Geospatial Analyses and Visualisations on Maps

Many aspects of business are related to location. It is important where the customers come from, how much time it takes them to get to your outlet, where the competitors are located, where it is worth opening new outlets. The neighbourhood and demographic potential of the areas around the points of sale is important. When analysing the location, we calculate the time of travel taking into account the road conditions depending on the time and day of the week. We use various sources of data on demography and income. We take into account infrastructure development plans, new housing investments, population migrations. A natural form of communication of the results of such analyses are maps. The experts of Data Science Logic use interactive maps in the html format to easily visualise phenomena relevant for managers. This enables a better and more intuitive understanding of their impact on business and making the right strategic and tactical decisions. The map visualisations can also be part of interactive dashboards.

Clients: Leading marketer from retail industry, Marketer from retail industry, FMCG company

Optimisation of Catalogue and Leaflet Distribution Areas

Despite the constant increase in importance of e-commerce and electronic channels of customer contact, many companies still devote a significant part of their budgets to traditional forms of communication such as flyers, catalogues or in-store leaflets.

Their preparation, printing and distribution entails significant costs. It is therefore important to measure the effectiveness of activities based on printed materials and optimise the use of the budget allocated to such activities. In other words, you have to answer the question: what is the ROI of the activities carried out and where to distribute, when, how often and what is worth distributing to make the ROI as high as possible with a limited budget.

By analysing data from existing distributions and combining them with data from cash-register systems, transactions recorded on loyalty cards, website traffic and data characterising the demographics of distribution areas, it is possible to build a model estimating the impact of printed materials on sales and other KPIs relevant to the company’s business (e.g. new customers registering in the loyalty program).

With the help of the model created, it is possible to predict the impact of a specific type of material at a specific place and time and to carry out reliable simulations of what would be the effect of abandoning planned activities. In other words, it is possible to simulate what would happen if the costs were reduced and the in-store leaflets were not distributed in a certain area. Would the decrease in revenue and margin prove to be higher than the savings on printing and distribution costs? It is also possible to identify areas where distribution would be more beneficial.

The effect of the optimisation is a list of areas with suggested expenditures, in which it is worthwhile to carry out the distribution. The prepared list is always accompanied by a map, which helps to better understand the recommendation. The optimisation can be carried out at the level of a municipality, postal code or smaller defined areas (e.g. sectors used operationally by companies distributing unaddressed mail).

For additional synergy effects, the analysis of distribution of printed material should be combined with a more comprehensive optimisation of other marketing channels. It allows for the recognition of interactions between different channels and the identification of situations in which different channels strengthen their impact and those situations in which distribution channel cannibalisation is observed. This allows you to answer questions such as: whether the in-store leaflets are more effective in combination with internet advertising, geotargeted for a given area, or whether it is an unnecessary duplication of expenses and it is more effective to separate these activities in space and time.

Clients: Leading marketer from retail industry

Marketing Campaign Optimisation

Maximise the ROI of each PLN spent on marketing

As budgets for marketing campaigns grow, the question of how well these funds are being spent and whether they could be better invested becomes increasingly important. Measuring and evaluating the effects of the campaigns is therefore becoming a necessity. By combining data from multiple sources and using analytical tools, you can track which activities translate most into conversion to purchase and return on investment.

The greatest possibilities of measurement and optimisation are offered by mailing within the framework of direct marketing. Knowing the addressee of the message and being able to observe their later behaviour (it is worth doing so, both for online and offline business), let’s you connect the marketing action with the consumer’s later reaction.

The scoring models can help you to optimally select the promoted product range, message content and the moment of sending to maximise the probability of conversion and the cart value. In the case of periodical campaigns, it becomes important to manage the frequency of mailings, so as not to tire customers and not to evoke the feeling of “spamming” their mailboxes. Similar tools can be used for sending paper catalogues and marketing mailings. Due to the higher unit cost of traditional mailings, the optimisation of the recipient list using scoring models can bring even higher benefits compared to electronic mailings.

The scoring models based on the analysis of geospatial data can also support the processes of optimising the distribution of leaflets and unaddressed mail.

