The next service provider is AI

Silent service

Tuesday, 4:05 pm. The truck is at loading dock 2, fully loaded. All that is missing are the shipping documents. A click in the software produces an error message, but no documents, and without them the truck cannot leave the yard. At 4:10 pm the ticket is logged. At 4:40 pm the truck is on the motorway.

The service provider solved the problem in half an hour. People in the shipping department will still be talking about that Tuesday weeks later. Nobody talks about all the days before, when the documents simply came out of the printer. Good service is silent. Nobody sends flowers for it.

This has little to do with ingratitude. Bad experiences simply stick more than good ones, and for services this has even been measured: a failure lowers satisfaction more than a good day raises it.^A1 So almost everyone in a company has an opinion about service providers, and most of the time it is not a good one.

With exactly this experience in mind, companies are now buying the next one.

The first article in this series was about a decision for which there is no position on the org chart. A rush order, five departments, each checking by its own rule, and an agent that makes a better proposal in one minute than the whole company does in two days. The agent was not the problem. What was missing was someone to take responsibility for its proposal. At the end there was a sentence that carries more than there was room for: the Mittelstand has run this experiment before, when it handed the operation of its IT to service providers. Those who outsource operations without building the management capability end up with less control and not one cent less responsibility.

This text takes that sentence at its word. At the latest since Kodak handed large parts of its IT to service providers in 1989, companies have outsourced on a large scale, and research now knows fairly well when it works.^A2 The question is what of this applies to a service provider made of software, with the authority to act. The other side of the table joins the discussion: Lukas Wiegand runs IT operations for mid-market companies with BUCS IT GmbH.

Power from the socket

Managed services means a service provider takes over the ongoing operation of a clearly defined part of the IT and is permanently responsible for it. Servers, network, workplaces, backup, security. Payment is usually per workplace or per month, with agreed response times and availability.

The model has been growing for years. In a German market study of companies with revenue of 100 million euros or more, two thirds of IT managers say it will be highly important in the coming years. In a global survey, two out of three companies use it, compared with just under half two years earlier.^A3

There is a simple reason why it grows so smoothly. Basic IT has become a standard product, like electricity. The writer Nicholas Carr predicted this as early as 2003 and caused a lot of irritation with it: when everyone has the same technology, it no longer gives anyone an advantage. It becomes a cost that everyone has to pay.^A4 His advice to managers was correspondingly sober: watch the vulnerabilities more than the opportunities. That is how every utility works. Nobody praises the power company for a week of electricity. You call when it is gone.

From the service provider's point of view, the division of labor looks like this:

“We operate and are responsible for the basic infrastructure, and we provide the foundation for secure and stable IT operations. What each company builds on this foundation, and how it is used, is the customer's responsibility to steer.”

What the customer gets for this is tangible: more stable technology, better documentation, reliable response times for critical systems, monitoring, backup and security. The money is usually well spent.

And yet BUCS IT sees the same thing again and again, especially where customers do not follow the recommendations on steering. The technology gets better, the day-to-day business does not necessarily. Responsibility becomes more formal, but not more effective. The costs appear neatly on the invoice, but management finds it harder to say what it gets for them.

How can operations run better and the business still run worse?

Let the contract fix it

The first answer is almost always a better contract. Service description, response times, escalation levels, annexes to annexes. Both sides try to write down in advance everything that could go wrong. The contracts get thick and still have gaps, because no contract can foresee every case. This insight earned the 2016 Nobel Prize in Economics. The researchers' conclusion: because no contract is complete, what matters is who has the right to decide when the unforeseen case occurs.^A5

In daily practice, the gap usually shows right at the start. When a service provider takes over a piece of IT that was never described, it first has to be described. Taking stock of systems, clarifying dependencies, talking to the departments, drawing boundaries. In the experience of a managed service provider, four to five out of ten new customers arrive without usable documentation, and catching up takes two to ten person days.^A21 Then comes the question that later sparks the conflicts: is this a project that is paid for once, or does the customer silently expect it as part of operations?

The customer wants to get rid of the topic, but does not know exactly what it takes. The service provider wants the contract and does not know what it does not know. Both only find out in operation. Sooner or later someone thinks the sentence everyone knows who has ever been responsible for a service contract: if I had known that, I would have written it in.

It is surprising how rarely the contract is needed in daily business. As long as things run, it stays in the drawer. Only when something goes wrong does it come out, and then both sides put up walls. It is meant as a tool that describes services, boundaries and responsibilities. It becomes a shield when the specific case has no owner in the company.

