Outsourcing vs insourcing: the position-by-position decision matrix for SMEs

Before opening a job description or contacting an offshore provider, you need four data points per function. Not three. Not six. Four. Each criterion is scored from 1 to 5. The combined score gives you a clear direction. Here is how each axis works.

A one-off task justifies neither a permanent contract nor a dedicated collaborator. A daily task does. The frequency criterion separates what belongs to a project (website redesign, ERP migration) from what belongs to continuous operations (accounting entry, customer support, commercial follow-up). Score 1 if the task occurs less than once a month. Score 5 if it occupies someone every day. Between 1 and 2: freelance or one-off service. Between 3 and 5: you need someone on a continuous basis, either in-house or through a dedicated outsourced collaborator. Frequency immediately eliminates false hiring needs. How many SMEs have hired a full-time developer when the real need represents 15 hours per week? And how many refuse to outsource a commercial follow-up role that runs 8 hours a day, 5 days a week? The frequency filter sets things straight. If the workload is constant and predictable, outsourcing with a dedicated collaborator becomes an option to evaluate seriously. If the workload is erratic, you will lose money either way, unless you find a flexible model.

Criticality measures the direct impact on your revenue or operational survival if the task is poorly executed or interrupted. A production bug in your SaaS application: criticality 5. Updating your LinkedIn page: criticality 1. Score 5 when a failure blocks your activity or causes you to lose clients. Score 1 when a 48-hour delay changes nothing. The prevailing belief says: "what is critical stays in-house". This belief is false if left unqualified. Critical customer support can be outsourced with a dedicated collaborator, trained on your processes, connected to your CRM, provided that management is structured and the person is not juggling ten clients at once. What matters is not where the collaborator is located. It is the level of control you retain. A demotivated French employee on a critical task puts you just as much at risk as a poorly managed provider. The real question: do you have the means to supervise and correct in real time? If yes, criticality alone is not enough to decide between in-house and outsourced. You need to cross it with the other axes. We will see how to read the combinations in section 3.

Complexity measures the time required for a new collaborator to become autonomous on the task. An SDR following up prospects on a structured script: complexity 2. A back-end developer who needs to understand a legacy architecture of 40,000 lines: complexity 5. Replacement cost measures what you lose (time, money, knowledge) when the person holding the position leaves. If the knowledge is documented and the process is reproducible, the replacement cost is low. If all the expertise lives in one person's head without documentation, the replacement cost skyrockets. Score each axis from 1 to 5. A position with low complexity (1 or 2) and low replacement cost (1 or 2) is an ideal outsourcing candidate: ramp-up is fast, dependency on the individual is low, and operational risk in case of departure is contained. Conversely, a position with complexity 5 and replacement cost 5 requires either strong insourcing, or an outsourcing model with thorough knowledge transfer and permanent documentation. As l'analyse du coût réel du turnover offshore shows, ignoring replacement cost means underestimating the true TCO of any option.

The theory is set. Let us move to application. Here is how the matrix behaves across the most common functions in French SMEs with 1 to 50 employees. For each function, a score and a recommendation.

Take an SDR (Sales Development Representative) whose job is to prospect, qualify and book meetings. Frequency: 5/5, it is daily work. Criticality: 3/5, one day without prospecting does not kill anyone, but a week without pipeline does. Complexity: 2/5, the script exists, the tools are standardised. Replacement cost: 2/5, a well-trained new SDR gets back up to speed within two weeks. Total score: 12/20. Classic outsourcing profile. The frequency justifies a full-time position. The low complexity and low replacement cost make the outsourcing risk minimal. And the moderate criticality means management can remain light: a weekly check-in is enough. Today, an SDR on a French permanent contract costs between 3,200 and 4,500 euros gross per month; including employer contributions it often exceeds 5,000 euros. A dedicated collaborator, trained on your tools, integrated into your CRM, costs three times less and produces the same call volume. The commercial follow-up position follows exactly the same logic. If you are leaving quotes without follow-up because nobody is chasing them, the problem is not the process: it is the absence of resource. And that resource does not need to be sitting in your open-plan office.

