Free automation platforms: what free actually costs
Four routes to automating without a subscription — built-in features, free tiers, self-hosting and scripting — and an honest accounting of what each one costs you instead of money.
Osman Ramadan11 min read
On this page
Everything free costs something. Usually time, sometimes reliability, occasionally a limit you discover at the worst moment. That is not an argument against free — plenty of automation should be free — but it is the thing missing from most lists of free tools.
This page takes the four genuinely free routes and prices each in what it actually takes from you. If what you want specifically is a comparison of the managed free tiers against each other, Zapier alternatives that are free does that in detail; this page is the wider question of whether free is the right answer at all.
What we'll cover
Four routes to free
Included features in software you already pay for. Free in the strongest sense: no new subscription, no new credential, nothing extra to maintain.
Free tiers of commercial platforms. Free until a limit, and the vendor's intention is that you exceed it.
Self-hosted open source. No licence fee, and infrastructure and maintenance that you own.
Scripts you write. No cost at all except the writing and the owning.
These are not interchangeable and they fail in different ways. The order above is roughly the order to consider them, which is the opposite of the order most people consider them in.
Free because it is already included
The most overlooked route and the one with the best economics.
What you probably already have: workflow rules in your CRM, automations in your project tool, filters and rules in email, recurring invoices and payment reminders in your accounting software, scheduled reports in your analytics, and scripting in your spreadsheet application. Google Apps Script is free with Workspace and reaches Sheets, Docs, Gmail, Drive, Calendar and Forms, which covers a surprising share of small-business automation.
Why it is genuinely better, not just cheaper. A built-in feature sees its own product's data directly, with no integration to configure, no credential to expire, and no second system to keep running. The most common automation failure — an expired authorization — cannot happen to something that was never a separate connection.
The cost: it stops at the product boundary. The moment a process spans two systems that do not know about each other, built-in features cannot help, and that boundary is the honest dividing line for when you need something else.
The reason people skip it: these features are unglamorous, poorly marketed by the vendors themselves, and usually switched off by default. Spending an hour reading what your existing tools already do is the highest-return hour in this entire subject.
Free because someone else is paying for your evaluation
Free tiers are marketing, which does not make them useless — it makes their limits deliberate.
The limit that stops you is rarely the one advertised. Volume allowances are generous enough to look reasonable; the constraint that actually bites is usually a cap on steps per automation, which rules out anything with a condition, or a daily rather than monthly allowance that does not accumulate.
What free tiers are genuinely good for: confirming a specific connection works before paying, learning a builder, and running a handful of genuinely simple automations permanently.
What they are poor for: anything the business depends on. Free tiers typically come with shorter log retention, lower priority, and no support commitment. When something fails on a Friday, you have no route to an answer and possibly no record of what happened.
The trap to watch for: building around the limit. Splitting one process into three two-step automations to fit a cap produces something permanently more fragile than the thing you were avoiding paying for. If your automation's shape is being decided by a pricing tier, you have started paying in complexity instead.
Free because you run it yourself
Self-hosted open source is free of licence fees and is the route most often misdescribed as simply free.
What it genuinely gives you: no per-run cost at any volume, data that never leaves your infrastructure, and independence from a vendor's pricing decisions. At high volume with existing infrastructure, it is dramatically the cheapest option and that is not marginal.
What it costs instead of money. A server and a database, which are cheap. Then updates, backups, monitoring, certificate renewals, and restoring it when it stops at an inconvenient moment. In a steady state that is perhaps one to two hours a month; in a bad month it is considerably more.
The arithmetic that decides it. Two hours a month of engineering time at a fully loaded rate exceeds essentially every managed plan on the market. So self-hosting is cheaper only when that time is genuinely free to you, and it is genuinely free only when somebody already runs servers as part of their job.
The failure mode to recognise: "somebody will learn." The maintenance then lands on top of their actual work, it is invisible until something breaks, and these arrangements tend to end with a migration back to a managed plan a year later, having cost more in total than paying would have.
When it is unambiguously right: regulatory constraints that rule out managed services, very high volume, or an organization where one more service on existing infrastructure costs close to nothing.
Free because you wrote it
The oldest route and still frequently correct.
A script plus a scheduler handles an enormous amount of real automation. Pull from an API, transform it, write it somewhere, on a schedule. For one or two stable sources this is an afternoon and costs nothing ongoing.
What it gives you: total flexibility, no vendor, no limits, and something you understand completely.