Being thick is not a fault. Detailed contracts work better than loose ones, as research showed back in the nineties.^A2 Contract and trust work best together. An analysis of almost 150 studies comes to exactly this result, and it is particularly clear for services.^A6 So the contract alone is not enough. What complements it is a relationship, and that needs someone on each side to lead it.

What a company should clarify before outsourcing an operation is shown in Fig. 1.

Four of these questions are about who decides. They are often skipped, and there is almost always a lack of awareness that someone with a mandate is needed at all.

The telephone game

What goes wrong in operations usually starts with the ticket.

An employee reports: “It doesn't work.” What is meant is an invoice that does not go out, an order that is stuck, documents for a truck that is waiting. The ticket says “printing doesn't work” or “system slow”. What depends on it and how urgent it is, is not in there. So the service provider asks back. The ticket is solved technically, the employee gets in touch again anyway, and a follow-up ticket is opened. Often the cause was not in the basic IT at all, but in permissions, an interface or another vendor. In this game of telephone, the information is usually lost at the same point: between the business department and the first technical translation.

It gets really expensive when two service providers are involved. Fig. 2 shows the typical sequence.

Anyone who has ever sent a rush order through the company knows the pattern: each department checks by its own rule and is right on its own terms. Here, two service providers stick to their contracts and both do everything right. The business department bears the damage, although no single service provider did anything wrong. What was missing was someone responsible for the interplay.

On top of that, everyone is measured by something different. The service provider by response time and closed tickets, IT management by availability and budget. Management is measured by the result, which is the sum. But it rarely sees what a single decision contributes to that sum.

For some decisions this is easy. If material costs rise, it shows up in the cost of materials, and you can calculate through to EBITDA what it costs. For an IT service the path is rarely that clear. If a weaker service means that an order sits longer or a piece of information gets lost along the way, it lands somewhere in the result. Where exactly and how much can only be traced back with considerable effort, and daily business leaves no time for that. So it stays invisible.

It is rarely measured either. From practice there are estimates for customers without a steering role. Of the tickets outside the basic IT provider's scope, 30 to 40 percent are affected: either a ticket that was already closed is reopened because the first solution did not solve the problem, or the description differs considerably from the solution that works in the end. Each ticket needs three to four times as many contacts as with customers that have a steering role, and the solution takes 30 to 60 percent longer.^A21 These are values from experience, not measurements. The direction is clear all the same. Research calls this the hidden costs of outsourcing: effort that is in no business case and quietly eats up the planned savings.^A7

The problem grows with every application, every interface and every additional service provider. At manufacturing customers there are between 2 and 30 applications, 10 to 100 interfaces and often four service providers at the same time.^A21 And manufacturing is less forgiving. A problem in the office costs productivity. A problem on the line or at the loading dock can immediately cost delivery capability, revenue or contractual penalties.

Who fills the gap when nobody steers? Fig. 3 gives the sobering answer from practice.

In this experience, only one or two escalations a year reach management.^A21 That sounds like little. Above all, though, it means that almost everything is decided beforehand, and much more often by the service provider than necessary. Nobody delegated these decisions to it. There is simply nobody else there.

What works

What makes the difference is well researched, and it is rarely the service provider.

One of the most cited studies on IT outsourcing followed 61 decisions on outsourcing or doing it in-house in 40 companies in the nineties and asked whether the expected savings materialized. The best-known figure from it: those who outsource almost everything mostly fail, those who outsource selectively mostly do not. Less well known is a second analysis. It shows who made the decision, see Fig. 4.^A2

Where business and IT decide together, it works in almost nine out of ten cases. Where management outsources alone, it goes wrong more often than right. The size of the IT department made no difference in the same study. So even with a small IT function, outsourcing can work well.

The study is old, the finding is not. Research over the last ten years keeps coming to the same result: the capabilities of the client matter at least as much as those of the provider. What has shifted is which capabilities these are. In the past, the company's own IT had to understand the technology. Today departments buy software themselves, and one task above all remains in-house: translating between what the business needs and what the service providers deliver.^A8

This is exactly the task that is rarely filled. Only just over one in ten of the larger companies surveyed has a dedicated position that steers service providers. In the global survey, almost seven out of ten companies say their steering is not mature, and just as many have brought services back in-house in recent years, mostly to regain control over quality.^A9

First and foremost, it takes a mandate, meaning someone who is allowed to decide. And it takes better translation between what is supposed to have an effect in the company and what IT delivers. More tickets do not help with that. A service provider quickly notices how to recognize such a person, as Fig. 5 shows.

