Rafa Merino
Algoritmos de recomendación de ubicación de referencias en almacén
In an environment where speed and precision define logistical competitiveness, the correct allocation of warehouse locations becomes a critical factor. We transform this complex decision into a strategic advantage through intelligent algorithms that learn from the actual behavior of your operations.
Our models analyze consumption patterns, product characteristics, order flows, and layout details to recommend the optimal location at any given time. It's not just about organizing a warehouse: it's about maximizing its potential.
With a personalized recommendation system for your business, we reduce travel distances, increase picking and replenishment productivity, improve service levels, and enhance inventory stability. This digitalization goes beyond data analysis, driving tangible results from day one.
Decision algorithm for inspection in expeditions
We've transformed traditional inspection into a smart, agile, and highly efficient process. Thanks to our advanced classification algorithm, we can accurately predict which shipments require review, optimizing resources and improving operational quality.
We develop models with up to 98% classification capacity (AUC), allowing inspections to be focused where they truly add value. The result is a faster, safer, and more cost-effective process.
What benefits does your organization receive?
Reduction of inspection costs, freeing up resources for higher impact operations.
Error reduction up to 150 ppm, driving operational excellence and customer satisfaction.
Data-driven decision making, without relying on subjective criteria or manual processes.
Fast and scalable implementation, integrable with any existing operating system or digital platform.
Digitization of picking
Transformamos tu proceso de picking en un flujo inteligente, preciso y totalmente trazable. A través de nuestros modelos avanzados analizamos en profundidad las variables que influyen en los errores de preparación —desde factores operativos internos hasta condiciones externas como la temperatura o la demanda puntual— integrando para ello datos de sistemas internos y fuentes externas.
Aplicamos técnicas de analítica avanzada y correlación para identificar qué parámetros impactan realmente en la fiabilidad del picking. Esto nos permite diseñar recomendaciones y automatizaciones que reducen errores, mejoran la productividad del personal y elevan la calidad del servicio al cliente.
El resultado: decisiones basadas en datos, operaciones más eficientes y un picking digitalizado, consistente y preparado para escalar.
Reference correlation matrix in order book
Discover where the real opportunities for efficiency lie in your operation.
Our Reference Correlation Matrix analyzes in depth the actual behavior of your order portfolio to identify which products are ordered together most frequently.
Through an intuitive visual map, where the intensity of the color reflects the degree of association between references, you can detect hidden patterns and make strategic location and slotting decisions with a direct impact on your operating costs.
References with the highest correlation become ideal candidates to be placed together, allowing for less travel, faster preparation, and reduced time and effort on each order.
Ultimately, this tool transforms your data into smart actions that optimize your warehouse, improve productivity, and enhance the customer experience.
Map of frequently used references in warehouse
Optimize your warehouse from the first glance.
Our High-Use Reference Map transforms the way you visualize and manage your locations, allowing you to immediately identify where your most critical products are and how to improve their layout.
In the initial situation—represented on the left—the references are distributed without a strategy based on their frequency of use: the red dots (high use) and blue dots (low use) appear scattered and misaligned with the work routes. This causes unnecessary movement and longer setup times.
After applying our methodology, as shown on the right, high-demand items are relocated to the most accessible and efficient areas of the warehouse. The impact is immediate:
Fewer operator movements
Significant reduction in order preparation time
Smoother and more organized processes
This map not only visualizes the change: it demonstrates how smart digitization turns your data into operational decisions that generate real savings and sustained performance.
Analysis of reference movement distribution
Understanding how each item moves within the warehouse is key to transforming logistics operations. Our movement distribution analysis reveals, at a glance, which products experience the most activity and which have more sporadic patterns.
Based on the ordering of the references according to their number of picks, we identify patterns that allow us to apply highly personalized location strategies: highlighting the items with the highest turnover, optimizing routes and freeing up space where it really contributes to efficiency.
The result is a strategic vision that drives more precise decisions, reduces operational times, and enables intelligent storage policies based on real data. Because digitization isn't just about measuring: it's about uncovering hidden opportunities and turning them into a competitive advantage.
Hypothesis testing to identify causes of deviations in the logistics process
In an increasingly complex logistics environment, deviations do not occur randomly: they have specific causes that can be measured, compared and corrected.
Our hypothesis testing service allows you to scientifically validate which variables truly influence your operational results.
Through advanced data analysis, we compare the behavior of different populations—orders, routes, suppliers, and operational centers—to detect significant relationships between variables and outcomes. This allows us to distinguish noise from real signals and turn insights into actionable conclusions.
In the example shown, we identified a strong correlation between lead time and certain ordering patterns, revealing operational behavior that had gone unnoticed by the team. With this type of analysis, your organization can:
Discover what factors explain delays, cost overruns, or errors.
Prioritize improvements with a direct impact on the service.
Make decisions based on evidence, not perceptions.
Optimize processes under a truly “data-driven” approach.
Digitizing is not just about automation: it's about understanding what is happening and why it is happening.
With our hypothesis testing, we transform your data into clear and actionable knowledge so your operations can move forward with precision.
Development of a predictive algorithm
Our process for creating predictive algorithms combines advanced analytics techniques, machine learning, and deep operational knowledge to anticipate behaviors, optimize resources, and reduce uncertainty in your supply chain.
The visualization shows the comparative accuracy of different predictive models. Using a box plot, we evaluate their performance and stability, allowing you to identify which algorithm offers the best reliability for your operation. This rigorous analysis is the foundation for selecting, fine-tuning, and implementing the solution that maximizes value for your business.
We integrate these predictive capabilities into your daily processes so you can forecast demand, anticipate needs, avoid inefficiencies, and make decisions based on real data, not intuition. Your supply chain, smarter and future-proofed.
Preparing data for sales prediction
Before building any predictive model, we refine and structure the data to ensure reliable and actionable forecasts. Our process combines advanced analytics and expert judgment to detect patterns, seasonality, and potential outliers that could distort actual demand.
Through dynamic visualizations—such as heat maps of sales increases by period and product—we identify atypical behaviors and opportunities to improve data quality. The result: a solid, clean, and consistent foundation that allows your models to accurately anticipate trends, reduce uncertainty, and enhance business and operational decision-making.
We digitize your data so your forecasts are smarter, faster, and more profitable.
Testing sales predictive models
We evaluated multiple demand forecasting approaches, combining established statistical techniques with advanced artificial intelligence models. We tested everything from classic algorithms like ARIMA and Holt-Winters to deep learning architectures such as CNN (univariate and multivariate) and LSTM, capable of capturing complex and seasonal patterns.
After an exhaustive phase of experimentation and validation, we selected an optimized hybrid model, designed to maximize accuracy and improve demand forecasting. The result: a more robust predictive capability, enabling more confident planning, reduced stockouts, and strategic decision-making based on reliable data.
Prescriptive algorithms for stock levels and replenishment
From a digital platform we adjust safety stocks in a personalized and dynamic way, achieving a perfect balance: better service with lower inventory levels.
Replenishment orders are generated automatically based on recent consumption histograms, incorporating both demand trends and seasonality. This allows us to anticipate actual needs, reduce overstock, and optimize product availability at all times.
With our solution, your company benefits from agile, accurate, and data-driven management, transforming logistics into a tangible competitive advantage.
