Data Blog by Lizeo

How to analyze tyre sell-out prices in 5 steps?

How to analyze tyre sell-out prices in 5 steps?

Analysing and monitoring consumer price changes in any market requires identifying and defining the business questions that this study needs to answer:

  • What are the sell-out price dynamics in my market?
  • What is my price positioning in relation to my competitors?
  • On which product panel? In which segment?
  • In which geographical area? On which channel?
  • Is my profit margin preserved? etc.
Asking these questions will allow you to define a scope and as such, select and guide the choice of data required for these analyses, whether internal or external.
For external data, it is often data on competitors’ prices, collected online or in the field. For internal data, it can be sales volumes, sell-in prices, profit margins, etc. In most cases, both of these types of data are required to conduct a meaningful analysis.

Let’s take the case of the tyre market:

 Define the scope for online price collection

To correctly define a scope of online price data collection, several questions need to be answered:

What is quality consumer price data?

The definition of a quality price data lies in its ability to meet several criteria:
  1. This first criterion is to precisely answer business questions you have defined beforehand. As such, when analysing this data, the extraction of findings will be directly related to the previously mentioned issues.
  2. The second is the notion of confidence : origin of data (traceability), repeatability in the collection process and independence of the supplier in relation to the market. Indeed, a supplier cannot sell its products in the target market and produce its own data.

What level of accuracy is required to provide a consistent and detailed price analysis?

The level of data accuracy has a significant impact on the results of an analysis. Indeed, the average difference in price for a tyre, with or without a manufacturer’s marking (e.g. AO marking for Audi), for a same size, can vary by several percentage points. The inability to distinguish them will distort the analysis of the average price for a specific size.

When to collect data?

The lifespan of price data is inherently very short. Therefore, its freshness is a criterion of crucial importance. In e-commerce, the frequency of re-pricing is very high, often day-to-day in the tyre industry, and sometimes on an intraday basis in other industries.

The collection frequency is therefore of utmost importance to obtain the most accurate analysis possible: an average price per week will be more accurate if prices are collected daily.

What parameters are involved in the collection of online price data?


  • Geographical: what country, region, states?
  • Commercial websites: number, location, etc.
  • Products: What brands, what product ranges?
  • What segment: Touring, Truck, Motorcycle, etc.
  • Sizes: 16, 17 or 18 inches



  • Complete size: 205-55-R16-91V
  • Technical markings: Runflat, Manufacturer Marking (OE), Rim protection, etc.



  • Collection frequency: daily, weekly, etc.

The parameters involved in collecting online price data are scope, depth and freshness.

How to collect data in an automated, industrialized and controlled way?

The collection of online consumer price data requires the use of tools which “visit” targeted commercial websites to collect the information needed, without disrupting them (ethical collection).

These tools must be smart and efficient since an e-commerce website is bound to change regularly: new design, new layout, new URL, etc.

These changes have a direct impact on data collection tools: website change = loss of data collection.

In order to keep these tools up to date, you need a team that monitors each website on a daily basis to ensure the operational capability of the system, 24 hours a day, 7 days a week.

Clean, match and unify data

Now is the time that ‘Data’ specialist expertise comes into play.

The description of a tyre sold online is one of the elements that can vary significantly from one e-commerce website to another.

The list of attributes which constitute the description of a tyre and which can vary are as follows:

  • The brand
  • The product name
  • The technical markings (runflat, OE Marking)
  • The tyre manufacturer code


To be able to attach the prices collected to the right tyre, it is necessary to “understand” and decipher the key elements displayed which allow this product to be identified in order to compare them with a ‘specific’ reference base of existing tyres (current or in the past) and validated (with an official source attesting it). This is the data unification step.

This referent database is the cornerstone of an efficient matching system to deliver data with an unparalleled level of quality.

The comparison of data collected with this referent database checks that the information found online for this tyre exists: does this tyre come from this brand? with its technical attributes? etc.

