Preparing for the SME Voucher
The voucher programme’s success depends on matching with the right supplier. The setup in Türkiye’s AI Action Plan runs like this: the SME declares its need through a digital platform, programme management runs a pre-assessment, then matches the business with a suitable supplier from an accredited pool.
The weak link sits in the first step. A business that describes its need wrongly gets matched wrongly. Then the voucher gets spent and no result appears. What is lost is not the money but four months.
What Is the Problem?
BU BÖLÜMÜN ÖZETİ
- The need gets described by tool name
- Whatever the supplier sells gets bought
- The work ends when the voucher ends
Three scenes repeat in supported projects.
The need gets described by tool name
“We need a chatbot” is not a need definition. It is a tool name. The actual need might be this: customer questions answered at night too. The same need can have three different solutions, and the chatbot may be the most expensive of them.
Whatever the supplier sells gets bought
If the need is vague, the supplier’s product shapes the need. That is a reversed order. The business ends up owning a tool for a problem it never had. The tool stops; the problem continues.
The work ends when the voucher ends
The most common ending. The system runs while support lasts, then goes unmaintained. Because no maintenance plan or continuity condition was discussed at the start. With nobody owning it, the system gets quietly abandoned.
Why Does It Happen?
BU BÖLÜMÜN ÖZETİ
- Applications go in without a process scan
- There is no measurement point
- Continuity belongs to nobody
Three reasons.
Applications go in without a process scan
The voucher gets announced and someone says “let’s do something”. But an application made without knowing which process needs support is a random choice. Random choices give random results.
There is no measurement point
The current state of the work was never measured. So the gain cannot be measured afterwards either. The plan sets a 10-percent productivity threshold; where the before is unknown, that threshold cannot be checked.
Continuity belongs to nobody
The supplier installs and leaves. If nobody inside owns the system, the work slowly stops. This is not a technology problem but an ownership problem.
How Is It Done?
BU BÖLÜMÜN ÖZETİ
- Step 1: describe the need as an outcome
- Step 2: take the baseline measurement
- Step 3: write the continuity terms upfront
Preparation completes in three steps.
Step 1: describe the need as an outcome
Do not write a tool name; write a result. “Order questions answered without human intervention, average response time down from two hours to five minutes.” That description leaves the supplier room. They find the cheapest, most suitable solution. Write a tool name and you are stuck with that tool.
Step 2: take the baseline measurement
Measure the current state for two weeks before applying. How many hours, how many items, how many errors. That measurement does two jobs. It gives your application a concrete starting point, and it proves the gain at the end. An unmeasured project ends with “it went well, I think”.
Step 3: write the continuity terms upfront
Discuss three things with the supplier from the start. In whose name will the system be set up, and who holds the data? After support ends, who maintains it and at what cost? Who inside will learn it and take it over? These three questions are the voucher-side counterpart of the sovereignty framework. Unwritten in the contract means unasked.
How Long, Where to Start?
BU BÖLÜMÜN ÖZETİ
- Two weeks of prep, four months of project
- The real cost is staff time
- First step: pick one process and start measuring
Short preparation, long-running return.
Two weeks of prep, four months of project
The need description takes a day, the baseline measurement two weeks. The plan then allows at most four months from experiment to pilot. So preparation is small next to the project. But it decides the outcome.
The real cost is staff time
The voucher covers software, cloud and service costs. What you pay is the hours. At least one person needs a few hours a week. Name that person upfront. Projects where “we’ll all look after it” do not move; everyone’s job is nobody’s job.
First step: pick one process and start measuring
Filter with three questions: which job repeats, eats time and is measurable? Pick one. Start measuring today. When the call opens you will hold a defined need and a real baseline number. Those two separate your application from the rest — and most applications have neither.
The Common Mistake
BU BÖLÜMÜN ÖZETİ
- Using the voucher instead of training
- Choosing the biggest problem
- Talking to one supplier and not planning the end
Three traps.
Using the voucher instead of training
If the team does not know how to use the tool, the voucher will not help. Basic capability first, tools second. That is why the level ruler comes before the voucher.
Choosing the biggest problem
Picking the most complex process for the first project is tempting. But complex processes take longer to set up and are harder to measure. Spend the first voucher on a settled, well-patterned job. Hard problems come in the second round.
Talking to one supplier and not planning the end
Even with matching in place, speak to more than one provider where possible. Different offers against the same need also clarify the need itself. The second omission happens at the end. If the project’s finish is not written at the start, it does not finish; it just stops. End date, handover plan and maintenance cost belong in the contract.
Frequently Asked Questions
Sık Sorulan Sorular
The most productive results come from repeating work with clear rules. Order and request classification, document reading and data entry, first response to customer questions, stock and demand forecasting, image-based quality control. They share three traits: countable output, orderly input, tolerable error margin. Do not pick sensitive work whose decisions reach the customer directly as your first project.
The match is a suggestion, not an obligation. If you described your need properly and the offer does not fit, object or ask for another provider. Accepting a bad match and regretting it later is the most expensive route. The programme tracks success thresholds too, so a mismatch hurts both sides.
Voucher programmes are generally grant-like, and repayment is not expected from a project run in good faith. But terms vary by programme; read how failure is defined before signing. The real loss is not money but time: four months spent, no result. What prevents that is the measurement point set at the start.