ALCHEMAI CUSTOMER
Not all customers are created equal - your customers are motivated by different things. In addition much of your value comes from a subset of your customers. Understanding customer should be fundamental to any business strategy, after all your business is your people. Here we outline our approach.
We have found that businesses (often through no fault of their own) are designed to focus on products as opposed to the people purchasing the products. But our customers are not equal – different things motivate different people (B to C) or businesses (B to B) and every customer who uses your product or services intersects with your company in a unique way. This is why we make customer the centre of our analytics.
Traditional approaches here involve segmentations into different customer types – behavioural segmentations, RFM segmentations and demographic segmentations. These are often useful for marketing purposes. However in digital, top tech companies are able to segment customers in a much more granular way, indeed Amazon, Meta and big tech companies focus on the minutiae of individual customer behaviour in determining which messages to optimise.
We perform the equivalent on your data, maximising the value often overlooked ‘First Party Data’ which you possess while combining this with the best third party data.
Our Alchemai Customer Product uses data engineering and machine learning to integrate many data sources into a tactical single view of customer. We then perform machine learning to develop a suite of propensity models and scores which in turn feed into reporting and scenario planning front end tools.
Data ETL
We ingest your (encrypted) first party data which often will include a transactional dataset (which includes key dates such as the date a customer leaves or joins), a customer dataset and a series of product related datasets. This step often involves significant data wrangling to bring disparate sources of data together. We employ an 80:20 approach to this, developing a tactical prototype very early on.
Part of this process involves washing your first party data with third party data sources – this enables us to enrich your data with extra segment level information. Some of this is freely available, some of this is brought in through APIs with our partners.
The result is a powerful modelling dataset which is carried forward into the modelling.
Machine Learning and Predictive Modelling
We then apply our cutting edge algorithms to the model dataset in such a way that we can rapidly run hundreds of models and test their accuracy, while exploring bespoke situations such as specific years, specific customer segments and specific products linked closely to your hypotheses.
The result is a multi-propensity score applied to all of your customers in terms of their likelihood to perform certain actions (leaving, buying a new product, upsell) in addition to an ideal customer profile for new acquisitions.
Visualisation and Strategic Decision Making
The front end of our product is a scenario dashboard which shows both what happened historically (the diagnosis) and the likelihood of what will happen going forward. Part of this can be a customer level next best action as well as an optimisation solution to maximise customer value.
This is always done in conjunction with the strategic decisions you are looking to answer – the key to any analysis is its actionability and we seek to ensure what we do gets implemented in your business and helps in its transformation.