Hire someone, build it yourself, put an agency on retainer, or buy a fixed-price build. Here is the real cost of each, including the parts nobody quotes you, and an honest account of when it is not us.
The options
| Factor | DIY (Zapier / Make) | Hire Someone | Retainer Agency | Aplos AI |
|---|---|---|---|---|
| Upfront cost | Tool subscription only | $0, plus recruiting spend | Often a setup fee, then month 1 | $2,000–$25,000 one-time |
| Ongoing cost | Tool fees that scale with task volume | Salary, taxes, benefits, forever | Monthly retainer, forever | $0/month to Aplos |
| Hidden costs | Your hours building and debugging | Recruiting, ramp time, turnover | Lock-in, scope creep, exit cost | Vendor tools, $20–150/mo direct |
| Time to value | Weeks to months (learning curve) | 2–4 weeks to hire, then ramp | Fast start, open-ended finish | 1–2 weeks single, 2–3 multi-tool |
| Technical skill needed | High (you build it) | Low (they do the task) | None (they build it) | None (we build it) |
| Quality of result | Depends on your skill and time | Depends on who you hire | Usually good, rarely finished | Built and tested against real data |
| Whose accounts is it in? | Yours | Yours, but the know-how is theirs | Often theirs, check the contract | Yours, always |
| You own it after? | Yes | Partly (lives in their head) | Only if the contract says so | Yes: full handoff docs |
| What happens if it breaks? | You fix it, on your night | Call them, wait, hope they stay | Covered, while you keep paying | 30-day fix window, then docs and a fixed quote |
| Key-person risk | You are the key person | High (they resign, it stops) | Moderate (account manager churn) | Low (documented, in your accounts) |
| Best for | Solo operators, one simple workflow | Judgment work, real daily volume | Continuous change every month | SMBs who want it done right once |
Cost columns describe how each pricing model behaves, not a specific vendor quote. Dollar figures elsewhere on this page are modeled assumptions, labeled where they appear. Aplos ranges are our published fixed prices.
Option one
This is the default answer in most small businesses, and it deserves more respect than automation vendors usually give it. A person is the most flexible thing you can buy. They handle exceptions, they notice when something looks wrong, they can be told to do a different job on Tuesday, and they carry context nobody wrote down. There is no workflow on earth that talks a furious customer off a ledge or reads a room on a vendor call.
Hire when the work requires judgment rather than rule-following. Hire when the work changes shape week to week, because every change is a rebuild in an automated system and a five-minute conversation with a person. Hire when the volume is high enough that a person is busy for most of a working day, since a role that is half idle is a role you will resent paying for. And hire when the work is relational, because customers know the difference between a person and a sequence, and in some businesses that difference is the product.
The clearest signal is this: if you cannot write down the rules the work follows, you do not have an automation candidate. You have a job.
The salary is the sticker price, not the invoice. On top of base pay you carry employer payroll taxes, benefits, paid time off, equipment, and the recruiting spend that got them in the door. Then there is the ramp period, where you are paying full freight for partial output while somebody else on your team is training instead of working.
Modeled assumption, not a measured figure. Our own automation versus hiring breakdown published the model we use: a $40,000 base salary, employer payroll taxes around 7.65 percent, benefits at roughly 25 to 30 percent of salary, and $3,000 to $8,000 in first-year recruiting cost. That puts a $40,000 hire between $55,000 and $65,000 in year one. Your real numbers depend on your state, your benefits plan, and how you recruit.
Two costs almost nobody prices in. The first is management overhead: someone has to assign, review, and unblock this person, and that someone is usually you. The second is turnover. Administrative roles turn over, and when they do you pay the recruiting cost again, run the ramp again, and lose whatever process knowledge never made it into writing.
Two to four weeks to hire if the market is kind, then two to eight weeks of ramp before output is reliable. You are typically a month or two from the day you decide to the day the work is actually getting done well.
People break differently than software. They get sick, they take vacations, and eventually they resign. When a person is the process, their two weeks notice is your two weeks notice. Everything they knew that was never documented walks out with them, and the replacement starts from the beginning. This is the single biggest argument for turning the repeatable part of their job into a system: not to get rid of the person, but so that the system survives them.
