Procurement Automation: Technology Limits Without Process Redesign

Context

The procurement software market grew by 14% annually between 2019 and 2023, according to Gartner. Vendors promise 20-30% reductions in acquisition costs through automated approvals, electronic quotations, and ERP integration. However, evidence from real-world implementations reveals a different pattern: companies that automate purchasing processes without prior redesign achieve only marginal gains—often below 5%—and face persistent organisational resistance.

The reason is structural. SME procurement automation digitises the existing workflow. If that workflow includes redundant approvals, poorly defined categories, unconsolidated suppliers, and inconsistent specifications, the software simply replicates inefficiency at greater speed. A purchase request that required four manual approvals over three days now passes through four electronic approvals in three hours—but still requires four approvals.

This article examines why the promise of automatic efficiency fails without prior intervention in procurement architecture. It analyses evidence from implementations in European SMEs, identifies the causal mechanisms explaining the gap between expectation and outcome, and explores the implications for decision-makers under pressure to "digitise purchasing" without clarity on what this entails.

The issue warrants in-depth analysis because the mistake is costly. Procurement software for SMEs costs between €15,000 and €80,000 in annual licences, plus implementation consulting. When the tool fails to deliver value, organisations rarely revert—they become locked into a system that no one uses correctly, with incomplete data and hybrid processes (part digital, part manual) that are worse than the previous state.

The State of the Evidence

The Hackett Group published a longitudinal study in 2022 involving 340 companies that implemented e-procurement platforms between 2018 and 2021. The central finding: companies that automated without process redesign achieved an average reduction of 4.2% in total acquisition cost, compared to 18.7% for those that redesigned first. The difference is not marginal—it is an order of magnitude.

McKinsey reported in 2021 that 60% of digital procurement implementations in European companies with fewer than 500 employees fail to meet initial savings targets. The report identifies three recurring causes: incomplete or duplicate supplier data (78% of cases), lack of category taxonomy (65%), and unchanged approval processes (82%). None of these are technical problems—they are governance issues that software exposes but does not resolve.

Research from the MIT Center for Transportation & Logistics (2020) shows that procurement automation delivers value in specific contexts: purchasing categories with standardised specifications, consolidated suppliers, and volumes that justify structured negotiation. For ad hoc, low-value, or variable-requirement purchases, automation adds friction without reducing cost. The study documents cases where processing time increased after automation, as users bypassed the system to maintain agility.

APICS (Association for Supply Chain Management) published 2023 data from 180 European industrial SMEs. Companies that implemented category management before automating reported adoption rates of 85% in the first six months. Those that automated first reported 34%. The difference is explained by clarity: when categories are defined, users know where to classify each purchase; when they are not, every request generates doubt and exceptions.

Data from INE (Survey on ICT Usage in Companies, 2023) shows that 23% of Portuguese SMEs with more than 50 employees use e-procurement software. However, only 8% report full integration with financial management systems, and 12% maintain parallel manual processes for "special cases"—which account for, on average, 40% of purchase volume. This suggests that most implementations are partial, limiting returns.

The literature converges on one point: SME procurement automation delivers value when processes are designed to be automated. When they are not, the tool becomes a record-keeping system—it documents transactions but does not improve decisions. The question is not whether to automate, but what to do before automating.

The Mechanisms

Digitising Approvals Without Redesigning Authority

The most common failure mechanism is the digital replication of approval flows designed for analogue contexts. In a Portuguese industrial SME studied by COTEC in 2021, purchase requests above €500 required approval from the department director, CFO, and CEO. The rule was historical—set when the company had 30 employees and the CEO knew all suppliers.

When the company implemented procurement software, it digitised the flow without questioning the rule. The result: the CEO began receiving 40-60 approval notifications per week, most for consumables or routine maintenance. The rejection rate was below 2%. The system added traceability but not value—the CEO became a digital bottleneck instead of an analogue one.

The solution requires redesigning authority based on risk and category. Purchases in pre-negotiated categories, with approved suppliers and within defined limits, do not require individual approval—they require periodic audits. Out-of-category purchases or those above material thresholds warrant scrutiny. But this distinction must be in the process design, not just in software configuration.

Research from the University of St. Gallen (2022) documents that companies redesigning authority before automating reduce the average number of approvals per request from 3.2 to 1.4, without increasing risk. The key is to replace transactional control with category control: instead of approving each purchase, approve the category strategy (suppliers, specifications, limits) and audit compliance.

