A supply chain coordinator at a metal components manufacturer in Kayseri tried three different no-code tools over the course of last year. She hit the same wall each time: the tool was deployed, the training was done, but there was no framework for clearly describing what problem actually needed solving. Which data does who see, who owns which decision, which step is being repeated manually — none of this was written down. The tool arrived, sat on top of an empty procedure, and fell out of use within a few weeks. This pattern sits at the blind spot of the citizen developer debate: no-code is not about opening a software library for employees. It is an excuse to teach process literacy. Those are very different things, and most Turkish SMEs are conflating them at cost.When the pandemic forced remote work on production and supply chain operations overnight, the centralized IT model hit a serious bottleneck. IT capacity stayed flat while field unit demand for digital solutions tripled almost immediately: warehouse entry and exit tracking, supplier approval forms, quality control record templates, shift notifications. None of these were complex software. But each required opening an IT ticket, waiting in queue for weeks, and defining a project budget. No-code platforms — Microsoft Power Apps, Google AppSheet, and comparable tools — entered exactly this gap. Among mid-scale Turkish manufacturers, interest in these platforms grew noticeably from late 2020 onward, particularly in firms that already had an ERP system but were struggling with the small operational gaps that ERP could not reach. The problem in the Kayseri case was not the tool — it was the organization that arrived unprepared.The citizen developer concept has been maturing since Gartner first framed it around 2015 and 2016. In the Turkish SME context, it is frequently misread as ‘give workers a tool and let them figure it out.’ This interpretation is both organizationally risky and unfair to the employee. A genuine citizen developer program operates on three layers. The first is problem definition competency: can the employee structurally describe a repeating process, map a data flow, and identify decision points? The second is tool competency: does the employee understand the logic of a no-code platform — triggers, actions, conditions — and know the platform’s limits? The third is governance competency: who approves what the employee builds, who maintains it, what data does it touch, and does it fall under KVKK obligations for data processing? All three layers need to appear in the training program. In most Turkish SMEs, only the second layer gets taught; the first and third are ignored. The result is polished-looking templates that get abandoned within months.An automotive parts supplier in Eskişehir built a supplier price update form on Power Apps at the end of 2020. The form was ready in two weeks, with zero lines of code written to IT. So far, a success story. But six months later, it emerged that no retrospective corrections could be made, that when a supplier submitted multiple revisions it was unclear which value should be treated as current, and that some price data had appeared in a shared workspace visible beyond its intended audience. From a KVKK standpoint, that third issue deserved serious attention. IT eventually stepped in and restructured the entire setup — not just the form the employee had built, but the whole underlying arrangement. This reads like a representative scenario, because the pattern — real speed gain, deferred governance cost — is one I have seen repeatedly in Turkish manufacturing and supply chain firms in the early stages of no-code adoption. No-code platforms do not hide your data from you. But without a governance framework, you cannot track where that data flows either.Does this mean these tools should not be used? No — but before deploying them, an organization should have answered three questions. First: which process types genuinely accept local, decentralized solutions, and which need centralized control? A practical rule: processes that touch ERP data, generate financial records, or serve more than one department need IT-supported development, not no-code. Conversely, single-unit operational tracking processes that live outside the ERP boundary are strong no-code candidates. Second: what share of the employees designated as citizen developers have received even basic training in process mapping or data flow description? If this share is clearly a minority, purchasing platform licences is a premature step. Third: is there a review and approval mechanism for applications that employees build — or is each worker simultaneously the architect, tester, and maintainer of their own solution? If the answer is ‘no mechanism,’ within a year the organization will have accumulated an invisible software inventory of unmaintained, inconsistent applications. In Turkish mid-scale manufacturing, this ‘shadow application’ problem is only now becoming visible. Within a few years it will be the largest cleanup item on IT leaders’ agendas.The urgency of the pandemic period pushed many Turkish SMEs to acquire licences and deploy fast. That response was understandable in context. But by 2021 the work has shifted: it is no longer about selecting a tool — it is about designing a competency program that extracts real operational value from the tool already chosen. The practical steps are concrete. Start by producing a ‘process opportunity inventory’: which repeating, manually executed, ERP-external processes exist in your operations, who manages them, and where does the data go? Then begin with pilot participants trained not on the tool but on problem definition — those who can draw a flow diagram and write down a decision point consistently outperform everyone else in no-code environments. Finally, keep a simple application log: who built what, what data does it touch, who is responsible for maintenance. This log can be five rows in a spreadsheet. It is a discipline question, not a technology question. Had the coordinator in Kayseri applied this sequence before trying her third tool, she would most likely be running a working system today. The tool did not change. The preparation was missing.
This article was originally published in Turkish by Gökhan MERCANOĞLU on May 31, 2021. The English edition has been reviewed and edited by the author.