Bad service data is expensive | May 2026

The problem nobody budgets for

You can design the best coverage model in the industry. Position stock in the right cities. Contract the right couriers. Staff the right technicians. And then the engineer shows up on site, and the batch of the part on the work order does not match the part in the machine.

The visit fails. A second truck roll gets scheduled. An emergency shipment follows. The SLA clock keeps running.

When these incidents get reviewed, the root cause usually gets coded as logistics or technician error. But trace it back far enough and it almost always starts with data. The asset record said one configuration. The machine had another. The entitlement was ambiguous. The service BOM was not up to date. The dispatched part was valid for the product family but wrong for that serial number.

Aberdeen Group research puts it plainly: insufficient or incorrect parts on site account for 51 percent of failed first-time fixes [1]. That is not a technician problem. That is a data-to-dispatch problem.

And the customer impact is real. A recent McKinsey survey of 250 senior executives found that 70% saw no improvement from their service providers in the past decade. Their top concerns? Poor spare parts availability and slow turnaround times for repairs. 80% percent are planning to change their sourcing strategy in the next two years [2]. When the data behind dispatch is broken, it shows up in exactly the metrics that make customers leave.

The data elements that drive uptime decisions

Service execution depends on a chain of decisions made in seconds: which asset failed, what the customer is entitled to, which part is needed, where to source it. Each decision draws on data. When the data is right, the chain works. When it is not, you get wrong parts, delayed dispatches, and unnecessary escalations.

Six data elements sit on the critical path.

Installed base. The foundation. Knowing what is installed, where, and in what condition. Dekker et al. show that installed-base size, together with part-life data, explains spare parts demand better than historical sales alone [3]. If the install-base record is wrong, the demand signal is wrong. Everything downstream inherits the error.

Serial-to-configuration linkage. The machine shipped five years ago is not the machine in the field today. Field upgrades, component swaps, and engineering changes mean the current configuration can differ significantly from as-shipped state. When the system cannot link a serial number to its actual configuration, the wrong part gets selected [3].

Service BOM. The managed view of what breaks and what replaces it. Siemens defines it as providing a neutral representation of a product with defined service-related definitions, tightly coupled to the physical BOM for accurate visibility of the asset configuration throughout its life [4]. An incomplete service BOM means technicians work from a parts list that does not reflect reality.

Entitlement status. What a specific customer, site, or serial number is actually allowed to receive. Parts? Labor? Loaners? PM? When entitlement data is scattered across disconnected contract and service systems, operations delays action while someone checks. We covered this in the March article: if entitlement is vague, execution becomes slow, expensive, and inconsistent.

Supersession and interchangeability. Parts get discontinued and replaced. If the supersession chain is not maintained in the ordering system, planners stock obsolete SKUs and technicians arrive with parts that no longer fit. McKinsey describes this as the “two-speed challenge”: products with 30-year lifespans depend on components with 5-year lifecycles, and each discontinuation cascades into inventory, service, and commercial problems [5].

Installed location. If the system does not know where the asset physically sits, dispatch planning breaks. Dekker et al. describe how customers relocate systems without informing the provider, causing misplaced inventory and increased servicing costs [3].

Where the data breaks: three scenarios

Data center. A server chassis fails. Dispatch selects a board from the nearest stocking location. The technician arrives and discovers the chassis was upgraded eighteen months ago with a different backplane. The board does not fit. Second dispatch, expedited shipping, SLA breached. Root cause: the install-base record was never updated after the field upgrade.

Medtech. An imaging system goes down. The service desk cannot confirm whether the contract covers a specific high-value module. Three hours pass while commercial operations reviews contract language. Clinical schedule disrupted. Root cause: entitlement was defined at the contract level but not mapped to asset components.

Telecom. A parts order comes in for a legacy router. The part number was disconnected two years ago. The replacement sits in stock five hundred kilometers away, but the supersession chain was not maintained in the ordering system. Time to restore doubles. Root cause: engineering knew about the supersession, but the service parts master did not.

Why “we have an ERP” is not the answer

The common response to data quality concerns is that the data exists somewhere in the ERP, or the service platform. And technically, that is often true. Records exist. But existing is not the same as accessible, accurate, and usable at the moment dispatch needs to act.

Didriksen et al. studied this empirically in an offshore operation managing over 10,000 spare parts. They found 50 relevant data fields scattered across 11 data tables, split between the plant-maintenance and materials-management modules. Maintenance navigated one module. Procurement navigated the other. Each department built a different decision basis from fractional data. Two earlier approaches using document-based methods failed because integrating the fragmented data was too expensive and error-prone. Only when an integrated data model was introduced did the project succeed: 15.1 percent stock value reduction, 76–91 percent improvement in resource efficiency, 4–5 percent decision quality improvement [6].

Service data is structurally harder than manufacturing data. It changes in the field. It depends on technician input. It crosses organizational boundaries. And it degrades over time. McKinsey makes the same point from the commercial side: the aftermarket involves many varied field-replaceable units with intricate relationships, resulting in more heterogeneity and unpredictability than new-product supply chains [5]. That heterogeneity is a data problem before it becomes a logistics problem.