Lack of Category Taxonomy

Procurement software assumes purchases are organised into economically logical categories. Yet most SMEs organise purchases by requesting department or historical supplier, not by spend category. When automation is implemented without taxonomy, each user classifies purchases differently—"office supplies" in one department is "consumables" in another—and the database becomes unusable for analysis.

APICS (2023) data shows that SMEs with implemented category taxonomy can consolidate 60-70% of spend into 8-12 main categories. SMEs without taxonomy show dispersion: 40-50 "categories" created ad hoc, many overlapping. This fragmentation prevents structured negotiation—it is impossible to identify total volume in a category to negotiate with suppliers.

The cost is direct. A distribution SME with €8M in annual purchases, studied by IAPMEI in 2022, discovered after software implementation that it was buying cardboard packaging from 14 different suppliers, with prices varying by 35% for equivalent specifications. The information was in the system—but no one saw it, because "packaging" was classified as "warehouse materials", "logistics consumables", and "other".

Category redesign is not an academic exercise. It requires analysis of historical spend (12-24 months), grouping by functional and economic similarity, and assigning category owners. Categories should have sufficient volume to justify active management (rule of thumb: minimum €50,000/year) and clear boundaries to avoid overlap. Only after this structure is defined does automation add value—by enabling spend tracking by category, supplier comparison, and consolidation opportunities.

Fragmented and Unvalidated Supplier Data

Procurement automation depends on structured supplier data: tax ID, address, payment terms, product categories, performance history. Yet most SMEs keep this data in multiple systems—accounting, ERP, departmental Excel sheets—without a validation process. When software is implemented, data migration exposes duplicates (same supplier registered with slightly different names), outdated information (contacts no longer valid), and gaps (suppliers without assigned category).

Research from Universidade Católica Portuguesa (2021) documents that industrial SMEs have, on average, 18% duplicate supplier records and 31% of records not updated in the past three years. Automating on this basis generates errors: quotation requests sent to wrong contacts, approvals blocked because the supplier lacks a category, spend reports double-counting the same supplier.

The solution requires data cleansing before automation. Typical process: export all supplier records, identify duplicates (fuzzy matching algorithms detect name variations), validate critical information (tax ID, IBAN, contact), assign categories, and flag inactive suppliers. Manual, intensive work—but unavoidable. Companies that attempt automation without this cleansing face months of reactive corrections, with users losing confidence in the system.

Hackett Group (2022) data shows that companies investing 60-80 hours in supplier data cleansing before go-live achieve adoption rates 40% higher in the first six months. The time is not wasted—it is a prerequisite for automation to work.

Incomplete Integration with Financial Systems

The value of SME procurement automation depends on integration with ERP and accounting. When integration is complete, the purchase request automatically generates a purchase order, which triggers warehouse entry and accounting entry. When integration is partial, each transition requires manual intervention—and the system becomes slower than the previous process.

IDC (2023) data shows that 55% of e-procurement implementations in European SMEs have partial integration: requests and approvals in procurement software, but invoice entry and payment in separate systems. This creates duplicated work (same information entered twice) and risk of error (discrepancies between systems requiring manual reconciliation).

The problem is often technical but also organisational. Full integration requires APIs between systems or middleware, adding cost and complexity. But it also requires process alignment: if procurement uses one approval flow and accounting another, integration does not solve the issue—it exposes inconsistency. Companies discover, after go-live, that they must redesign processes on both sides for integration to work.

McKinsey (2021) documents that companies treating integration as a technical project fail in 70% of cases. Those treating it as a process redesign project—mapping end-to-end flows, identifying friction points, and redesigning before integrating—achieve a 65% success rate. The difference lies in recognising that integration is not software configuration; it is organisational alignment.

Lack of Category Governance

Even with taxonomy defined, automation fails if there are no category owners with authority to manage suppliers, negotiate terms, and enforce compliance. In an SME without category governance, each department continues buying from familiar suppliers, even when better alternatives have been centrally negotiated. The software records the transaction but does not change behaviour.

APICS (2023) research shows that SMEs with dedicated category managers (even part-time) can consolidate 40-50% of spend into negotiated contracts within the first 12 months. SMEs without category managers maintain dispersion: 80% of spend in spot purchases, with no leverage of volume. The difference is not in the tool—it is in the responsibility structure.

