Taking full control

• Article 6 of 6
• Database Marketing, September 2006
• View a PDF of the original article ~ 1.6MB

Page 1 | Page 2 | Page 3 | Page 4 | All 4 Pages

In part, this can be done through technology. A ‘hub’ or bus is typically required for CDI to integrate disparate systems, since it’s unlikely that any organisation is going to rip out its existing infrastructure in favour of a new system. Since data can therefore stay in existing databases, just linked through the hub to other databases, marketing is still able to add fields to records in its own database. The differences will be that the address data either need no longer resides in marketing’s database or will be derived from a cleaned database; and that other departments will be able to access marketing’s data, just as marketing will be able to access their data.

This will be especially true in SOAs, where databases and services will be loosely coupled together to answer specific business needs. “Loosely coupled is how it’s going at the moment,” says Ed Wrazen, VP International of Harte-Hanks Trillium Software. “It’s difficult to get one database that suits the whole business.”

This effectively gives the optimum set-up for both marketing and the rest of the organisation, with marketing able to do live, instantaneous ETL – hopefully, with very little T, if the rest of the organisation is doing its job properly – perform any specific operations it wants on the data it has in its own database, and then give back the results of its work to other departments.

The challenge will be in ensuring that other departments are getting data quality to a level that suits marketing’s purposes and that they’re exposing any data they might have in a way marketing can access.

Although many vendors will argue that board-level sponsorship by a marketing-friendly sponsor will be the best way to ensure this, board-level sponsorship of any IT project is almost impossible. And for longer term projects, the turnover of sponsors in the organisation will mean it unlikely that enthusiasm for the project can be maintained over time.

Instead, a business unit that receives input from other units within the organisation – including marketing – should be in charge of any CDI/MDM project. In particular, the experience of marketing with data cleansing can be of help and should help ensure that the data is in a form close to or potentially even in exactly the state required by marketing for its particular needs.

This approach will avoid some of the classic problems found during data warehousing projects, for instance, which could be seen as one of the first real attempts to do CDI. “Everyone understands the idea and its value, but if you hand ownership to IT, people get annoyed,” says SAS’s Kellett “Pretty soon, you get databases springing up all over the place. You can only solve this by getting people sitting down and talking instead of rushing around.”

Indeed, says Kalido’s Hayler IT should never be placed in charge of this unit. “If it is, it won’t work. They shouldn’t make the judgements about what’s right and they don’t have the clout in the organisation to make the project work.” It’s potentially the finance department that may well end up as ‘data stewards’, he says.

By passing off data stewardship to this near-autonomous unit, most of the concerns about additional work for marketing can be avoided as well. The unit should ensure there are business rules that are applicable to all departments and that everyone can use and modify the data in the CDI without unduly affecting the other departments.

This will require changes in the organisation’s culture: one of the biggest problems in CRM, for instance, was the refusal of sales people to enter data or enter it in the correct format. Says Datanomic’s Tuck, “On any project of this type, people need to understand the value of information. It needs to be evangelised. Otherwise, people won’t be motivated to accurately record data. In call centres where the sole measurement of performance is call volume, the most common names entered are ‘Siebel’ and ‘…’. Rudimentary data verification can fix that, but you need deeper cultural changes. That’s by no means easy – you have to make the case to them.”

Page 1 | Page 2 | Page 3 | Page 4 | All 4 Pages