A day every SME knows
Studies show where knowledge workers really lose their time and which Nexoria modules give it back. With a concrete example calculation for a 30-employee company: €503,000 in time value per year with €10,440 in licence costs.
07:30. A managing director switches on his computer. 247 unread emails. An office administrator searches SharePoint for a contract that a colleague filed two years ago. The phone rings at reception and nobody answers. In the website's live chat, a customer has been waiting for eleven minutes.
A day that every SME in Germany knows. Studies show how much working time on a day like this goes into searching, email and coordination.
Where the eight hours really go
Probably the most frequently cited figure in the industry comes from the report The social economy: Unlocking value and productivity through social technologies by the McKinsey Global Institute of July 2012: Around 19 per cent of working time goes into searching for and gathering information. The value was measured for interaction workers, not for knowledge workers in general. In a 40-hour job, that is almost eight hours a week, a whole day's output in which nobody produces anything and people only search.
On top of this, there are other time-consuming areas that have been studied several times in recent years:
- Email handling: Interaction workers spend around 28 per cent of their working time reading and answering emails (McKinsey Global Institute, “The social economy: Unlocking value and productivity through social technologies”, July 2012).
- Tool switching: Knowledge workers used an average of ten different applications per working day and switched between them 25 times a day (Asana, “Anatomy of Work Index 2021”, survey of more than 13,000 knowledge workers in seven countries, including Germany; in the following edition, “Anatomy of Work Global Index 2022”, the figure fell to nine applications). Every switch costs context, and the context then has to be rebuilt.
- Work about work: The same survey puts the share of working time that does not go into actual value creation but into status updates, meetings, duplicated work and coordination at 60 per cent. The figure is based on respondents' self-reporting; in the 2022 and 2023 editions, the value was 58 per cent. The publisher, Asana, is a provider of work management software.
These figures are not abstract. They translate very concretely into the everyday work of an SME when you work them through for each persona.
Administrative staff: where the mailbox hours disappear
A typical office administrator in an SME starts her day with the mailbox. Microsoft measures around 117 incoming emails per day for the average employee (Microsoft WorkLab, Work Trend Index Special Report “Breaking down the infinite workday”, 17 June 2025, based on anonymised Microsoft 365 usage data as of 15 February 2025, EU tenants are expressly not included). For German businesses, this is an indication, not a measured value. What is clear, however, is where the time goes: Which email is urgent? Which needs a reply? Which is phishing? Which can be archived?
At the same time, search queries come up. “Where is the current framework agreement with supplier X?” “Which travel expense allowance applies to business trips over 50 kilometres?” “What did we last document on customer complaint Y?” Each of these questions takes between five and fifteen minutes on average. Spread over the day, this adds up to a noticeable block. This example calculation does not derive the McKinsey share; it merely illustrates how such a share comes about in everyday work in the first place.
With an integrated AI platform, the calculation shifts significantly. A semantic knowledge base finds the source and points to the specific document and page number. A mail chat summarises incoming emails, suggests a reply and transparently flags phishing attempts. The office administrator makes the decisions; the computer does the groundwork.
Assumption for the example calculation: instead of 90 to 120 minutes of email triage, 30 to 45 minutes are enough. The rest comes down to pressing “Send” after a quick check. For document searches, the effort per query drops to a few seconds instead of several minutes. Per administrative employee, this adds up to around 90 minutes of recovered time per day, almost eight hours per week.
Executive management: when market analyses take days
At executive level, the time trap looks different. Here it is not about many small tasks, but about a few knowledge products that are all the more time-consuming: market analyses, pitch decks, strategy reports, board papers, customer presentations.
Creating a pitch presentation from a 30-page customer briefing traditionally takes between four and eight hours. Most of this is structural work, consolidating research and slide design. The actual subject-matter thinking, in other words the value-adding part, often accounts for less than a third of the time.
When Nex generates the structure, content and slides from the briefing itself in the AI chat, the ratio shifts fundamentally. The AI does the structural work in minutes. The managing director spends his time on the subject-matter refinement, in other words on the part he is paid for. A market analysis based on up-to-date web sources, which used to take two days of research and one day of writing, can in this calculation be produced as a first draft in 30 to 60 minutes and then needs an hour of human refinement.