A full-stack developer. Frequency: 5/5. Criticality: 4 to 5/5 (your product depends on code quality). Complexity: 3 to 5/5 (depending on the stack and codebase age). Replacement cost: 3 to 5/5 (architectural knowledge is often concentrated in one person). Total score: 15 to 20/20. Tense decision zone. This is not a categorical no to outsourcing. It is a signal: more structure is needed. An outsourced developer in a critical role works if knowledge transfer is organised from sprint 1, if documentation is imposed as a deliverable, and if the code review process is asynchronous and rigorous. A French developer costs you 5,000 euros per month and you get one pair of hands. For the same budget, three dedicated developers can work on your stack, provided technical management is structured. The matrix does not say "outsource everything". It says: when criticality and complexity are high, increase the management level, document everything, and choose a model where the collaborator is 100% dedicated to your project. Not a production centre dispatching tickets across ten clients. A developer who knows your code, your conventions, your tools. La méthodologie TCO sur 12 mois lets you precisely calculate the gap between insourcing and dedicated outsourcing on these technical profiles.

You have scored each position on the four axes. The total score ranges from 4 to 20. Here is how to read it, when to switch, and the pitfalls to avoid in interpretation.

A score between 4 and 10 means: recurring task, moderate criticality, low complexity, easy replacement. This position has no business being on your French payroll. Every month you pay a local permanent contract for this function, you are burning cash that you could invest in your core business. Functions that consistently fall into this zone: accounting entry, invoice chasing, level-1 customer support, lead qualification, CRM updates, e-commerce back-office, telephone secretarial services, candidate sourcing. These are high-frequency, low-technical-complexity positions. Their value comes from volume executed correctly, not from rare expertise. The model that works here: a dedicated collaborator, on a local permanent contract, trained on your processes, integrated into your tools. Not a freelancer who disappears when something better comes along. Not a call centre that pools its agents. One person who works for you and only for you. For high-volume support functions, l'outsourcing multifonction depuis un prestataire unique allows you to centralise several of these positions without multiplying points of contact.

Grey zone. The position could stay in-house. It could also be outsourced. The determining factor is no longer the raw score; it is your ability to structure the management. A position with criticality 4 and complexity 3 works in outsourcing if you impose: a two-week documented onboarding, weekly synchronisation rituals, direct access between your internal team and the outsourced collaborator, and living documentation. Without this framework, the risk of misalignment increases and the criticality of the position turns every error into a visible problem. This is the zone where most outsourcing arrangements fail. Not because the model is bad, but because the leader treats a score-13 position like a score-7 position: they delegate and disappear. 73 % des externalisations échouent avant 6 mois precisely for this reason. The ramp-up is not sized for the criticality level of the position. If you fall into this zone, ask yourself one additional question: do I have, internally, someone capable of supervising this outsourced collaborator during the first 90 days? If yes, go for it. If not, strengthen that point before launching anything.

High score across all four axes. Daily task, critical to the business, complex to execute, costly to replace. These are your key positions: CTO, senior sales manager, software architect, CFO. The first option is insourcing. If you have the budget and the recruitment succeeds, keep this position on a French permanent contract. Proximity, direct relationship, and shared culture bring real value in these strategic functions. The second option exists when the budget does not follow. A CTO at 80,000 euros per year is out of reach. A senior sales manager at 65,000 euros, likewise. In this case, outsourcing takes the form of a dedicated senior profile, with an adapted level of management and follow-up. The model does not change in principle (one collaborator, one client), but recruitment is more demanding, onboarding is longer, and management is tighter. The matrix does not forbid you from outsourcing a score-18 position. It tells you: if you do it without structure, you will regret it. And if you do not do it because you cannot afford to hire in France, you will be left with a gap in your org chart. The real risk is not outsourcing. It is inaction that leaves a critical position vacant for six months because local recruitment has failed three times. As l'analyse des recrutements locaux ratés shows, the cost of absence often exceeds the cost of risk.