¿How do you decide where to source wholesale purchases for your retail business?
¿How can you trust the quality of the manufacturing process – or the quality of the supplier’s delivery rate, response rate, and professionalism?
At Meru.com, we use Gigabytes of data and machine learning algorithms to help us and our clients focus on the best Asian manufacturers.
At Meru, we specialize in helping retailers in Mexico and Latin America make wholesale purchases from Chinese suppliers and manufacturers. For us that means providing both a marketplace of popular products (“hot sellers”), but also a shopping experience that feels closer to home – even though our products are sourced on the other side of the Pacific Ocean!
Not only do we take control of the logistics involved in physically delivering between the two continents – but we also build data driven algorithms that help remove all uncertainties that come with buying from Asia.
In this article we’ll take a look at how our team of data scientists address questions about which products to purchase, which suppliers to source these products from and how much trust we can place in these suppliers. After all, it is these kinds of issues that rightly dominate any Latin American retailer as they look to Asia to source quality products from wholesale suppliers that provide a healthy return in the local Mexican retail market.
Using data to choose suppliers
Researching and understanding manufacturers in Asia is a critical anchor point to Meru’s operation. We know that our customers here in Mexico are placing their trust in us that they are purchasing from suppliers of the highest quality. And so focusing on the production quality and business excellence of the Asian suppliers we partner with is an absolute priority.
Meru’s Supplier QA algorithm ensures we meet the above goals – firstly by applying strict controls on which suppliers we allow onto our marketplace - and secondly by ensuring the products we offer from those suppliers meet integrity, production, and safety quality assurances.
For the supplier QA quality algorithm, we utilize a technology stack built with Amazon Web Services. This cloud based, serverless architecture allows us to feed in data about potential suppliers that Meru might wish to engage with from a huge range of sources. Structured, unstructured, static, and dynamic data sources are all provided to the algorithm. Then, both supervised and unsupervised algorithms are used to, for example, identify those suppliers who exhibit operational behaviour that is statistically more likely to result in high quality manufacture. We use historical data to identify how credit scores and product reviews for each manufacturer have varied with time allowing us to assess how their business has grown and evolved. And we can use these insights to classify and cluster suppliers into graded ranking systems that drive how Meru engages with the supplier.
But moreover, because Meru has a data science, logistic, and management teams also present in China, we are able to leverage sources of data that can only be accessed from inside China itself. For example, in China, each manufacturer must register a “Unified Social Credit Identifier”, which ensures market regulation and adherence to quality and inspection codes. Accessing these datasets helps Meru’s Supplier QA algorithm track each supplier’s response to government quality and regulation inspections. And of course 3rd party private APIs and databases, again only accessible from inside China itself, add yet another layer of data insight that we can use to truly understand the manufacturers we work with.
Data driven supplier assessment
Meru’s proprietary algorithms, executed within AWS Lambda serverless compute architecture, takes all of these many and varied data points and generates filters the metrics we go on to use in grading the manufacturers. We can be as strict or relaxed as required. For example, discounting any supplier who doesn’t provide proper product safety certification. Or discounting suppliers who haven’t provided comprehensive information on their products, as a reflection of their business excellence. So, for example removing from consideration suppliers who have not provided accurate descriptions of their products, or detailed prices and lead times – because this is the very information a retailer here in Mexico needs to know to plan their own business.
We take advantage of a phalanx of over 20 independently acquired metrics against which we can assess a supplier’s business excellence, manufacturing quality, credit score, delivery rate, response rate, product quality, customer review score, and overall trustworthiness. It then becomes a trivial matter to only give further consideration to suppliers who score, say, score a B+ or above in our grading system. By effectively taking this “all you can eat” attitude to data acquisition we can build a comprehensive supplier QA algorithm, which will assess Asian manufacturers against a wealth of metrics and assessments – deciding in as strict a manner as we chose about whether the products of these manufacturers should be considered for our marketplace platform www.meru.com.