{"id":14447,"date":"2026-05-11T15:42:33","date_gmt":"2026-05-11T13:42:33","guid":{"rendered":"https:\/\/www.improven.nl\/?p=14447"},"modified":"2026-05-11T15:47:22","modified_gmt":"2026-05-11T13:47:22","slug":"ai-faalt-zelden-door-technologie-maar-bijna-altijd-door-data","status":"publish","type":"post","link":"https:\/\/www.improven.nl\/en\/nieuws\/ai-faalt-zelden-door-technologie-maar-bijna-altijd-door-data\/","title":{"rendered":"AI rarely fails because of technology - but almost always because of data"},"content":{"rendered":"
\n\t\t\n

\u201cAccording to Gartner, 85% of all AI models and projects fail due to poor data quality or a lack of relevant data. Many companies train their (generative) AI models on incomplete, disorganised or outdated datasets, resulting in substandard results.\u201d<\/em><\/strong><\/p>\n

This observation is confronting, but for many CIOs and CDOs it is especially recognisable. The ambitions are great, the tools impressive - and yet real impact often fails to materialise. Not because AI doesn't work, but because the basics are missing.<\/p>\n

AI speeds up what is already there. If data is messy, AI mostly speeds up the mess.<\/p>\n\t\t\t<\/div>\n\t\n

<\/div>\n\t
\n\t\t\n

Why AI projects are stalled<\/strong><\/h2>\n

In practice, Improven sees the same pattern recurring:<\/p>\n

    \n
  • AI is started when technology experiment<\/strong>, independent of strategy and processes<\/li>\n
  • Data quality and definitions are inadequate<\/li>\n
  • Ownership of data is diffuse<\/li>\n
  • The organisation is not included in the use and change<\/li>\n<\/ul>\n

    The result: pilots not scaling up, models not trusted and employees dropping out. AI lingers on promise, not results.<\/p>\n

    Improven's belief: data first, then AI<\/strong><\/h3>\n

    Improven deliberately chooses a different order. Successful AI does not start with the model, but with the data.<\/strong> Our Data and AI proposition therefore focuses on the foundation of AI and data-driven working:<\/p>\n

      \n
    1. Data quality as a prerequisite<\/strong> We bring insight into data flows, definitions and quality and demonstrably improve them. No AI on polluted or fragmented data, but working from one clear truth.<\/li>\n
    2. Data strategy and governance before tooling<\/strong> Together, we determine what role data and AI play in organisational goals. With clear governance: who owns it, which rules apply, how do we ensure privacy, compliance and quality? In this, governance is not a brake, but an accelerator.<\/li>\n
    3. Structure and ownership<\/strong> Data only gets value when it is clear what it means and who is responsible for it. Improven helps organisations make that ownership explicit - across departments.<\/li>\n<\/ol>\n

      Data management as missing link under AI<\/strong><\/h3>\n

      The often overlooked factor in AI success is strong data management.<\/strong> Companies want to \u2018do something with AI\u2019, but do not realise that you first have to get a grip on your data. Data management is more than managing databases; it is an overall picture of vision, organisation and discipline<\/strong> around data. And let that be just the link that makes innovation sustainable.<\/p>\n

      Building blocks of data management:<\/strong><\/h4>\n
        \n
      • Data strategy (direction)<\/strong> - Make sure there is a plan for the use of data<\/strong> in the organisation, linked to business goals. Which data is really important? Where are opportunities for value creation with data and AI? A clear data strategy prevents AI projects from becoming a wild guess<\/strong> become.<\/li>\n
      • Data governance (agreements)<\/strong> - Define who<\/strong> for which is responsible and how<\/strong> decisions about data are made. For example, appoint a data owner per domain and make agreements on privacy and compliance. Clear governance<\/strong> means that there is oversight and steering so that everyone knows what can and cannot be done with data. Improven ensures that governance workable<\/strong> and \u201cwith your feet in the clay<\/strong>\u201d is set up - no thick manuals in the cupboard, but concrete agreements that people can work with on a daily basis.<\/li>\n
      • Data quality & definitions (basics in order)<\/strong> - AI is as smart as the quality of its input. Data management includes the data cleaning, enrichment and monitoring<\/strong>. Important here is that core concepts are unambiguous: if the concept \u201ccustomer\u201d<\/em> in the CRM means something different from what it means in the billing system, then you have noise. One source of truth<\/strong> is the goal, with clear definitions and metadata so that everyone talks about the same data and AI models learn the right patterns.<\/li>\n
      • Continuous process (culture)<\/strong> - Good data management is not a one-off project<\/strong>, but an ongoing journey. Business strategies change, market data evolves, new data sources are added. That is why Improven helps organisations to embed data management as an ongoing process<\/strong> - e.g. through data stewards, periodic data quality checks and employee training. It is about cultivating a culture in which data is managed and improved as a strategic asset<\/li>\n<\/ul>\n

        Why good data management makes all the difference<\/strong><\/h3>\n

        Those who take their data management seriously build a head start. The reward is that AI does not remain stuck as a trial balloon, but grows into a structural value creator<\/strong>. Unlike organisations bogged down in loose experiments, companies with their data in order are able to Roll out AI more widely and achieve lasting results<\/strong>.<\/p>\n

        With Improven's integrated approach - connecting business, data and IT<\/strong> - we ensure that AI initiatives do not founder on data quality or resistance, but are instead supported by a clear data compass. This is how we help organisations make the step from \u2018we want something with AI\u2019<\/em> to \u2018AI that works\u2019<\/strong> - From hype to sustainable benefit through informed data management.<\/p>\n

        Generative AI: sober, targeted and controlled<\/strong><\/h3>\n

        Improven is decidedly realistic about generative AI. The technology offers opportunities, but it is not a panacea. That's why we follow one simple principle: goal before tool<\/strong>.<\/p>\n

        We start with concrete use cases where generative AI really saves time or increases quality. Small, manageable and measurable. At the same time, we recognise the risks: hallucinations, bias, privacy and dependence on source data. We address these upfront - not afterwards.<\/p>\n

        AI only works when people work with it<\/strong><\/h3>\n

        Technology and data are just two sides of the coin. The third is people. Improven guides organisations in adoption, change and embedding in processes<\/strong>. Training, clear agreements and integration into daily work ensure that AI does not remain a loose experiment, but becomes part of how the organisation works and decides.<\/p>\n

        From AI ambition to AI that works<\/strong><\/h3>\n

        The Gartner figures make one thing clear: anyone serious about using AI needs to invest seriously in data. Grip on data, processes and people determines the success of AI.<\/strong><\/p>\n

        Improven helps organisations exactly there. Not by selling AI, but by making it data foundation, governance and organisation<\/strong> in order. Thus, AI grows from hype to tool - and from experiment to structural value.<\/p>\n

        Improven helps organisations move from \u2018we want something with AI\u2019 to \u2018AI works for us\u2019.<\/strong><\/p>\n\t\t\t<\/div>\n\t\n\n\t

        \n\t\t
        \n\t\t\t\t\t\t\t\t\t\n\t\t\t\n\n\t\t<\/div>\n\n\t\t\n\n\t\t
        \n\t\t\t\n\n
        \n\n\n\t
        \n\t\t\t\t
        \n\t\t\t\n\t\t\t\n\n\t
        \n\t\t\t\t\t\t\t\t
        \n\t\t\t\"\"\t\t\t\t\t\t\n\t\t\t\t\t<\/div>\n\t\t\t\t\t\t\t\t\n