Digital advertising optimisation is possible with the use of advanced attribution models , which make it possible to estimate the impact of individual media and touch-points on conversion (also in terms of value).

For ATL activities in “traditional” media (TV, radio, press) econometric models can be beneficial. The conclusions based on such models help to decide on the size of budgets and their distribution among different channels. They also enable simulation of various scenarios and predicting sales reactions to changes in the amount and allocation of advertising budgets.

The promotional actions are an important aspect of marketing activities. Such mechanisms always entail a cost for the company, whether in the form of lost margins (discounts and price reductions) or products given “for free”. Measuring the actual return on such investments is usually a big challenge. Various promotional activities overlap. The phenomena of cannibalisation inside and between product groups are observed. What is more, the promotion may shift some of the demand from future periods, thus cannibalising the sales in subsequent weeks. Depending on the industry and product range, the strength of individual effects may vary. Nevertheless, the assessment of the effects and success of a promotion based only on the increase in the sales volume over its duration is insufficient and may lead to wrong conclusions, which in turn lead to sub-optimal decisions and allocation of the marketing budget. A multi-faceted analysis of the real impact of promotions on business is actually possible only on the basis of advanced analytical methods. The core is always to ‘clean up’ the sales generated during the promotion period from the sales which, without the promotion, would include other products from the same product group or would be generated in future periods. In turn, increased sales of goods from complementary categories should be added (the so-called halo effect). This is the only way to obtain information about actual additional sales as a result of the promotion (incremental sales).

The models built according to appropriately selected methods can estimate the strength and direction of the promotion’s impact, taking into account additional factors, such as promotions taking place at the same time for a different product range, seasonality, price flexibility, advertising support or the actions of competitors. The model can be used as a basis for the development of simulations of different variants of the promotional calendar and its optimisation not only for one specific action but for the whole sequence of actions properly planned in time.

Clients: Leading marketer from retail industry

Anti-churn Analyses (Customer Churn)

Keep your most valuable customers who are considering leaving you

It is difficult and expensive to acquire new customers. In the majority of mature industries, when the market is not growing as dynamically, you have to drag customers away from your competitors, which often involves offering large discounts and other types of costly benefits, and spending on advertising. In this situation, it is therefore crucial to keep your own customers and prevent them from going to the competition. By using predictive models, it is possible to predict which customers are most at risk of churn at a given time (e.g. during the next quarter). With this knowledge, appropriate retention campaigns can be directed to endangered customers.

Later, with the analytics, it is possible to measure the effectiveness of such campaigns and estimate their ROI. In combination with models that estimate customer value, it is possible to prioritise the retention of the most valuable customers, whose loss would be the most significant burden on the company’s budget. It is also possible to allocate the budget for retention measures accordingly. For example, the most valuable customers can be targeted with a telephone campaign supported by an attractive discount offer, while less valuable customers can receive the emails. The key to the success of such a strategy is to forecast the risk of churn as precisely as possible. This is possible thanks to predictive models, as customers usually send signals (sometimes quite subtle) that indicate an increased probability of churn. The more data sources can be used by the models, the more accurate the forecasts (e.g. data on complaints and contacts with the customer service are of great value), but often even the transaction data themselves are a good basis for building models.

The analysis of customer churn makes it possible not only to identify those most at risk but also to understand the factors that increase the probability of churn (e.g. purchase of a specific product or service, visit to a specific point of sale, too long waiting time for a complaint to be processed). This allows you to identify areas within the company where changes are necessary to increase consumer satisfaction and retention.

The phenomenon of churn usually refers to businesses based on subscriptions or contracts (for example, telephone subscriptions), but it can also be defined and analysed for relationships based on irregular transactions. Advanced analytical methods can, in such cases, answer the question of what is the probability that the consumer without a transaction in the last 9 months will return to the store. The answer depends, of course, on the industry, but also on the history of a particular consumer and the moment in their life cycle (the identification of which is possible thanks to customer life cycle models). It is worthwhile to start with an analysis of basic data sources and gradually expand the tool, increasing the accuracy of forecasts and deepening understanding of the phenomenon.