This text is built around those breaking points. An in-house hire breaks on leave and narrow specialisation. A freelancer breaks on availability and on being a single point of failure. An agency breaks on the queue and on anonymity, and we write that about ourselves, because we sell the third model. At the end comes the hourly threshold above which your own team costs less than our subscription.
Three models, three different distributions of risk
An in-house hire: you employ a developer exclusively. You get someone who knows your business and why that odd wholesale discount must work exactly as it does. You pay for time, not outcomes, including time when nothing happens. Availability and competence risk sit entirely with you.
A freelancer: you buy a task or a pool of hours from one person who also has other clients. You get flexibility and usually the lowest rate for real work, and pay for work delivered. Continuity risk stays with you, because nobody can step in.
An agency or software house on a subscription: you buy a team and a process, not a person, so you get several competences and a contractual commitment. You pay whether or not there is work that month. Availability and substitutability risk move to the supplier, priority risk stays with you.
What an in-house developer really costs
We print no salary figures: these are rough market ranges that differ heavily by country, so any number here would be wrong for most readers.
The rule that travels across markets: the full cost of a developer is a multiple of the gross salary. Contributions, leave, equipment, tooling, recruitment and the risk of departure sit on top of the number in the offer letter.
The items that usually vanish:
- paid leave: entitlements differ by country, and with sick days and holidays you pay for twelve months and get about eleven,
- equipment: a laptop and peripherals amortised over three years,
- licences and training: an IDE, testing tools, conferences and courses, every year,
- recruitment and onboarding: the cost of the process, plus one to three months in which the new person learns your code,
- risk of departure: with turnover around two years, that cost returns on a cycle.
A store needs not one competence but six: frontend, backend, server configuration, payments, integrations with ERP and couriers, and technical SEO. Nobody is good at all six. You hire someone excellent at two or three and buy the rest externally anyway, or accept mediocre work. Not a complaint about employees, just arithmetic.
The freelancer: a single point of failure
A freelancer is the best choice when a task has a clear beginning and end. Rebuilding a listing template, an integration with one wholesaler, performance fixes on one view, migrating a dozen email templates. Known scope, known result, no time pressure. There you get the best price to quality ratio, and no reason to pay a subscription.
A freelancer fails the moment the word "now" appears. An outage on Sunday evening. A holiday in Black Friday week. Three days of illness. No cover, because there is nobody to assign, and no documentation, because the person who remembers everything never writes it down. This is a single point of failure, exactly as one server without a backup is. The model works until the day it stops, and then there is no plan B.
The agency on a subscription: the queue and anonymity
What you buy from us and from any comparable firm: several competences under one contract, substitutability (if someone is ill, another picks the ticket up), a guaranteed response time, documentation a colleague can read, and ticket history in one place.
What you do not buy, and it is worth hearing before you sign. You have no person dedicated to you: whoever does your task on Monday does another client's on Tuesday. Your ticket sits in a queue, placed by classification, not by how much you care. There is also anonymity risk: under a poorly run contract nobody on the supplier side knows your business well enough to ask "are you sure you want this before the weekend".
So one criterion matters more than price. If the contract does not define ticket classification and the response time attached to each class, it is a promise, not a commitment. A commitment can be enforced and measured, a promise cannot. We write more on that in our text on how technical support differs from an SLA.
The break-even point, calculated on our rates
Our support packages: Basic 250 USD for 10 h, Advanced 500 USD for 20 h, Premium 1,000 USD for 40 h. An hour inside a package costs 25 USD, above the pool 40 USD. Service window: business days from 8 to 18.
Step by step, assuming you stay on Premium and buy the overage:
- 40 h: 1,000 USD,
- 60 h: 1,000 USD plus 20 h times 40 USD, 1,800 USD in total,
- 80 h: 1,000 USD plus 40 h times 40 USD, 2,600 USD in total,
- 100 h: 1,000 USD plus 60 h times 40 USD, 3,400 USD in total,
- 130 h: 1,000 USD plus 90 h times 40 USD, 4,600 USD in total.
On the other side sits the hire: nominally 168 h a month, but after leave, holidays, meetings and non productive work, realistically from 120 to 135 h.
Read the ladder in reverse, which is how to do this without quoting salaries. Take the fully loaded monthly cost of one developer in your market, the multiple described above. Subtract 1,000 USD, divide by 40 USD, add the 40 hours from the package: that is the volume at which both options cost the same. Below 4,600 USD of loaded cost the break-even falls under 130 h, inside what one person can deliver.
The conclusion that works against us: above 120 h a month your own team is cheaper in most markets and you should build it. Two costs then appear that this calculation omits. First, management: someone must set priorities, review code and run appraisals, from 10 to 20 percent of a manager's time. Second, 120 h a month is no longer maintenance but continuous development, counted differently. If you consume from 10 to 40 h, the subscription is cheaper, with no room for discussion. The billing models we covered in our text on subscriptions versus hourly billing.
The mixed model that works from 250,000 to 3.5 million USD in annual revenue
In this group of stores the best arrangement has three elements. One person on the client side who knows the business and may decide: not necessarily a developer, but someone who understands promotions and the warehouse. An external technical team on a subscription that executes. And separately, supervision of availability: monitoring and reaction to what stopped working, a different service from the changes you commission.
Separating decision from execution works because these are two competences. The person deciding must know the business context and be available during the day; the team executing must know the code and have substitutability. When both sit in one head, every holiday stops decisions and execution at once, and priorities get set by whoever called last.
Five questions to ask each of the three models
The same questions, put to all three, show the difference faster than any price comparison:
- Who picks up the phone on Sunday evening, and on what basis? With us, reaction outside business hours belongs to a separate, highest SLA level, and a single Sunday intervention outside the subscription costs 80 USD per hour. "They will surely pick up" is not an answer.
- Who takes over if this person disappears for two weeks? You want a name or a role, not a reassurance.
- Where are the code, the access credentials and the documentation? The repository should be yours.
- Who watches expiry dates of the SSL certificate, domain, module licences and platform version? A calendar task, and it needs an owner.
- Who is liable if the store is down for six hours? Check what is written, because lost sales are excluded in every honest contract, ours included.
Formal responsibility: the data processing agreement and the response time
There is one more layer, not a matter of preference but of law. The company maintaining your store has access to your customers' data, so it is a processor, and a data processing agreement is required (Article 28 GDPR). It must define the subject matter and duration of the processing, the categories of data and of data subjects, confidentiality, security measures, sub processing, help with data subject requests and breach notification, return or deletion of data, and audit rights.
With an external supplier you get this in the package, with a response time commitment. With a freelancer you usually have neither: an invoice, access to the admin panel, nothing else. That does not rule out freelancers. It means the agreement must be signed with them too, and if nobody did it, the gap is on your side, not theirs.
Run your own case: add up the hours commissioned over the last three months and compare with the calculation above. Support packages and availability supervision levels are listed at /en/sla-help-desk. If the result says hiring is cheaper, we will say the same.