This person does not have to be a head of IT. One BUCS IT customer does not have one. The company is too small for its own head of IT, but too complex for simple IT: production is closely interlinked with IT, and legal quality requirements demand complete, verifiable documentation. The company has an IT coordinator. This person connects the organization with the service providers and keeps the collaboration running. One person with a mandate, without a department of their own.

Research supports this lean form. For smaller companies, one study shows that trust between management and service provider has slightly more influence on success than formal control. Another survey found that control only helps up to a point and does harm beyond it.^A10 What is needed, then, is someone whom both sides trust and who is allowed to decide. Whether this is also the head of IT depends on how much the business demands of IT.

Why is this person so rare? You do not see them as long as nothing escalates. The role is seen as pure administration, costs money before it delivers anything, and is therefore expected to run on the side. After outsourcing, internal IT is often sized too small, and many confuse the contract with steering. The gap only becomes visible at the first major incident.

There is no fixed threshold beyond which it tips over without this person. It becomes critical when several things come together: more than one critical service provider, several sites, many business applications, and nobody who is allowed to set priorities.

The picture is only fair when the other side is included:

“The service provider has to operate transparently, cleanly and professionally. But the customer has to state what effect, priority and boundary this operation should have in the company.”

The new service provider

Now a service provider is coming in that is not a company.

An AI agent gets a task, access to data and systems, and rules it has to keep to. It works independently, decides within its scope and reports results. In the language of outsourcing, that is a service provider with a service description. For both, the same applies: if nobody in the company is responsible for the effect, the system works by its own logic. A large consultancy that accompanied agent projects for a year concludes that an agent is onboarded more like a new employee than installed like software.^A11

On its first day, the agent knows nothing about the company. It has to be briefed, with goals, context and examples, and it needs boundaries it can recognize. After that it is steered and reviewed, and the collaboration can improve over time.

With one difference. A human service provider gets to know the operation because people remember what went well and what did not. An agent does not do this on its own. Whether it remembers depends on whether someone has set up a memory. Without it, it starts from scratch every time, and it only gets better if a person maintains its briefing and boundaries.^A12 This improvement, too, needs someone to organize it.

How close the parallel is becomes clear in the introduction. What applies when handing over a piece of IT to a service provider can be transferred almost word for word to an agent. Fig. 6 puts the two side by side.

The comparison also holds for maturity. Many believe they first have to get their data in order before the first agent. Outsourcing has an answer to this that fits one to one:

“Maturity does not mean perfect data. Maturity means: a service that can be described, deviations that can be recognized, and named responsibility.”

Good data often only comes with good steering.

In one respect, the new service provider can even do more than the old one. An agent with access to the data of the entire chain can see what daily business never calculates: how a postponed order or a lost piece of information works its way through to the result. It can tell apart what is merely a precondition for acting and what actually changes the result. This requires that someone shows it this chain, releases the data and defines which effects count. So the transparency that is missing today only emerges where someone steers the agent.

Two current figures show how far companies have come. According to a large international survey, only one in five has a mature governance model for agents. And market researchers expect more than 40 percent of agent projects to be cancelled by the end of 2027.^A13 That sounds familiar: two thirds consider managed services important, just over one in ten companies has a dedicated position that steers them. The same mistake, one generation later.

Where the comparison falls short

A comparison is only as good as its limits. Lukas Wiegand puts the core into one sentence: a service provider can be questioned and held to a contract, an agent decides faster, less intuitively traceably and potentially at a much greater scale. Besides the missing memory, there are six points where an agent differs from a service provider. Fig. 7 summarizes them, and each one makes steering more important.

Liability. A service provider can be held to a contract and, in the worst case, subjected to recourse. An agent cannot. The European AI Act therefore leaves the obligations with the company that deploys it. For high-risk applications, it must assign a person who is competent and authorized to oversee the agent. A planned directive that was supposed to make it easier for injured parties to claim damages was withdrawn by the EU in 2025.^A14 Responsibility stays in-house, and with it the steering role.

