data points, and the quality of your data itself becomes paramount. 

  • Tools – Customizable pricing tool to engage with your advanced BI tools
  • Data Management Data lake
  • Organisational Structure – Bring data scientists and engineers on board to report to the pricing head as well 

Automation in Action 

When you reach phase three of automation, you are ready to implement a feedback loop. This process acts as a failsafe to ensure pricing outliers are manually reviewed. 

When a quote is requested, the system will use a series of thresholds to decide if a pricing analyst is needed. These thresholds include 

  • Price Level – If a spare part generally retails for an extremely high price, it will not be handled by automation alone
  • Revenue Impact – When the potential impact on the company’s revenue is low, there is no need to involve analysts 

If the part is able to pass these thresholds, the price will be given to the customer without interference. Otherwise, an analyst will take over the case. 

Another way that team members can get involved with a pricing matter is if you receive a customer complaint. This means that if a miscalculation slips through the cracks, you can address it and update the algorithm for the future. 


Digitizing your pricing model is complex, especially when dealing with the scale of many spare parts operations. When you take the time to create automations that handle the low-risk quotes, you free up your team to give larger deals the time and attention they deserve.

How useful was this post?

Click on a star to rate it!

Average rating 0 / 5. Vote count: 0