The Power of data quality in eCommerce repricing

June 21, 2023

Read time


‘Garbage in, garbage out’ has always been the information processing industry’s mantra

This old saying’s universal applicability came under the spotlight again in the recent coronavirus pandemic, when the significance of the methods used to obtain target data and the ability to clearly and precisely interpret those data as the basis for sound political decisions became blatantly obvious.

Non-representative samples, flawed data collection methods, the carryover effects of diverse errors and many other factors are all included in the massive volume of data that has to be synthesised, analysed and will inevitably affect the political chain of events. The complexity of this process is made evident by the apparently contradictory information being reported and the resulting public uncertainty. The data situation in consumer markets is similarly complex.

Human error will always result in data inconsistencies.

There are countless factors at play as a result of the multitude of suppliers and products, individual consumer preferences, and intense competition. A smart pricing strategy for the entire product portfolio is a key strategic success factor for companies operating in competitive markets. The practice of price management in conjunction with context-dependent, competition-based pricing (countries, markets, price portals, target groups etc.) or ‘dynamic pricing’ is one of the most effective profit drivers. It increases gross profit versus competitors (by 1-3%), which has a direct positive impact on EBIT of up to 20%.

garbage in garbage out concept

Repricing basics

Obviously, data are the bedrock of dynamic repricing. The better the data quality, the more effective the repricing. In his blog, marketing expert Neil Patel (Google, General Motors) explains how much ‘bad data’ are costing companies, and suggests that these companies could realistically generate 70% more revenue with cleaner data. Data underpin repricing strategies and business decisions in the same way that foundations hold up a building. Here’s what can go wrong: Incomplete data It would be a whole lot easier if retailers used identical names for products. But that doesn’t happen. Tommy Hilfiger is sometimes referred to as Tommy or abbreviated to TH. The eBay shop ABC is called on Amazon. From a computer’s perspective, each of these different names is a distinct brand or supplier. Humans find it relatively easy to make the connections and put them in the right category (although it can be a time-consuming process). For a computer program, however, this is an impossible feat.

Aimondo’s artificial intelligence solves the problem with pattern recognition as part of the machine learning process. It intelligently places queried datasets into the right context. At the same time, a learning algorithm ensures that reliability improves with every query, which saves time, costs and errors. AI can recognise, sort and filter diffuse data sets from the entire global database in a matter of minutes. To make reliable decisions, you need reliable data.

Big data

Inaccessible data

Data accessibility is another important factor alongside data quality, and a web design change is just one of many potential obstacles. That’s the easy part for Aimondo. Things get a lot more complex if there’s a Captcha to solve. Captchas are multilayer security mechanisms to differentiate between humans and bots. For example, online portals use them to ensure that only human users can log in, and to protect the resource from malware attacks. Captchas can be a real challenge, but they’re nothing our artificial intelligence can’t overcome, especially since it continuously learns and improves.

Incorrect data

Human error will always result in data inconsistencies. For example, a supplier may have made a typing error and listed a €10 product on his website for €100. Our algorithm can recognize these deviations from the regular price range and exclude conspicuous data sets from the analysis. When the product is a camera, it’s important to know if the price is for the camera with or without lens, or maybe even an entire starter set. These are important factors that have to be taken into account. Otherwise the quality of the data is severely compromised and could easily lead to incorrect assumptions that cost you dear.

how data should travel within the organisation

Complete and reliable data

Most suppliers tend to use data from two sources only: Global Trade Item Number (GTIN) or EAN code. In this respect, Aimondo is in a league of its own. With the help of neuronal pattern recognition, our AI is capable of capturing product data without a GTIN or EAN. It can even read data out of photos and product images. In fact, it recognizes and captures up to 99.9% of all product offers (vs. competitors ~55%). That’s globally unique.

Say goodbye to incomplete and error-prone information and base all your business decisions on top-quality data.

Take advantage of artificial intelligence, machine learning and neuronal networks: the state-of-the-art technologies for tactical and competitive pricing, comprehensive competition analysis and competitive pricing optimisation. Everything else is yesterday’s news.

Whether you’re a brand, retailer, online portal or manufacturer, you’ll notice a permanent improvement in your profits and market position.

Alex Rose, Tech Lead at Aimondo

Alex Rose, Tech Lead at Aimondo


Tech Lead with a background in academia, Phd in Computer Science

free ebook

Step-by-step guide on maintaining high profitability in retail.

Download a guide