The challenge for SMEs is scale. Companies with €10-30M in annual purchases cannot justify full-time category managers for all categories. The solution is prioritisation: identify 3-5 categories representing 60-70% of spend, assign owners (department heads with additional responsibility), and define category KPIs (savings, supplier consolidation, compliance). Residual categories remain decentralised but with clear rules (approved suppliers, spend limits).

The Portuguese Case

Portugal has specificities that amplify the challenges of SME procurement automation. INE (2023) data shows that 94% of Portuguese companies have fewer than 10 employees, and only 1.2% have more than 50. This fragmentation limits procurement economies of scale—company purchase volumes are small, bargaining power is low, and investment in structured processes is hard to justify.

The Bank of Portugal (Economic Bulletin, 2023) documents that Portuguese industrial SMEs have an average EBITDA margin of 8.2%, below the European average of 11.4%. Margin pressure makes procurement critical—every percentage point of cost reduction directly impacts profitability—but also limits capacity to invest in systems and processes. Companies face a trade-off: invest in automation or maintain operational flexibility with manual processes.

AICEP (2022) data shows that Portuguese exporting SMEs have more internationalised supply chains than the European average: 38% of raw material purchases come from outside the EU, compared to 28% in Europe. This internationalisation adds complexity to procurement—different currencies, long lead times, customs requirements—that standard software does not always accommodate. Companies automating without adapting processes to this reality face operational friction.

IAPMEI (SME Report 2023) documents that only 15% of Portuguese SMEs with more than 50 employees have a dedicated procurement function. In most, purchasing is the responsibility of production, operations, or finance directors, who manage procurement as a secondary activity. This lack of specialisation limits redesign capacity—there is no internal knowledge of best practices in category management, structured negotiation, or supplier management.

But there are signs of change. COTEC (2023) data shows that industrial SMEs participating in procurement mentoring programmes (COMPETE 2020, Portugal 2030) managed to reduce purchasing costs by an average of 12% in the first year, through supplier consolidation and structured negotiation—without automation. This suggests significant room for improvement through process redesign, and that automation can amplify these gains if it follows, not precedes, redesign.

The Portuguese experience reinforces the central argument: automation without redesign digitises inefficiency. But it also shows that redesign is possible even with limited resources—it requires focus on critical categories, use of external expertise (consulting, mentoring), and an incremental approach. Companies attempting to automate everything at once fail; those that redesign one category at a time, automate when the process is stable, and scale progressively, have a higher probability of success.

Management Decisions

Decision-makers under pressure to "digitise procurement" should start with a diagnostic question: what problem are we trying to solve? If the problem is lack of spend visibility, the solution may be analysis of existing data (accounting extracts, purchase orders) before investing in software. If the problem is approval time, the solution may be redesigning authority, not automation. If the problem is supplier pricing, the solution is structured negotiation, which requires volume consolidation—automation helps, but does not replace this.

A structured approach begins with spend mapping. Export 12-24 months of purchases from existing systems (ERP, accounting), classify by functional category, identify top 10 suppliers per category, and calculate price dispersion for equivalent products. This exercise—20-30 hours of analytical work—reveals where the opportunity lies. If 60% of spend is concentrated in 3-4 categories with multiple suppliers and price variation above 15%, there is room for consolidation. If spend is dispersed across dozens of small categories, automation will have limited return.

Second step: redesign critical processes. Choose 1-2 high-impact categories (significant volume, multiple suppliers, recurring purchases), map the current process (who requests, who approves, supplier selection criteria), identify inefficiencies (redundant approvals, lack of specifications, absence of negotiation), and redesign. The new process should have risk-based approvals, clear specifications, approved suppliers, and a category owner. Test for 3-6 months before automating.

Third step: supplier data cleansing. Export records from all systems, identify duplicates, validate critical information (tax ID, contact, payment terms), assign categories, flag inactive suppliers. Create a maintenance process: who validates new suppliers, by what criteria, and how information is updated. Without clean data, automation generates errors that erode confidence in the system.

Fourth step: select software aligned with maturity. SMEs with unstructured processes should avoid complex platforms that assume category governance, ERP integration, and sophisticated workflows. Better to start with simple tools (request management, supplier catalogue, basic approvals) that adapt to existing processes, and scale as maturity increases. Companies with redesigned processes can justify more robust platforms, but should require proof of concept before commitment—test with one category for 2-3 months, measure impact, and only then scale up.