Speed and scale. A service provider makes errors in individual cases. An automated system repeats the same error until someone stops it. The best-known example is not even from AI. In 2012, a faultily installed trading algorithm at a US broker sent out more than four million orders in 45 minutes, and the loss was over 460 million dollars.^A15 Agents are improving quickly. But there is still a long way between “works most of the time” and “works reliably”. In a 2025 test with realistic office tasks, the best model tested at the time completed just under one in three tasks entirely on its own.^A16

Traceability. You ask a service provider why it did something and get an answer you can check. A model also provides reasoning. But that need not be the reasoning by which it actually decided. The vendors' own research shows this.^A17

Gray zone. This may be the most important difference. A person who notices that an instruction does not fit the case stops and asks. A service provider, too, ends up deciding itself when nobody answers its question, but at least it asks. An agent rarely does. Studies from 2023 to 2026 show that language models now often recognize ambiguity, but ask back in only a few percent of cases. And even the newest models would rather answer with false confidence than admit a gap.^A18 With an agent, someone therefore has to draw the boundary so that it does not get into the gray zone in the first place.

Relationship. When things are difficult with a service provider, a lot can be resolved in conversation. You call, understand the other side's situation, find a way that is in no contract. An agent is more stoic. It can express understanding, even very well. But its result does not get better because of it, and the frustration is at least as great as with a service provider. People demonstrably feel less understood as soon as they know that the understanding reply came from a machine. How quickly they withdraw trust from an algorithm after an error, on the other hand, is disputed.^A19 Whether people distrust an agent too quickly or too slowly, the consequence for steering is the same. Those who punish errors too harshly switch the agent off too early. Those who accept them too easily check too little. With a service provider, trust carries part of the steering. With an agent this trust does not exist, so review has to take over that part.

Contract. There is a contract with the software vendor, but not with the agent itself. It only gets instructions and rules, and they are just as incomplete as any outsourcing contract. With people, law, culture and relationship fill the gaps. The agent lacks this framework.^A20 With a service provider, contract and relationship carry the load together. With an agent, the relationship drops away. What remains are the instructions, which are never complete, and the person who sharpens them every day.

Silent again

At the beginning there was a truck at loading dock 2, and all the days before it, when nobody talked about the documents.

Good steering is the same. When it works, you do not see it. No ticket is opened for the third time, no department waits for two service providers to agree. The IT coordinator from the example will never appear in an annual report. Things simply run.

It will be no different with the agent. Where it works, hardly anyone will talk about it. Somewhere there is a person who knows which decisions the agent makes, who knows its boundaries, evaluates its errors and makes its briefing a little better every week. Nobody sends flowers for that.

Whether this person already exists in your own company can be settled with a single question: who in our company is, for the first agent, what the IT coordinator is for the service providers? If no name comes to mind, that is the answer.

Appendix: the evidence in detail

ExpandCollapse

For those who want to know more: the studies behind the statements in the text, with sample and original wording.

