The first 90 days of a digital transformation in a small company should produce one measured improvement to one process, owned by one person, before any significant software is bought. Most advice on the subject assumes a programme office and a budget line that renews every year. A 40-person company has a handful of people who are already fully booked, a system everyone works around, and money that had to be argued for.
That is a different problem with a different answer: two weeks of looking, two weeks of choosing, a month of simplifying and a month of shipping and measuring. After the plan come the questions that decide whether it survives a real company: cost, size, failure modes, and a worked example.
What does digital transformation mean for a company of 20 to 200 people?
For a company of 20 to 200 people, digital transformation means changing how a few core processes run, one at a time, so that each change is measured and kept. It is not a platform purchase or a multi-year programme. The work is done by people who already have full-time jobs, so the scope has to fit beside those jobs, and each step has to show a result before the next one is funded.
That describes most of the Swiss economy. The Federal Statistical Office counted 52,476 small enterprises with 10 to 49 employees and 9,791 medium-sized ones with 50 to 249 in 2023. SMEs make up more than 99% of market enterprises and employ roughly two thirds of the people who work in them.
Smaller firms also start further back. Eurostat's Digitalisation in Europe reports that in 2025, 71% of EU SMEs reached a basic level of digital intensity, against 96% of large businesses and about 20 points short of the EU's 2030 target of more than 90%. Switzerland is not in that series, so treat it as the shape of the problem rather than a Swiss measurement.
The more useful number is who does the work. In 2024 only 14% of small EU enterprises employed an ICT specialist, against 78% of large ones, and in only about a third of small enterprises did their own staff perform ICT functions, according to Eurostat's survey of ICT specialists in enterprises.
Read that as a design constraint, not a deficiency. The plan below assumes no IT department and an outside supplier who looks after the servers. The operations lead, the finance lead and the managing director do most of the work, a few hours a week each.
Days 1–15: find where the time actually goes
Not a full maturity assessment. The SECO SME portal recommends assessing digital maturity before any change or tool purchase, and the instinct is right: look before you buy. At this size, narrow the look to three processes your people complain about, and follow one real order, one invoice and one hire from start to finish.
Write down every handoff, every re-keying and every wait. The columns: step, person, system, time working, time waiting, and whether anything was redone. Sit with the person doing the work rather than asking them to describe it, because the description is always the official version.
You are looking for one number: how much of the elapsed time is waiting versus working. In the companies we have walked through it usually lands between 60% and 85% waiting. Waiting is cheap to fix compared with buying software, because most of it is an approval nobody needs, a handoff between two people who could talk, or a document sitting in someone's inbox while they are on holiday.
Do this with a notebook or a shared spreadsheet. Instrumentation comes later, if at all. The notebook is the baseline for day 90.
Days 16–30: pick one process and one owner
One. Not a portfolio. The candidate should meet three tests. It happens often enough to measure inside a quarter. Someone can be made accountable for it. And improving it does not require another department's permission.
That last test eliminates most attractive-looking candidates, and it protects your quarter. A process that crosses sales, production and finance holds the biggest gains, and it is also where a first project dies in meetings.
The owner is the person whose week gets worse when the process is slow, not the person most enthusiastic about technology. They need the authority to change how the process runs and a few protected hours a week, written into the calendar. In a 20-person company that is often the managing director. In a 150-person company it is usually a department head, and the managing director's job is to defend their hours.
What should an SME digitalise first?
An SME should digitalise first whatever is currently held together by one person's memory or one shared spreadsheet. That is where the operational risk sits and where the payback is fastest, because you are replacing an unmanaged dependency rather than optimising something that already works. Order entry, quotation and time to invoice are the usual answers in Swiss SMEs.
| Candidate | Sign it is the bottleneck | What to measure | Typical first change |
|---|---|---|---|
| Quotation | Customers chase quotes; one person knows the prices | Days from request to quote sent | One price list, one template, fewer approvals |
| Order entry | Orders retyped from email into the system | Minutes per order, error rate | Structured order form or import |
| Time to invoice | Work finished weeks before it is billed | Days from delivery to invoice | Invoice triggered by delivery, not by month end |
| Onboarding a hire | Accounts and equipment arrive late | Days until the person can work | One checklist with one owner |
Analytics, dashboards and AI are almost never the right first move. They describe problems you already know you have. If AI is on the owners' minds, the preparation is in the five data foundations an SME needs before AI, and most of it overlaps with the process work here.
Days 31–60: change the process before buying anything
The most common expensive mistake is buying software to automate a process nobody has simplified. You then pay to encode the mess, and every future change costs money.
Take the process you picked. Remove the steps that exist only because of how the old system worked. Reduce approvals to the minimum anyone would actually defend. Agree on one place where the data lives. Then see what is left. Frequently it no longer needs the software you were about to buy.
Where it does, you now have a specification instead of a wish list, which is the difference between a two-week vendor evaluation and a six-month one. Be wary of concluding that the answer is an ERP because larger companies have one. Eurostat's e-business figures show that in 2025, 41% of small EU enterprises used ERP software against 89% of large ones. Plenty of well-run small firms work without one. The signals that you have outgrown spreadsheets and point tools are set out in when a small company actually needs an ERP, and running the selection properly is covered in ERP selection for the Swiss mid-market.
Days 61–90: ship it and measure against the notebook
Deploy to the one process, with the one owner, and compare against the baseline you wrote by hand in the first two weeks. Not against a projection. Against the notebook.
