functional-art_slideshow_imageOr what the art of cooking and database management have in common

by Frédéric Cavro

Hotdog stand or fine cuisine? Lunchtime special or seafood buffet? – As in the culinary arts, database management is aimed at satisfying the tastes of the target group. There is an enormous range: depending on the customer’s requirements, simple dishes or elaborate creations are produced – with or without topping.

Transferred to the world of database management, this means anything is possible. Starting from standard reporting, progressing to analysis methods, such as online analytical processing (OLAP), then on to forecasts. The information “dish” produced depends on the customer’s requirements and the capacity of the expert database chef. In addition, the quality of the information ingredients is crucial because the data models produced depend on the output from the kitchen, dining rooms and garden of the database manager and customer and the status of the “information hypermarket”. This means data about the environment, purchasing power and geo-information.

Database management operates like creative cooking. “Creativity is involved in putting together the appropriate ingredients to make an exquisite meal. The challenge is identifying which ingredients I need to produce a certain flavor. This equates to identifying which database I should use in the analysis process,” outlines Frédéric Cavro, the Managing Director of the database management specialist, SAZ Marketing, in Frankfurt. Database management first has to deal with objectives and methods:

  • What is the advertising objective? Is it selling? Increasing turnover or raising the company profile etc?
  • How can this be achieved using direct marketing?
  • Which methods and media (with their various opt-in problems) can be used to penetrate the correct target groups?

This is important because, for example, different addressees react differently to mailings. Some prefer snappy self-mailers to remind them of monthly Ikea offers, even though they are already familiar with them. Others want fancy packages and incentives enabling mail recipients to actually arrange a test drive in a new VW cabrio.

Stable Models

When these points have been clarified, the information dining rooms have to be filled – database managers call this ETL processes – derived from extract, transform and load. Here the customer’s data is closely examined: which information is historically and currently available for long-term consideration? Furthermore, it is important to know what previous experience should be used in the data model: has direct marketing already been carried out? If so, using what methods and what were the results? “This is important to guarantee the consistency of the information used in the analysis,” stresses Carsten Fronia, the Marketing Research Manager at the SAZ-IT subsidiary, SAZ smart.net, in Garbsen. Without this consistency – for example the standard formatting of customer age when considering various periods – a model will not be stable later. In addition to the structure of the ingredients, their hygiene must also be up to standard. The basic principle of data mining is “garbage in = garbage out”. This means: if poor ingredients are used, it is no wonder that the meal does not taste very good. Therefore, Cavro recommends that great importance is placed on data hygiene. This starts with existence and plausibility tests and extends to making any corrections which are necessary because end customers have moved or died. As a preventative measure, advertisers should, for example, compare their customer data with all current change of address data available on the market. In special cases, information “chilli” may even be required: if a strategy is aimed at the special case of risk and debt management, enquiries with the residents’ registration office should also be included in data cleansing to determine the customer’s actual place of residence, advises Cavro.

Producing Indicators

Depending on the objectives, additional information can be used in the information kitchen and dining rooms (data model), such as data offered for sale in the “hypermarket”, for example lifestyle information or geo-data. Then database management provides suggestions for operative databases which may contain accounts information, goods in stock, sales data and turnover information. “A top database chef must ensure even at the preparation stage – setting out the kitchen utensils – that he obtains and structures all data which has an influence on the marketing objective. This provides the opportunity to lay the foundations for better results at the preparation stage,” says Fronia. Strengths have to be worked out. An example is the daily press: how does a regional newspaper manage to classify why the sales in certain places are below expectations? Certain variables – they may concern the detail and frequency of reports about certain towns and communities – must be used as possible indicators and considered as a whole. While the top chef fries onions, pours red wine over the duck and juggles with spices, the database manager works on the structure. Fronia calls this dispositive database management. In doing this, using the example of a retail chain, he examines relevant purchasing characteristics of customers: age and place of residence at the time of purchase. Statistical estimates are also required: How is the average customer’s behavior expressed in relation to the place of residence? An important role is played by the distance to the next place of purchase – and thereby the customer’s willingness to drive to the shop which results in increased sales. Depending on the branch network, different assumptions by companies with branches lead to different strategies: i.e. Supplier A’s next shop may be a few kilometers away whilst some cities have a few branches.

A top database chef masters the information equivalent of hotdogs as well as gourmet cuisine. This is like Drugstore Chain B, because, with its strong presence, this promises to result in higher sales. “Chains, for example, want to know in which areas of Germany many of their customers live, so they can make out patterns,” explains Fronia. To do this, they have to relate this to branches with high sales and examine how these companies work together with chains. Are markets cannibalized or do sales increase due to overlapping catchment areas? For example because these branches can be reached more quickly and have a stronger market presence. Fronia: “The findings result in the customer’s strategic decision on how the data model should be adapted.” In the dispositive area, information defined as relevant by database management is linked together. For instance, it links together the place of residence and the sales data of customers using the customer number. Streets or post-code areas can even be allocated to environment structures. Irrelevant information is, by contrast, reformatted or simply ignored.

Bridging the Knowledge Gap

The culinary art is gradually approaching its peak. The wild thyme may admittedly have run out. But top database chefs are people with ideas. They have to consider ingredients with a similar flavour to wild thyme. In database management this means: how can blind spots be overcome by transcription using other variables? We will stay with the example of the branch: as database management does not know the size of the branch, it determines the turnover per business and divides the markets into turnover classes. “To map marketing requirements as accurately as possible, I need several variables to bridge a knowledge gap,” explains Fronia. This applies, for example, when determining soft factors or factors which are difficult to measure, such as customer loyalty. Fronia: “I have to construct an informational aid by mapping measurable factors.” Database management is an art like haute cuisine. Quality manifests itself even before preparation – i.e. when entering the data garden and filling the information stores. Chefs achieve peak performance when they know the tastes of their guests – hotdogs or coq au vin? Database managers also have to know what their customers want. Note: the process differs depending on whether response optimisation or an increase in sales is on the agenda. “In the end, the data model should map reality as closely as possible,” summarises Cavro. In doing this, database managers also have to bridge the gap if certain information is not available. Cavro: “This is where a distinction can be made between the good and bad suppliers on the market. A top database chef can prepare both hotdogs and haute cuisine.”

SAZ Marketing GmbH in Frankfurt specializes in B-to-C applications. This includes database management, address services and list broking as well as address modification and cleansing processes. This is centered on the IQ-Base knowledge database and the smartADDRESS product range. This involves various tools to check, correct and add to addresses, including telephone numbers, information on changes of address and the deceased.

This is the third post in our series Modern Database Management. The previous two articles can be found here.