Option two
We are not going to pretend this is a bad idea, because for a lot of people it is the correct one. Zapier and Make are genuinely good products. They have hundreds of prebuilt connectors, the editors are approachable, and a person with no engineering background can put a working automation together over a weekend. Anyone telling you that you must hire a professional to connect a form to a spreadsheet is selling you something.
If you are a solo operator or a very small team with one simple workflow, do it yourself. A contact form that creates a CRM record and sends a confirmation email is a textbook two-step automation. It will take you an afternoon, the tool will cost less per month than lunch, and paying anyone four figures to build it would be a waste of your money. We have told prospects exactly this on audit calls and sent them away with a sketch instead of a quote.
DIY also wins when you actually enjoy it. Some founders like building systems, and the fluency you gain from wiring your own stack together is worth something on its own. If you are one of those people and you have the hours, you do not need us for the first few workflows.
Two things get underestimated. The first is your time, both to build and to maintain. Building is the fun part and it is not the expensive part. The expensive part is the Thursday afternoon six months later when a vendor changes an API, the automation starts failing silently, and you spend three hours in a run history figuring out which step stopped returning data. Whatever your hour is worth to the business, that is the real bill, and it recurs.
The second is how these platforms price. Both Zapier and Make charge by consumption, per task or per operation. That model is cheap when you run a hundred records a month and unpleasant when you run tens of thousands, because a single multi-step workflow consumes several units per record. Plenty of businesses discover this the month after a marketing push works. Check your own usage against your plan's tier before assuming the cheap tier holds, and if you want the full breakdown of how the three major platforms price at volume we wrote one up in n8n vs Make vs Zapier.
You fix it. That is the whole answer, and it is fine right up until the week you are slammed. There is no support line that will debug your specific workflow, because the vendor supports the platform, not what you built on it.
Option three
The traditional agency model is a monthly fee for a standing relationship. You get a team that knows your business, a queue you can drop requests into, and someone else carrying the maintenance burden. Plenty of good agencies run this way and their clients are happy.
Retainers earn their keep when the work never ends. If your operations change every month, if you are running continuous experiments, if you have a queue of process work that will still be full a year from now, then a standing team is faster and cheaper than repeatedly scoping, quoting, and onboarding someone new. There is real value in a partner who already knows which of your CRM fields are lying to you.
The same is true if you have no internal technical capacity at all and no intention of building any. Somebody has to own the systems. If that will never be anyone inside your company, paying an outside team to own them permanently is a legitimate structure, not a scam.
The line item is the retainer. The cost that hurts is what the structure does to incentives. A retainer is paid for availability, not for completion, which means finishing your work quickly is financially unrewarded. This is not usually a matter of bad faith. It is just that nobody at a retainer agency is measured on making themselves unnecessary.
Then there is lock-in, and it is worth being specific about the mechanism. If the automations were built in the agency's accounts, on their platform tenancy, under their credentials, then what you have been buying is access, not an asset. Stop paying and the workflows stop running. Ask two questions before you sign anything: whose accounts does this live in, and what exactly do I keep if I cancel next month. The answers tell you whether you are buying or renting.
Fast to start, because there is no scoping negotiation for every request. Open-ended to finish, because there is no defined finish. The first automation might land quicker than a fixed-price build. The question is what month twelve looks like.
It gets fixed, and that is genuinely the retainer's strongest feature. Maintenance is included and you do not think about it. Just be clear-eyed that you are paying for that insurance every month whether or not anything breaks, and that the insurance evaporates the moment you stop.
Option four
This is what we do, so read the rest with appropriate suspicion. The model is deliberately narrow: we scope your workflow, quote one fixed number in writing, build it inside your accounts, hand over documentation and a walkthrough, and stop billing. A single automation runs $2,000 to $5,000. A multi-tool build that connects several systems runs $8,000 to $25,000. A full custom platform or AI system runs $40,000 to $100,000 and up. There is no hourly rate and no retainer.
Fixed-price builds fit businesses that have identified specific, repeatable, rule-based work and want it handled permanently without adding a monthly obligation. The typical fit is a company between roughly five and fifty people with two to six workflows worth automating, enough complexity that DIY would eat weeks of founder time, and not enough continuous change to justify a standing team.