Assumption for the example calculation: with two to four such knowledge products per week (for the executive team of a business with 30 employees), five to ten hours per week are recovered.
Customer contact: response time as a lever
The third persona is not an employee of your own but the customer. And this is where there is potential that hardly shows up in any analysis.
The lever is response time. An enquiry that lands in the inbox in the evening and is answered the next morning competes with a provider whose chatbot answers in seconds. The reason is mundane: a customer who has a question at 11 pm does not wait until the next working day. They keep searching and end up with the next provider.
The telephone is also a hidden pain point. Suppose a medical practice, an insurance office or a trade business misses between 10 and 30 calls a week because nobody answers. The person is talking to a customer, on their lunch break or has already finished for the day. An AI phone assistant can take these calls. The example calculation assumes that it books 60 per cent of them directly into the calendar and passes the rest on to the right person with a context summary.
Assumption for the example calculation: a chatbot answers around 70 per cent of standard questions. For a service team with two full-time employees, this corresponds to freeing up around 1.4 FTE. A VoiceBot that handles 60 per cent of calls itself also avoids lost appointments that previously simply never came about.
The invisible layer: processes and daily tool hopping
The biggest time trap is not any single module but switching between them. If you work in the email program in the morning, maintain notes in the CRM at midday and update tasks in the project management tool in the afternoon, you lose time every day through context switching alone.
This time does not go into tasks but into switching back and forth between tools, picking up the train of thought again and entering the same information twice in different systems. Because the effort is spread over many short interruptions throughout the day, it does not appear as a separate item in any time tracking.
An integrated platform that brings AI chat, knowledge base, email, project management and process automation together in one interface shrinks exactly this layer. Added to this is the effect of Automations: recurring workflows that are set up once and then run automatically. Examples include approvals that pass through the responsible levels above a set amount, an annual GDPR audit that starts on its scheduled date, and onboardings that go through the same steps for every new employee.
Adding it up: what this means in euros
At first glance, these figures seem abstract. A concrete example calculation shows what they mean for a typical SME.
Assumptions: 30 employees, of whom 20 in administrative roles, 5 in executive or management roles and 5 in direct customer contact. Average fully loaded hourly rate: €50. Working weeks per year: 47 (after deducting annual leave and public holidays).
Administrative staff: 20 people × 90 minutes of recovered time per day × 5 days = 150 hours per week × 47 weeks × €50 = €352,500 per year.
Executive and management roles: 5 people × 7.5 hours of recovered time per week (mean of 5 to 10) = 37.5 hours per week × 47 weeks × €50 = €88,125 per year.
Customer contact relief: A relief of 1.4 FTE in the service area corresponds to around €63,000 in fully loaded costs per year (1.4 × €45,000 average fully loaded annual service salary).
Subtotal: Around €503,000 per year, for a 30-employee business. Platform costs, by contrast: thirty users on the Base plan at €29 per user per month result in an annual licence of €10,440. In terms of time alone, the investment pays off by a factor of 48. The times recovered per role are orders of magnitude, not measured values.
This calculation does not include further effects: faster decisions, lower onboarding costs for new employees, customer churn avoided through night-time response time and the silent risk that an unanswered telephone causes day after day.
What the studies do not show
Efficiency can be measured. Other effects cannot, and that is precisely why they are left out of Excel models.
Consistency. A central AI knowledge base gives every employee the same, correct answer. There are no longer three versions of a travel expense policy, only one of which is current. Whoever asks reliably gets the right answer, regardless of whom they ask and how long that person has been with the company.
Onboarding. A new employee who can ask about all internal processes, policies and contacts via chat from day one finds their way around faster than if they had to get every answer from colleagues. With a workforce that is partly renewed every few years, this is a significant lever.
No loss of knowledge. In the past, when a long-standing employee left the company, part of the knowledge went with them. A knowledge base that has taken in all correspondence, all contracts and all minutes over the years keeps this knowledge in the company, even after the person has left.
Understand data flows. At Nexoria, files, the knowledge index and semantic search are located in Germany; Mistral AI generates the AI answers in France. Other services and enabled integrations are listed under Security. This information helps you assess your own deployment.