Mid-to-large retail and financial services companies in Turkey are navigating two simultaneous pressures on customer data. On one side sits the Turkish Personal Data Protection Law, KVKK, which has been in force since April 2018. On the other, for companies doing business with European markets, GDPR enforcement runs without interruption. Both frameworks fundamentally redefined what it means to collect, store and use customer data. The problem is that many companies launching a customer data platform project treat legal compliance as a separate workstream. The legal team works in isolation, the technology team works in isolation, and the result is an architecture that is neither legally defensible nor operationally sustainable.
A customer data platform, or CDP, is a system that consolidates customer data from multiple channels into a unified profile and makes that profile available for segmentation, personalisation and analytics. In retail, point-of-sale data, e-commerce behaviour, call centre records and email interaction converge in a single customer view. In financial services, transaction history, credit application data and mobile app behaviour come together in the same way. This consolidation capability is what makes CDP genuinely valuable. It is also what makes a poorly configured CDP a significant liability under KVKK. The law requires companies to specify the purpose for which personal data is collected, the legal basis on which processing rests, and how long data will be retained. If the CDP architecture does not answer these questions before go-live, the cost of achieving compliance after the platform is operational grows considerably.
The first element that must be embedded at the centre of the architecture is consent and legal basis recording. Under KVKK, explicit consent is one of several lawful grounds for processing, not the only one. Contract performance, legitimate interest and legal obligation each provide separate grounds for different data categories. The CDP data model must reflect these distinctions. In practice, this means that each customer record should carry not a simple consent flag but a granular map showing which data category is processed on which legal basis. In Turkey, many retail companies are currently operating CRM systems that do not make this distinction at all. A CDP migration offers an opportunity to close that gap, but migrating without first resolving ambiguous legacy consent records simply transfers the problem into the new platform.
The second critical design decision is purpose-based processing. Both KVKK and GDPR prohibit using data for purposes beyond those for which it was originally collected. The CDP segmentation engine must respect this boundary. If a customer consented to receive emails under a loyalty programme, that data cannot be used for profile-based advertising targeting without additional consent. Implementing this at platform level means assigning purpose tags to each data set and configuring segmentation rules so they can only access data tagged with a compatible purpose. This is technically achievable, but it requires marketing teams to answer the question ‘why are we targeting this customer’ for every campaign they run. Marketing teams in Turkey are not yet accustomed to this discipline. Technical controls without process design do not hold.
The right to erasure, sometimes called the right to be forgotten, is the requirement most frequently overlooked in CDP architecture. Article 11 of KVKK grants data subjects the right to request deletion of their personal data. For the CDP to honour this request, the platform must know in advance which source systems each customer profile connects to and how a deletion instruction will be triggered in each of those systems. In practice, this means the CDP must operate a deletion orchestration mechanism that covers not only its own data store but also the connected e-commerce infrastructure, email platform and analytics tools. In Turkey, this orchestration is still handled manually in most organisations: when a deletion request arrives, teams intervene in each system separately. This approach both increases error risk and threatens compliance with the thirty-day response deadline that KVKK sets.
Data minimisation is another principle that CDP design teams routinely underweight. The assumption that more data always produces better personalisation is in direct tension with the legal framework. KVKK requires that only data necessary for the stated processing purpose is collected, and that data is deleted or anonymised when retention is no longer justified. In CDP terms, this means retention policies must be defined at platform level and enforced automatically. If a customer has not made a purchase in two years and no active consent or contractual relationship exists, that profile should trigger automatic archiving or anonymisation. Managing these policies manually becomes effectively impossible as data volumes grow. Building retention logic into the platform from the start is far less costly than retrofitting it later.
A company that wants to build CDP and KVKK compliance together should start not with a technology decision but with an organisational one: can the legal, data and marketing teams produce a shared data processing inventory? Without that inventory, neither platform selection nor architectural design can proceed soundly. In the current environment in Turkey, where foreign CDP licence costs represent a meaningful budget line given exchange rate pressure, clarifying architectural requirements before platform selection also prevents unnecessary customisation spend. When consent recording, purpose tagging and deletion orchestration are built into the design rather than added afterwards, the CDP becomes both legally defensible and operationally sustainable. Adding these three elements after the platform is live is, without exception, more expensive than building them in from the beginning.
This article was originally written in Turkish by Gökhan MERCANOĞLU on June 3, 2019 and has been automatically translated into English and other languages using machine translation.