1. A1 Bad weighs more. Baumeister and colleagues summarize psychological research as follows: “Bad is stronger than good, as a general principle across a broad range of psychological phenomena.” Mittal, Ross and Baldasare measured this for services, in three customer surveys with more than 27,000 responses in total from health care and car retail: weak performance on one attribute affects satisfaction and repurchase more strongly than equally good performance. Roy F. Baumeister, Ellen Bratslavsky, Catrin Finkenauer, Kathleen D. Vohs: Bad Is Stronger Than Good. Review of General Psychology 5(4), 2001. Vikas Mittal, William T. Ross, Patrick M. Baldasare: The Asymmetric Impact of Negative and Positive Attribute-Level Performance on Overall Satisfaction and Repurchase Intentions. Journal of Marketing 62(1), 1998. ↩
2. A2 Who decides determines success. 61 sourcing decisions in 40 companies in the USA and the UK, 1991 to 1995, 145 participants. Measure of success: expected cost savings achieved, evaluated for 48 decisions with a discernible outcome. Total outsourcing (more than 80 percent of the IT budget external) 29 percent successful, selective 85 percent. Joint decision by management and IT 87 percent, IT management alone 74 percent, management alone 50 percent, management alone in decisions to outsource 40 percent (tables 3 to 5). Detailed contracts more successful than loose ones. The size of the IT function does not distinguish success. Starting point of the wave was Kodak's outsourcing in 1989. Mary C. Lacity, Leslie P. Willcocks: An Empirical Investigation of Information Technology Sourcing Practices: Lessons from Experience. MIS Quarterly 22(3), 1998, pp. 363 to 408. ↩
3. A3 Managed services are growing. Two thirds of IT managers say managed services will be highly important in future (121 user companies with revenue of 100 million euros or more, a quarter of them above 1 billion euros). Globally, 67 percent use managed or operate services, two years earlier 45 percent (more than 500 executives). Europe 2025: 38.9 billion US dollars annual contract value for outsourced IT services, 57 percent cloud. Lünendonk: IT-Sourcing-Trends 2025/2026, p. 22. Deloitte: Global Outsourcing Survey 2024. ISG Index Europe, Q4 2025. ↩
4. A4 IT becomes a utility. “What makes a resource truly strategic … is not ubiquity but scarcity.” IT is becoming a cost “that must be paid by all but provide distinction to none”. Carr's advice to managers: “Spend less”, “Follow, don't lead”, “Focus on vulnerabilities, not opportunities”. Counter-position: McFarlan and Nolan, according to whom companies do differ in how they use IT. Nicholas G. Carr: IT Doesn't Matter. Harvard Business Review, May 2003. F. Warren McFarlan, Richard L. Nolan: Why IT Does Matter. HBS Working Knowledge, 2003. ↩
5. A5 No contract is complete. From the announcement of the 2016 Nobel Prize in Economics to Oliver Hart and Bengt Holmström: “Because it is impossible for a contract to specify every eventuality, this branch of the theory spells out optimal allocations of control rights.” Royal Swedish Academy of Sciences, press release 2016. ↩
6. A6 Contract and trust complement each other. Poppo and Zenger find for information services that formal contracts and relational governance complement each other and together improve performance. The meta-analysis by Cao and Lumineau of 149 studies covering 33,051 business relationships confirms this, with a stronger effect for services. Laura Poppo, Todd Zenger: Strategic Management Journal 23(8), 2002, pp. 707 to 725. Zhi Cao, Fabrice Lumineau: Journal of Operations Management 33 to 34, 2015. ↩
7. A7 Hidden costs. Survey of 50 companies on their outsourcing projects: the planned savings often get eaten away “by costs that managers can't pinpoint”. Jérôme Barthélemy: The Hidden Costs of IT Outsourcing. MIT Sloan Management Review 42(3), 2001. ↩
8. A8 What has to stay in-house, then and now. Feeny and Willcocks name nine capabilities that must remain internal despite outsourcing, including leadership, linking business and IT, informed buying, contract facilitation and vendor development. A review of 174 studies (2010 to 2014) confirms the role of client capabilities. A study of 200 companies in 13 countries found in 2025 six ways to maintain outsourcing performance in a crisis, most of them through client capabilities. For cloud services, the customer's technical IT knowledge has mixed effects, because departments buy for themselves. David F. Feeny, Leslie P. Willcocks: Sloan Management Review 39(3), 1998. Mary C. Lacity, Shaji A. Khan, Aihua Yan: Journal of Information Technology 31(3), 2016. Ilan Oshri, Federica Angeli, Julia Kotlarsky, Jatinder S. Sidhu: Journal of Strategic Information Systems, 2025. Stephan Schneider, Ali Sunyaev: Journal of Information Technology, 2016. ↩
9. A9 Steering is rarely filled. Only 11 percent have a dedicated provider management anchored in the organization, 57 percent have none (Lünendonk, sample as in A3). Almost 70 percent say their vendor management function is not fully mature. 70 percent have brought services back in-house within five years, the most frequent motive at 68 percent being better control over quality and performance. Deloitte itself offers consulting on this topic. Lünendonk: IT-Sourcing-Trends 2025/2026, fig. 26. Deloitte: Global Outsourcing Survey 2024. ↩