Fifth step: phased implementation. Do not automate all categories simultaneously. Start with a pilot category (significant volume, redesigned process, clean data), implement, measure results (cycle time, compliance, savings), adjust, and only then expand. The big-bang approach—go-live for all procurement in one day—has a failure rate above 60%, according to Hackett Group (2022). An incremental approach allows for learning and adjustment.

The decision to automate procurement is not binary. It is a sequence of decisions: which categories to redesign first, which processes to change, which data to clean, which tool to use, what implementation pace. Companies treating automation as an IT project—buying software, configuring, going live—fail. Companies treating it as an organisational transformation project—redesigning processes, changing behaviours, building capability—are more likely to deliver sustainable value.

For SMEs with limited resources, the critical trade-off is between breadth and depth. Attempting to automate all procurement with current processes delivers little value. Deeply redesigning 2-3 critical categories, automating those, and scaling progressively delivers more—but requires patience and discipline. The pressure to "digitise quickly" often leads to the first option; the evidence favours the second.

Decision-makers should also consider alternatives to full automation. For low-value, high-variability categories (ad hoc purchases, non-standard requirements), maintaining a manual process may be more efficient than forcing automation. For critical categories with few suppliers, investing in relationships and negotiation may deliver more value than investing in software. Automation is a tool, not an objective—it should be used where it adds value, not where it adds friction.

Limits and Unknowns

The evidence analysed has limits. The cited studies focus on industrial and distribution SMEs, with recurring purchases and relatively standardised categories. Service companies, with more variable purchases and lower volume per category, may have a different dynamic—automation may add complexity without sufficient return. Specific research in this context is limited.

Second limitation: most studies document results at 12-24 months. Long-term impact—whether automation improves negotiation capacity over contract renewal cycles, or whether accumulated data enables predictive supplier risk analysis—remains under-documented. There are anecdotal cases of emerging value after 3-5 years, but systematic evidence is scarce.

Third limitation: rapidly evolving technology context. Procurement platforms with generative AI, capable of generating specifications, comparing proposals, and suggesting alternative suppliers, are beginning to emerge. If these tools reduce dependence on structured processes—if they can operate effectively with less clean data or less defined categories—the argument of this article may weaken. But so far, evidence from real implementations is insufficient to validate this promise.

Finally, the article assumes process redesign is viable. In contexts of high staff turnover, organisational cultures resistant to change, or leadership unable to impose new rules, redesign may fail—and automating dysfunctional processes may be preferable to maintaining manual status quo. This is a risk management issue that each organisation must assess in its own context.

Next step: if this topic requires an executive decision, Macro Consulting can support with Digital Transformation, linking diagnosis, priorities, and execution.

Sources

• Gartner (2023), "Market Guide for Procurement Software", Stamford, CT
• The Hackett Group (2022), "Digital Procurement Transformation: Performance Study 2022", Atlanta, GA
• McKinsey & Company (2021), "Why procurement digital transformations fail — and how to get them right", McKinsey Quarterly, Q3 2021
• MIT Center for Transportation & Logistics (2020), "Automation in Procurement: When Does It Pay?", Cambridge, MA
• APICS (2023), "Supply Chain Management in European SMEs: Benchmark Report", Chicago, IL
• INE — Instituto Nacional de Estatística (2023), "Inquérito à Utilização de Tecnologias da Informação e da Comunicação nas Empresas 2023", Lisbon
• COTEC Portugal (2021), "Digitalização de Processos em PMEs Industriais: Estudo de Casos", Lisbon
• University of St. Gallen (2022), "Procurement Governance in Medium-Sized Enterprises", Institute for Supply Chain Management, St. Gallen
• IAPMEI (2022), "Estudo sobre Práticas de Procurement em PMEs Portuguesas", Lisbon
• Universidade Católica Portuguesa (2021), "Qualidade de Dados em Sistemas de Gestão de PMEs", CATÓLICA-LISBON, Lisbon
• IDC (2023), "European SMB Technology Survey: Procurement & Supply Chain", London
• Banco de Portugal (2023), "Boletim Económico — Outubro 2023", Lisbon
• AICEP (2022), "Internacionalização das PMEs Portuguesas: Relatório Anual 2022", Lisbon
• IAPMEI (2023), "Relatório Sobre a Situação das PME em Portugal 2023", Lisbon