This allows data collected online to be systematically and automatically categorised:

  • Unknown product: is it new?
  • Known product: attachment of the collected price, its source and the collection date
  • “False” product: the combination of information is not possible – the product does not exist


Finally, the most outrageous prices are also filtered (e.g.: a touring tyre sold for less than 20€ is unlikely to be real)

The combination of a highly qualitative tyre database, matching technologies, Machine Learning algorithms and product marketing expertise provides quality data with a high level of completeness.

This will allow you to focus on your business (the analysis) and gain in efficiency. Indeed, analysts and Data Scientists spend between 50 and 80% of their time cleaning data before they can start handling it!

Data Scientists spend between 50 and 80% of their time cleaning up data before they can start manipulating it!

Enhance data through tire market knowledge

In this step, the data is cleaned, structured and unified. Prices are attached to the right product, which has a unique description.
However, analyses conducted on this level of data do not yet provide sufficient accuracy.
Indeed, the segmentation of the tyre manufacturer, the vehicle segment, the season, the manufacturer marking or the European Article Numbering – EAN or Global Trade Item Number – GTIN are examples of filtering and analysis elements which are of critical importance when tracking market prices. Unfortunately, this data is not available online, or very rarely with a very low level of quality.
It is therefore necessary to enhance the data collected online with ‘offline’ data from catalogues, pricing scale, PriCat, all in a unified, overall and multi-brand view.
Enhancement data:
  • Tyre Manufacturer Segment (Third party): Premium (1), Medium (2), Budget (3)
  • Vehicle segment: Car, AllTerrain, Van, etc.
  • Season: Summer, Winter, AS, Nordic
  • OE Marking: Audi, Mercedes, BMW, Tesla
  • Technologies: Runflat, Seal, Extra Load, etc.

 Preparation and generation of the final data cube

The preparation step converts unified and enhanced data into a set of data (cube) that integrates your business rules, your internal data but also your vision of the market and your segmentation. Indeed, data on competitors’ prices is a key element in understanding the market but internal data is just as important.

Examples of business rules:

  • Segmentation of tyre manufacturers
  • Aggregation of data by channel
  • Aggregation of products by use
  • Calculations on an identical scope of sizes
  • Product clustering


Examples of Internal data:

  • Sales volumes
  • Sell-in price data
  • Profit margin
  • OE fitment (Tyre is original equipment on vehicle brands and models)


This is the step where the accuracy and consistency in the collection of online data is of particular importance. Indeed, to make sense of these 2 data sets, the granularity of online data must be at the same level. As such, these 2 data sources will be matched at the same level in order to compare the same elements.

This is the last step before starting to ‘consume’ data in business tools.

 Data analysis and extraction of findings

This is the last step.
This is the last step. It is now the time to inject data into one or several tools to work on it in various ways: analysis, simulation, modelling, monitoring, extraction of findings, decision making, etc. 
However, before injecting it; it is crucial to choose the right tool(s) to maximise the use of data and adoption, but above all, provide a precise answer to the business questions asked at the first step.
By establishing the profile of the data users as well as the intended uses: this is the user persona. His profile is built through a few simple questions:

  • What is his background? Technical, Non-technical.
  • What is his level of maturity / knowledge with respect to the data? Beginner, expert, etc.
  • What is his business? Price Analyst, Data Scientist, Statistician, etc.
  • What are his objectives? Study, monitor, decide, etc.
  • Who are the studies for? Management, Sales, Marketing, etc.
  • What is the frequency of use? Daily, weekly, monthly.


Depending on the users’ personae, preference will be given to:
  • A Data Science tool to handle and create views and data cubes, create scenarios and simulate them, compare scenarios, etc.
  • Business Intelligence tool which will perform “data discovery” and extract findings: advanced filtering, data mining, construction of indicators, etc.
  • Data Visualisation or dashboarding tool to share findings, make decisions quickly or manage an activity on a daily basis.

Why you should trust us with this methodology for analysing sell-out prices in the tyre market?

Lizeo Group supports its customers in their data-driven digital transformation and has been collecting tire price data online on a daily basis for more than 10 years on over 1000 e-commerce sites worldwide.

This represents approximately 11.5 Million price lines per day, in all currencies, which are cleaned, matched, enriched and ready to use.

In this article