What it costs: somewhere to run it, and ownership. A script knows nothing about retries, alerting, or credential rotation unless you write that too, and writing that properly is most of the work. Scripts also accumulate: every organization of a certain age has cron entries nobody can account for, written by people who left.
The honest test: will the person who wrote this still be here, and still willing, in a year? If yes, scripting is excellent. If the answer is uncertain, you are creating a dependency on an individual, which is the thing automation was meant to reduce.
Pricing the four against each other
| Route | Money | Time to set up | Ongoing cost | Fails when |
|---|---|---|---|---|
| Built-in features | None | Minutes to hours | None | Process crosses systems |
| Free tier | None | Hours | None until the limit | You need a third step |
| Self-hosted | Server only | Days | 1–2 hrs/month | Nobody owns the box |
| Your own script | None | Hours to days | Yours to maintain | The author leaves |
The pattern: the routes that cost no money cost attention, and attention is the scarcer resource in almost every small team. That is why the built-in features win so often — they are the only option on this list that costs neither.
When paying is the cheaper option
Worth stating plainly, because it is the conclusion most free-tool lists avoid.
When you spend more than an afternoon working around a limit. An hour of your time is worth more than the nine dollars that would have removed the constraint.
When the process matters. Once a failure would cost you something real, the absence of support, the short log retention and the lower priority stop being acceptable trades.
When nobody owns the infrastructure. Self-hosting without an owner is a liability that has not surfaced yet.
When you are building complexity to avoid a subscription. The clearest signal, and usually only visible in hindsight.
When the time would earn more elsewhere. A solo business or a small team has a genuine hourly rate. Two hours a month maintaining free infrastructure against a twenty-dollar plan is a straightforwardly bad trade at almost any rate.
The genuinely free options are the built-in features you already have, and they are worth exhausting first. Beyond that boundary, the question stops being what is free and becomes what costs least in total. For automations that span systems, that calculation frequently favours a managed platform: on CodeWords you describe the process in plain language and Cody, the automation builder, builds it, connects it to the tools you already use, and deploys it, with no infrastructure to own. Automations connect to more than 3,000 integrations. The free plan covers light use, with Pro at $39 per month and Business at $100 per month as usage grows; details are on the pricing page.
Frequently asked questions
What is the most genuinely free option?
The automation features already included in software you pay for. No new subscription, no credential to expire, nothing extra to maintain, and they see their own product's data directly. Spend an hour finding out what yours already does before looking anywhere else.
Can a business run entirely on free automation?
A small one, largely yes, if the processes stay inside individual products or are simple enough for a free tier. The constraint arrives with cross-system processes that need more than two steps, which is where every free tier draws its line.
Is self-hosting actually free?
Free of licence fees, not free. Budget one to two hours a month of somebody's time in a steady state, which exceeds most managed subscriptions when costed honestly. It genuinely pays at high volume, under data residency constraints, or where someone already runs infrastructure.
Why do free tiers limit steps rather than volume?
Because steps are what make an automation useful. A two-step cap permits a trigger and one action, which demonstrates the product and cannot run a real process with a condition in it. The volume allowance is the number people compare; the step cap is the one that decides.
What happens to my automations if I stop paying after upgrading?
Usually they are disabled rather than deleted, and anything exceeding the free limits stops working — a five-step automation on a two-step plan will not run. Check the specific behaviour before downgrading, because "kept but disabled" and "deleted after a period" are both common.
Is a free trial the same as a free tier?
No, and vendors are not always clear about it. A tier continues indefinitely; a trial ends. Confirm which you are getting before building anything you intend to keep.
Should I combine free tiers across several tools?
It works and it produces an estate nobody can debug. Three vendors means three places to check when something fails and three accounts to maintain. The subscription that avoids it is almost always the better deal.
How do I know which route my problem belongs to?
Ask whether the process stays inside one product. If it does, the built-in features are almost certainly enough and everything else is unnecessary. If it crosses systems, free tiers cover the simple cases and run out at the first condition. Beyond that you are choosing between owning infrastructure and paying a subscription, which is a question about whose time is scarcer rather than about tools.
Does free get more expensive as I grow?
The routes diverge sharply. Built-in features stay free indefinitely. Free tiers get abruptly expensive at the moment you cross a limit, which is by design. Self-hosting gets cheaper per unit of work as volume rises, which is exactly why it wins at scale and loses at small scale.