About Versible

Versible was born from the realisation that something is getting missed in our increasingly specialised world. Data Scientists. Data Engineers. Business Analysts. Strategy Consultants. Marketing Managers. Centres Of Excellence. Delivery Managers. Data Governance. Cloud Platform Engineers. These specialisms are necessary, but to deliver complex projects they create a lot of handoffs between individuals. Versible aims to minimise those handoffs by concentrating skills and knowledge within single people. The result is an ability to take-on and deliver complex work that large team can struggle with .

Versible Ethos

The Done Thing on a company website is to list virtues: Hard work. Dedication. Deep expertise. Focus on outcomes... But we think you will gain a lot more from reading some short point-of-view pieces on topics that highligh how Versible thinks and tackles challenges.

Deep Agile

Agile is more than how to arrange a to-do list. Agile is about a mindset. It runs beyond how you organise, to what your delivery culture is. If you appreciate this, how projects are planned becomes just a detail.

Read more...

Ones and Noughts

While The Cloud has revolutionised computing, fundamentally nothing has changed. Data is still ones and noughts which are stored and processed through logic gates. Compared to 20 years ago, we have orders of magnitude more computing power immediately on tap, and cloud services to make everything simpler. So why doesn't it always feel like this?

Read more...

Scaling Complexity

Why does delivering complex data applications take so long / cost so much / need so many people? Is this because we intuatively assume delivery effort scales linearly with complexity? But this mis-understands the nature of data application complexity, which stems from the relationships between parts of the solution, more than the number of parts per se. A solution with twice as many parts could have quadruple the number of relationships which drive the amount of time/cost/people needed. Recognising this forces us to challenge some deeply held concepts for how delivery teams should work.

Read more...

States of Coding

Ace coders can effortlessly create complex solutions using all the bells & whistles of their chosen language. Maybe. But really, the very best coders write code that just looks simple.

Code should be written primarily for other humans to easily understand. Suppress any temptation you might have to signal your personal expertise by writing complex code.

Read more...

Versible Models

For more than 20 years, the people behind Versible have been delivering complex data solutions. Often the same analytical and data modelling challenges come up again and again. A common reaction would be to treat these recurring models as core Intellectual Property to be protected. We disagree. We are happy to make these models source code available (which is not the same as free to use - read the LICENSE document). The instinct to protect these models inside a proprietory black box is short-sighted. The value is not the source code itself, but in how that source code is applied to solve real business problems in an impactful way.

These models are largely based on grocery retail, but can be applied to other settings which share similar product purchasing characteristics.

Customer Segment Bias

A model to measure product positioning by the type of customers that buy that product.

See the code on Github 🔗

Product Loyalty

A model for how loyally a product is purchased. This metric can be interpretted and used several different ways to help inform a variety of business decisions. For example, level of repeat purchase, or the importance of a product to the customers that buy it.

See the code on Github 🔗

Similarity & Role

A powerful approach to measuring the relationship between pairs of products. This uses probabilities to compare observed versus expected purchasing combinations. Although product level relationships is the most common use of this technique, the approach can be applied to any product-like entity like brand, category or department.

See the code on Github 🔗