Measure the same way as the first time, following real work end to end. If the new figure comes from a system report and the old one from a notebook, you are comparing two methods, not two processes.
If the number did not move, stop and find out why before starting the second process. A first improvement that cannot be demonstrated makes the second one politically harder, and in a small company political capital is the scarcest resource you have. The structure for turning that measurement into a case the owners can judge is in how to measure ROI on a digital transformation.
How much should a small company budget?
A small company should budget mainly for its own people's time in the first 90 days, plus a modest amount of outside help to structure the work. Software spend belongs in the second quarter, once the process is simplified and the requirement is real. Any proposal that starts with a platform and a licence count before anyone has followed an order end to end is selling a solution to an unexamined problem.
Count the internal time honestly, because it is the largest line and the one most often left out. A useful reference is the Federal Statistical Office's labour cost figure: an average of CHF 63.62 per hour worked across the Swiss economy in 2020, including employer social contributions. Three people giving four hours a week for thirteen weeks is 156 hours, or roughly CHF 10,000 at that average. Your own rates will differ, but the order of magnitude is the point: it is real money and should sit in the plan.
Outside help is worth paying for when it brings method, not when it replaces the owner. How to compare offers is in what a Swiss company should actually pay for consulting. If the work turns into genuine product or process innovation with a research institution, Innosuisse's innovation cheque funds a preliminary study of up to CHF 15,000 for organisations with fewer than 250 full-time employees. It pays the research partner, not a software vendor, so it fits a feasibility question rather than an implementation.
What changes between 20 and 200 people?
Between 20 and 200 people, the method stays the same but the owner, the communication and the dependencies change. At 20, the managing director usually owns the first process and can change it in a week. At 200, a department head owns it, more people must be told before anything changes, and the process touches more systems, so the quarter needs a written scope and a fixed weekly review.
At the small end, the owner is also running sales and payroll, and the change quietly loses every contest for their time. Protect two half-days a week, and treat a missed week as a finding.
At the larger end, the risk is a steering group that turns one process into five. Keep the scope to one process anyway. A company of 200 has more processes to improve, not more capacity to improve them in parallel.
In between, around 60 to 120 people, the usual trap is the outside IT supplier. The supplier controls the systems, nobody inside owns the configuration, and every change becomes a ticket. Name the internal person who decides what the supplier does, before day 31.
What usually goes wrong
The failures are predictable, and none is technical. They are the patterns in why ERP programmes stall in Swiss mid-market companies, compressed into a quarter.
| Failure | What it looks like | Fix |
|---|---|---|
| Too many fronts | Five workstreams, none finished | One process, one quarter |
| No named owner | Everyone consulted, nobody accountable | One person who can say no |
| Tool before process | The mess, now in software | Simplify first, buy second |
| No baseline | "It feels faster" | The notebook from the first two weeks |
| Consultant-dependent | Nothing works when they leave | Handover with a date |
The pattern is the same for AI pilots, which stall for want of an owner far more often than for want of a model. That argument is made in getting AI past the pilot stage in Swiss companies.
A worked example: a 45-person manufacturer
Take a hypothetical 45-person maker of machine components with an old ERP, a sales team of four and quotes that take a week. The figures below are illustrative, not a client case.
In the first two weeks the operations lead follows ten quote requests from email to sent quote. Each takes about five working days, of which roughly half a day is actual work. The rest is waiting: for the one person who knows the current surcharges, for the managing director's approval, and for a drawing check that could run in parallel.
In the next two weeks the sales lead is named owner and quote turnaround is chosen as the one measure. In the month that follows, the surcharges go into one price sheet that sales can read, the approval threshold is raised so that only unusual quotes need the managing director, and the drawing check moves to the start. No software is bought.
In the last month the operations lead follows another ten quotes. If turnaround has fallen from five days to two, the company has a result to show the owners and a specification for a quote configurator if one is still needed. If not, the notebook shows where the waiting moved, and that is the next fix. The same logic applied to the shop floor is in Industrie 4.0 in Swiss manufacturing.
After the first quarter
The second process is easier, because you now have a method, a demonstrated result and an owner who has done it once. That compounding is the actual transformation. Not a platform, and not a roadmap deck.
Before choosing the second process, write the one page that says where all of this is heading: the business outcome, the processes that produce it, and what you will stop doing. The format is in how to write a digital strategy for an SME on one page. Our digital transformation practice runs engagements in exactly these quarter-long increments, and our industrial page covers what that looks like for manufacturers and distributors.
Sources
- Federal Statistical Office (BFS) — Kleine und mittlere Unternehmen, enterprises and employment by size class, 2023
- Eurostat — Digitalisation in Europe, 2026 edition, digital intensity of SMEs and large businesses, 2025 data
- Eurostat — ICT specialists: statistics on hard-to-fill vacancies in enterprises, who employs ICT specialists and who performs ICT functions, by size
- Eurostat — E-business integration, use of ERP and CRM software by enterprise size, 2025
- SECO SME Portal — Digitalisation, a multi-step process, maturity assessment before change or tool purchase
- Federal Statistical Office (BFS) — Arbeitskosten, labour cost per hour worked, 2020
- Innosuisse — Innovation cheque, up to CHF 15,000 for a preliminary study with a research partner
Related reading
- How to write a digital strategy for an SME: it should fit on one page
- How to measure ROI on a digital transformation before you fund it
- When does a small company actually need an ERP?
- What should a Swiss company actually pay for consulting?
- Digital transformation consulting
Working on this right now?
Tell us where you are in two questions. A consulting partner reviews every enquiry within 2 business days.
