Real-time marketing techniques developed during the mid-1990s following the initial deployment of customer relationship management (CRM) solutions in major retail banking, investment banking and telecommunications companies. The intrinsic and prevailing 'heavyweight' nature of the key CRM vendors at this time, who were generally focused on major back-front office system integration projects, provided an opportunity for niche players within the campaign management application arena.

    The implementation of real-time marketing solutions through the late 1990s would typically involve a 10-14 week delivery project with 1-2 FTE expert consultants and often would follow an earlier outbound marketing solution implementation. This relatively lightweight delivery model had obvious attractions within the vendor sales cycle and customer procurement context but was ultimately to prove a disincentive for major systems integration services providers to partner with real-time marketing vendors.

    Real-time marketing solution implementation classically involves the server-side installation of a multithreaded core decisioning application server / interaction transactional-biased schema and supporting client components such as a 'fat-client' desktop campaign studio / rules editor, browser-based marketing user reporting interface and enterprise application APIs such as web services / Java components. Vendors typically will also provide legacy interfaces for COM, sockets and HTTP integration.

    Vendor solution approaches to real-time learning naturally vary but commonly, the underlying models utilize a naive Bayesian probability classifier, recognising that despite their apparently oversimplified assumptions, these classifiers have worked well in many complex real-world situations. To help gain acceptance with in-house specialist data mining stakeholders, the real-time solutions also support external model scores and execution within offer decision making.

    The dotcom 'bust' of 2000 inhibited the further development and implementation of item-based collaborative filtering techniques which, having been incorporated within real-time marketing solutions through the 1990s, should have been immediately attractive to online retailers managing hundreds of thousands (or millions) of products as opposed to a retail bank with a hundred propositions across savings, credit card and mortgage product lines.

    © 2017 Advanced Analytics Association. All Rights Reserved.

    Please publish modules in offcanvas position.