Digital procurement only works if employees actually use the systems. Yet this is precisely where things often fail – especially in indirect procurement, where many requesters only place orders occasionally. What is therefore crucial is less the introduction of a solution than its everyday use. For this to succeed, applications should be easy to understand, respond quickly and lead to completion without detours. If, on the other hand, interfaces are overloaded and contain too many fields, the font size is too small or the application responds sluggishly, users quickly switch back to e-mail or verbal requests.
Adoption should therefore be anchored in the project as a measurable target figure – not only as the number of logins, but above all as the decline in informal channels and the reduction in time per transaction compared with the baseline. The web metric Interaction to Next Paint (INP) can serve as an up-to-date benchmark. It measures how quickly an application processes user interactions and reacts visibly. An INP value of ≤ 200 milliseconds is considered good and ensures that interactions are acknowledged immediately for users – for example through a brief inline status message, a loading indicator or a confirmation notice [1]. If this threshold is exceeded frequently, the tendency to abort processes also increases.
In implementation, it is advisable to establish a robust baseline before go-live – in other words, before the point at which the solution is used productively. Among other things, this can describe what proportion of requirements has so far been handled by e-mail, how many minutes a requirements capture takes on average and how often procurement has to follow up or correct. This is followed by optimisation in short iterations, each with clearly formulated hypotheses, defined measurement points and documented release notes. Those who continuously monitor interaction latency via the INP value can detect at an early stage when the application reacts with delays to interactions – for example with dynamic form fields or live validations – and can take appropriate countermeasures.
Intelligent suggestions and real-time compliance – levers for acceptance in the process
Beyond interface design, AI-supported suggestion logics reduce the cognitive load: product group/material group, cost centre, delivery address and other mandatory attributes are suggested in a meaningful way, shortening the search and reducing follow-up queries. An example from day-to-day laboratory work: if a requester enters “pipette tips”, the system suggests the appropriate product group and adopts the standard delivery address and cost centre from the profile. The requester then only has to check and confirm. The correct assignment of product groups in particular controls many rules in the ERP system and is a frequent stumbling block.
The second lever is real-time compliance: budget, authorisations, contract bindings or pricing logics are checked directly during entry against the rules stored in the ERP – a core principle of real-time procurement. For users this means: in the event of a budget overrun, a clear note with options for action appears; if a supplier is blocked, the appropriate contract supplier is suggested. Incorrect paths are ruled out early, approvals run on the basis of valid specifications, and the touch rate decreases.
Performance and feedback go hand in hand. Every validation is an interaction and should deliver a visible signal within the defined response corridor, i.e. the maximum permissible response time. Only in combination with precise, understandable feedback do technical optimisations have their full effect. These include unambiguous error messages, positively confirmed fields such as “address verified” and short progress indicators for longer checks.
Guided intake and assistants – paths to complete data capture
Unstructured requirements are a reality in indirect procurement and are largely responsible for the lack of time resources in procurement departments. It is therefore crucial to make them suitable for system processing. A guided intake displays only relevant fields, ensures that all mandatory fields are completed and automatically enriches data – for example product group, contract or cost centre. This turns e-mails into qualified transactions with high rule compliance and a low touch rate. Dialogue-based assistants can complement this path: they can ask context-related questions, explain guidelines in clear language and suggest values. In this way, they simplify workflows and promote system usage – provided that governance and reliability are clarified.
Search and input aids must also noticeably facilitate the flow. Autocomplete and automatic address suggestions streamline forms when suggested values can be adopted directly and easily adjusted if necessary. This also reduces typing effort and errors. If suggestions are inflexible or feedback is unclear, the mental load increases – the result is more follow-up queries and abandoned processes.
Measuring success – from go-live onwards
Adoption can be measured in three categories. Firstly, the usage rate in the target group: what proportion of addressees – generally employees in companies and organisations – handle requirements via the system instead of by e-mail or verbal request. Secondly, the displacement of informal channels: to what extent mail and ticket volumes decline in favour of rule-compliant transactions. Thirdly, the time and touches per transaction: how many minutes elapse from the start of the requirement to approval, and how many manual interventions are necessary. In addition, a standardised usability indicator such as the System Usability Scale (SUS) can be used to compare perceived usability across releases; however, the decisive factor remains the link with hard process indicators.
Conclusion
High user adoption is not a product of chance, but the result of clear user guidance, AI-supported data enrichment and real-time compliance checks against ERP rules – underpinned by stable performance and cross-channel orchestration via guided buying. To this end, BeNeering [2] relies on an ERP-proximate architecture with real-time access to the system of record and provides a central entry point with guided buying that leads requesters to the appropriate path – catalogue, framework agreement, stock, marketplace or free text with intake – and through the entire ordering process. An intake assistant also supports the structured processing of inputs and free-text requirements and ensures correct, complete data.
Those who link these principles with a clean measurement set-up (usage rate, displacement of informal channels, time & touches) and improve them in short iterations achieve high self-service and no-touch rates in areas that can be standardised. The combination of guided buying, intake assistant and real-time procurement thus relieves the operational procurement function and makes digital procurement noticeably easier in day-to-day work – for occasional users as well as for experienced users.