Choosing a destination
Onshore, nearshore or offshore: how a Portuguese SME decides
Three questions settle almost every case: what data moves, how much time-zone overlap is indispensable, and what a point of friction per call is worth.
Corpshore Portugal editorial team
Written by the team that builds these operations. No individual byline: this is internally reviewed work, not personal opinion.
Published
Where does the decision start?
With the data, not the price. If the work touches health data, public sector data or specially sensitive data, the decision is made: it stays inside the European Economic Area and the conversation becomes where in Europe, not how much is saved in Manila.
If it does not, the field opens and the second question becomes the time zone. Work that needs real-time conversation with the in-house team works badly at more than three or four hours' difference. Asynchronous work, such as document processing or back office, tolerates twice that.
What does each model actually save?
Taking Lisbon as one hundred: Porto and Braga sit twelve to twenty per cent below without leaving the country; Poland around twenty-two per cent below, still inside the EEA; Brazil around forty-five per cent below, with standard contractual clauses; Uzbekistan and the Philippines more than sixty per cent below, with a time difference that changes the operating model.
Gross saving is rarely net saving. A cheaper destination with high attrition, more supervision and a lower first-contact resolution rate can end up more expensive per case resolved, which is the unit that matters.
What is the most frequent mistake?
Deciding everything at once. Most SMEs we speak to end up in a mixed model: what is sensitive and conversational stays near, what is high-volume and asynchronous goes where it is cheaper. Splitting by type of work rather than by department usually works better.
The second most frequent mistake is deciding without measuring the baseline. An operation that does not know how many cases it resolves per person per day, nor what a resolved case costs it today, cannot rigorously assess any proposal, and ends up comparing a supplier's rate with an internal number that was never calculated.
Which work travels well and which travels badly?
Work with clear rules, verifiable output and little dependence on context that is not written down travels well: document processing, data entry and validation, reconciliations, first-line support over a stable knowledge base, triage and classification. These are processes that survive being written down.
Work depending on relationships, judgement on unprecedented exceptions, or accumulated tacit knowledge travels badly. Key account management, negotiation, decisions on cases nobody anticipated, and anything where the right answer depends on knowing what happened two years ago with that particular customer.
There is a middle zone where most decisions are won or lost: work that looks routine and whose exceptions consume most of the time. If nobody has measured what percentage of volume is exceptions, the decision is being taken on the normal case, and the normal case is not what costs.
How long does it take to decide and to execute?
The decision, taken carefully, takes between six and ten weeks: four to six understanding the work as it is and measuring real volumes and exceptions, and two to four comparing models with loaded numbers on both sides. Shortening this is possible, and it is the origin of most outsourcing that goes wrong.
Execution to steady state takes between four and six months for a mid-sized operation, and the bulk is neither recruitment nor training: it is capturing the knowledge that today lives in the heads of the people doing the work. Without protected time from those people, that phase does not happen and the rest collapses onto it.
For an SME, the lowest-risk path is a pilot with ten to twenty per cent of volume over eight to twelve weeks, with the current team still standing. It costs more in the short term than transferring everything at once, and is substantially cheaper than discovering a problem with a hundred per cent of the volume already moved.
What should you ask a supplier?
Six, and the answers quickly separate those who have measured from those who have not. What is your attrition on this account and this role? How many people are dedicated and how many shared? Is the rate per contracted or per productive hour? Who does quality assurance and against what rubric?
And the two most revealing: what happens if service levels are missed three months running, and how much time and money it costs to exit this contract and bring the work back. A supplier who cannot answer the last one clearly is selling a dependency rather than a service.
It is also worth asking to speak to a client in a comparable setting, and noticing whether the supplier offers one who scaled back or left and returned. References picked only from the good cases say less than one where something went wrong and was resolved.
How do you measure whether the decision was good?
Not by cost per hour, which is the number that drove the decision and the worst indicator of the outcome. Measure cost per case resolved, first-contact resolution rate, end-to-end process time, and reopening rate. Those four together describe whether the work is being done; cost per hour describes nothing.
Add a measure of quality as perceived by the end customer, separated by market and by variant where there is more than one. An average across Portugal and Brazil hides the smaller market, which is usually where a problem is growing without showing up anywhere.
And measure the baseline before transferring. The most common and most expensive mistake is starting to measure after the transition, which makes it impossible to know whether what you are seeing is improvement, degradation, or the same performance as always seen with instruments for the first time.
What do you do when it is not working?
First, separate a process problem from a supplier problem. Many outsourced operations with poor numbers are faithfully executing a process that was already bad and whose weakness was hidden by experienced people compensating for it. Changing supplier does not fix that and costs another transition.
Then look at attrition on the supplier's side. If the team on your account turned over completely in nine months, the problem is neither training nor procedure: nobody stays long enough to get good. That is a conversation about conditions and job design, not about service levels.
And have the exit plan ready before needing it. Knowing how long it takes to bring the work back, what documentation is yours by contract and in what format, and who at the supplier is obliged to hand it over. Anyone starting these questions during the crisis negotiates from the worst possible position.
Which transition costs belong in the calculation?
The current team's time, which is the largest and the least budgeted. Knowledge capture, training and parallel running consume weeks from people who still have their own jobs, and treating it as zero cost because they are already on payroll is the most common way a business case looks better than it is.
Parallel running during the pilot, which is double work for six to twelve weeks. It is expensive and it is the only real safety net, and cutting it to save money trades a known cost for an unknown risk in the worst possible place.
The temporary performance dip, which always happens and whose depth and duration can be estimated. A ten to twenty per cent productivity drop over two to three months is normal in a well-run transition, and planning for it avoids the premature conclusion that the decision was wrong.
And the exit cost, which should be calculated before entering. How much time and money it would take to bring the work back, what documentation is yours by contract and in what format. Anyone who cannot answer this at the outset will find out while negotiating from a weak position.
What governance structure is needed?
Less than most contracts define and more than most SMEs set up. The essentials are one named counterpart on each side with real authority, a short weekly operational meeting and a monthly review with numbers, and an escalation path that does not depend on who happens to be available.
The monthly review should look at causes rather than only indicators. A service level met for three months with attrition rising is not a healthy operation, it is one consuming its reserves, and only someone looking at both numbers at once sees that.
The quarterly review should be able to conclude that the design is wrong. Many contracts have enough governance to manage execution and none to question the decision, and they end with operations meeting indicators for years without ever asking whether they are still the right model.
And names rather than roles. An escalation path that says account manager works in normal conditions and fails at eight on a Friday evening, which is when it gets used. Names, phone numbers and a named deputy is the minimum, costs nothing and works.
Frequently asked questions
- What is the maximum workable time difference?
- Around three to four hours for conversational work; far more for asynchronous work.
- Can I mix models?
- You can, and it is the most common outcome. Split by type of work, not by department.
- Is the cheapest the biggest saving?
- Not always. Measure cost per case resolved, not cost per contracted hour.
- When is the decision already made?
- When health, public sector or specially sensitive data is involved.
- How do you compare cost between destinations?
- Loading both sides the same way and measuring per productive hour, not per contracted hour.
- How much does attrition weigh in the decision?
- A great deal. A cheap destination with high attrition can cost more per case resolved.
- What always stays inside the EEA?
- Health data, public sector data and special categories of personal data.
- Where do you start a mixed model?
- With asynchronous, high-volume work, which is what moves most easily without losing quality.
Let us look at the numbers for your case
Tell us which processes you want to outsource, in which languages and at what volume. We come back with a euro estimate and an operating design, with no commitment.
We reply within 6 hours on working days. If you would rather write: info@corpshore.solutions