10. A10 Trust in the Mittelstand. Eight failed outsourcing projects in SMEs: “trust is slightly more important than control issues such as output-based contracts and structured controls in the governance of IT in SMEs.” A survey of 344 IT executives, not limited to SMEs, shows that governance only helps up to a point and does harm beyond it. Jan Devos, Hendrik Van Landeghem, Dirk Deschoolmeester: Rethinking IT governance for SMEs. Industrial Management & Data Systems 112(1/2), 2012. Study by LMU Munich, Journal of Strategic Information Systems, 2021. ↩
11. A11 Onboarding agents. “Onboarding agents is more like hiring a new employee versus deploying software.” And: “there can be no ‘launch and leave’.” McKinsey QuantumBlack: One year of agentic AI: Six lessons from the people doing the work, 12 September 2025. ↩
12. A12 Memory is an add-on. With the major vendors, memory for agents is a feature that is actively switched on or off. An MIT report names the lack of learning ability as the biggest hurdle: most systems “do not retain feedback, adapt to context, or improve over time”. The report's methodology is disputed. Anthropic: Memory Tool documentation. OpenAI: Memory FAQ. MIT NANDA: The GenAI Divide. State of AI in Business 2025. ↩
13. A13 Governance for agents. 21 percent have a mature governance model for agentic AI (3,235 executives in 24 countries). Forecast: more than 40 percent of agentic AI projects will be cancelled by the end of 2027, due to costs, unclear value and inadequate risk control. Both firms earn money from consulting on this topic. Deloitte: State of AI in the Enterprise 2026. Gartner, press release of 25 June 2025. ↩
14. A14 Liability. Article 26(2) of the AI Act: “Deployers shall assign human oversight to natural persons who have the necessary competence, training and authority, as well as the necessary support.” Applies to high-risk systems. The AI Liability Directive was withdrawn, Commission decision of 16 July 2025, published in the Official Journal C/2025/5423 of 6 October 2025. Regulation (EU) 2024/1689. Official Journal of the European Union. ↩
15. A15 Errors in series. Knight Capital, August 2012: more than four million orders in 45 minutes, loss over 460 million US dollars. Trading software, not AI. U.S. Securities and Exchange Commission, press release 2013-222. ↩
16. A16 Reliability of agents. In a simulated company, the best model tested in 2025 completed 30 percent of realistic office tasks entirely on its own. According to METR, many real tasks require reliability of over 98 percent. TheAgentCompany, arXiv 2412.14161, NeurIPS 2025. METR: Notes on time horizon limitations, 22 January 2026. ↩
17. A17 Reasoning and actual path. Two models mentioned a hint they had demonstrably used in only 25 and 39 percent of cases respectively in their reasoning. Anthropic, Chen et al.: Reasoning models don't always say what they think, 3 April 2025. ↩
18. A18 Gray zone. In 2023, GPT-4 resolved ambiguous statements correctly in only 32 percent of cases, humans in 90 percent. 2026: models recognize ambiguity in 60 to 80 percent of cases, but ask back in only 0.5 to 5 percent (preprint, tested up to GPT-4.1 and Claude 3.5). Current models such as GPT-5, Gemini 3 Pro and Claude Opus 4.5 also answer overconfidently in cases of doubt, because their evaluation rewards guessing. Alisa Liu et al.: EMNLP 2023. Su, Cardie: arXiv 2605.25284, 2026. Adam Tauman Kalai et al.: Nature, 22 April 2026, authors partly from OpenAI. ↩
19. A19 Trust after errors. Dietvorst, Simmons and Massey: “people more quickly lose confidence in algorithmic than human forecasters after seeing them make the same mistake.” Counter-findings: Logg, Minson and Moore found that people sometimes prefer to follow algorithmic advice, Qin, Zhou and Zhong that human errors cause more annoyance. Yin, Jia and Wakslak: people feel less heard as soon as they know the reply came from AI. Journal of Experimental Psychology: General 144(1), 2015. Organizational Behavior and Human Decision Processes 151, 2019. Computers in Human Behavior: Artificial Humans, 2025. PNAS, 2024. ↩
20. A20 Incomplete instructions. Human contracts are always incomplete, and law and culture fill the gaps. An AI lacks this external framework. Dylan Hadfield-Menell, Gillian K. Hadfield: Incomplete Contracting and AI Alignment. arXiv 1804.04268, 2018. ↩
21. A21 Practical values from a managed service provider. Estimates by Lukas Wiegand, BUCS IT, from his own customer base, not measured. 40 to 50 percent of new customers without usable documentation, catching up takes 2 to 10 person days. For customers without a steering level, 30 to 40 percent of tickets outside the basic IT provider's scope reopened (a ticket that was already closed is opened again because the first solution did not solve the problem) or wrongly described (major difference between problem description and the solution that actually works). Three to four times as many contacts per ticket, 30 to 60 percent longer resolution time. At manufacturing customers 2 to 30 applications, 10 to 100 interfaces, often four service providers. One or two escalations per year to the customer's management. ↩

About the author

Operations consultant · Mid-market companies · formerly COO

Christopher von Könemann advises mid-market companies on improving results and working capital and on setting up steering systems. He was previously COO of a mid-market group of companies.

LinkedIn · Contact

Share on LinkedIn

With contributions from Lukas Wiegand

Lukas Wiegand brings in the perspective of a managed service provider. With BUCS IT he is responsible for operating complex IT environments in mid-market companies, where technical performance, clear responsibilities and business priorities have to come together. bucs-it.de