Meanwhile, customers are getting impatient. 43 percent expect their providers to use data and AI to improve service speed. But less than 30 percent are actually using their suppliers’ digital offerings, citing unclear value and failure to address real operational pain points [2]. The gap between what data could do and what data actually does is where most service organizations are stuck.

The cost spiral: inventory, expedites, and repeat visits

Bad data does not just cause failed visits. It drives a cost spiral.

Excess inventory. When planners do not trust install-base data, they buffer. More safety stock, broader assortments, more forward-positioned SKUs. Van der Auweraer et al. show that tracking installed-base size provides the most value for spare parts inventory control, especially when the base is not constant [7]. But that value only materializes if the data is accurate. Without it, planners overstock declining products and understock growing ones.

Expedite spend. Every wrong-part dispatch creates an emergency. The cost difference between standard and express service parts logistics can be five to ten times, and it usually goes unmeasured because it is buried in transport lines rather than tracked as a data-quality cost.

Repeat visits. Industry-average first-time fix rates sit around 75–80 percent. Top performers reach 88 percent [1]. That gap is mostly a readiness gap: right part, right data, right preparation. And 80 percent of customers say they would pay higher prices for services that reduce total cost of ownership over time [2]. They are telling providers exactly what they want. The problem is that broken data prevents providers from delivering it.

The minimum viable service data model

This is not a master data management program. It is a checklist: the data objects and ownership rules that need to be in place before dispatch, parts selection, and entitlement validation can work.

If any row cannot be answered clearly for your top-criticality assets, the data model is not ready to support reliable execution.

Where a 4PL becomes the early-warning system

A lead logistics partner running the service parts network encounters data failures every day. They process the dispatch, pick the part, ship it, and hear back when it does not match. That position creates a feedback loop most organizations do not use.

Didriksen et al. found that logistics holds a key role in bridging siloed ERP modules because it sits at the interface between procurement and maintenance [6]. That is exactly the role a 4PL plays. Not as a data-cleansing vendor, but as an operational layer that surfaces failures in real time and feeds corrections back to source systems.

In practice, that means flagging recurring wrong-part dispatches and tracing them to BOM mismatches. Identifying supersession gaps by tracking orders for obsolete part numbers. Quantifying entitlement delays. Benchmarking site-level data quality by comparing exception rates. And providing the evidence that turns data improvement from an IT project into a business case with measurable returns. Some 4PL can even incorporate the item interchangeability rules, convertor style in their execution systems.

What to fix first

Not everything can be fixed at once. Here is the sequence that produces the fastest return.

First: entitlement clarity. The single biggest source of dispatch delay and billing disputes. Make it machine-readable. Map it to the asset or serial level. This alone removes a category of friction from every service interaction.

Second: service BOM and supersession. The biggest drivers of wrong-part dispatches. Clean these before investing in demand planning sophistication. Goldsmith and Sachs show that forecast accuracy degrades sharply when the data foundation is limited [8]. The BOM has to be right before the forecast can be right.

Third: installed-base accuracy. The hardest to fix because it depends on field discipline. Every technician, every install, every swap needs to close the loop. But it is essential for proactive and predictive models. Dekker et al. conclude that spare parts demand forecasts can be made considerably more timely and accurate using installed-base information [3]. The value is real, but only when the data is trustworthy.

Once the data foundation is reliable, you can design stocking and network positioning with confidence. Without it, even the best multi-echelon optimization is working from bad inputs.

Final thoughts

Service leaders do not need a two-year master data program. They need a minimum viable data foundation that makes dispatch, parts selection, and entitlement work today.

That means treating data as an operational asset, not an IT hygiene project. Assigning ownership for the six elements on the critical path. Using the logistics layer as a real-time feedback loop that catches data failures before they become customer failures.

The companies that treat data quality as a cost line will keep paying for it in expedites, repeat visits, and excess inventory. The companies that treat it as a service capability will deliver faster, spend less, and build the foundation for everything that comes next.

Author: Eyal Yossef, VP Supply Chain Solutions at Unilog

References

[1] Aberdeen Group, cited in AEX Software (2026). First-Time Fix Rate Improvement. aexsoftware.com/blog/how-to-increase-first-time-fix-rates

[2] McKinsey & Company (2025). How aftermarket service providers can meet new customer expectations. Survey of 250 senior executives, data collected May 2024.

[3] Dekker, R. et al. (2013). On the use of installed base information for spare parts logistics. Int. J. Production Economics, 143(2), 536–545. doi:10.1016/j.ijpe.2011.11.025

[4] Siemens Digital Industries Software. Service bill of materials (sBOM). siemens.com

[5] Vesco, S. / McKinsey & Company (2024). Why aftermarket and service are vital to OEMs—and how to excel.

[6] Didriksen, S.K. et al. (2026). An empirical data model for spare parts management. Applied Sciences, 16(1), 94. doi:10.3390/app16010094

[7] Van der Auweraer, S. et al. (2021). The value of installed base information for spare part inventory control. Int. J. Production Economics, 239, 108186. doi:10.1016/j.ijpe.2021.108186

[8] Goldsmith, R.L. & Sachs, A-L. (2025). Spare part demand forecasting in every phase. Int. J. Production Research. doi:10.1080/00207543.2025.2607643