It also fits people who have been burned by lock-in and specifically want the asset in their own accounts, with documentation, so they are never negotiating from a position where walking away means losing the thing.
The build price, once, plus the tools. Vendor tooling typically runs $20 to $150 per month paid directly to the vendors, and for most clients a good share of that is software they already subscribe to. Nothing goes to us monthly. If you want a second automation next year, that is a new fixed quote, and you are free to have someone else build it.
One to two weeks for a single automation, two to three weeks for a multi-tool build, measured from a signed scope. Everything starts with a free 30-minute audit call where we look at what you are running and tell you what is worth automating first, including when the answer is nothing yet.
For the first 30 days after delivery, we fix anything that is not working as scoped, at no charge. After that, you have the workflow in your accounts, written documentation, and a recorded walkthrough, which means your team or any competent builder can open it up. If you would rather we handle a fix, we quote it as a small fixed piece of work. You are never paying a monthly fee for the privilege of being allowed to ask.
If the process you want automated is still changing every few weeks, a fixed scope is the wrong instrument, and you should either wait until it settles or hire a retainer team. If you want a partner who lives inside your operations and picks up the phone daily, we are not that either. And if you have one simple workflow and a free weekend, build it yourself and spend the money on something else.
Good for this when…
Good for this when…
Good for this when…
Good for this when…
The decision
Most comparison pages end without telling you what to do. Here is the shortcut we actually use on audit calls. Find the row that matches your business, not the one that flatters it.
Build it yourself. Pick Zapier or Make, spend a weekend, and keep the money. At this size the constraint is cash, the workflows are usually linear, and the volume is low enough that consumption pricing is irrelevant. If it breaks you will have the time to look at it. Come back when you have five workflows or when one of them starts touching invoices.
This is where a fixed-price build usually wins, and it is the most common shape of business we work with. The tell is that your workflows have stopped being independent: a lead has to reach a CRM, a scheduler, a text message, and a report, and the state has to stay consistent across all four. That is exactly the scope where the DIY hours stop being a weekend and start being a quarter, and where a monthly retainer is buying availability you do not need because the list of things to build is finite. Start with one automation at $2,000 to $5,000, see how it lands, then decide about the rest.
Hire, and automate the part around the person. Almost every role has a rule-based crust: the confirmations, the reminders, the data entry, the status updates. Peel that off into a build so the person you hire spends their day on the part that actually needed a human. This combination is usually cheaper and better than either option alone.
Honestly assess whether the change is continuous or whether it just feels that way. If your processes genuinely rewrite themselves every month, a retainer team or an internal ops hire is the right structure and we are the wrong vendor. If instead you have a large but finite backlog, a multi-tool build at $8,000 to $25,000 handles more of it in one pass than a year of retainer months typically does, and you keep the asset.
DIY is off the table regardless of your workflow count. Raw API work, authentication, pagination, and error handling are engineering, and no-code tools stop helping at exactly that boundary. Your real choice is between a build and a person doing it by hand, and the answer usually comes down to how many records per month go through it.
Fix before you build. Take the free 30-minute audit call and find out whether what you have is repairable, because rebuilding something that only needed a credential rotated is a bad use of anyone's money. We say that out loud more often than you would expect from a company that sells builds.
Every figure below is a modeled scenario, not a measured result from a client. The arithmetic is shown so you can swap in your own numbers.
DIY (Zapier / Make)
The subscription is the small part. Model 40 to 60 hours of your time to design, build, test, and maintain one non-trivial workflow. Valued at $50 per hour, that is $2,000 to $3,000, and the maintenance share of it recurs. Platform fees sit on top and rise with task volume.
Hire: part-time admin
Modeled from our hiring breakdown: $18 per hour, 20 hours per week, 52 weeks is $18,720, plus roughly $1,400 in payroll taxes and $2,000 in basic benefits. Every year, and it rises with wages.
Hire: full-time employee
Year one on a $40,000 base salary, once you add employer payroll taxes near 7.65 percent, benefits at 25 to 30 percent of salary, and $3,000 to $8,000 of recruiting. Excludes management overhead and the ramp period. Source: our own published model, linked below.
Retainer agency
The number is whatever you negotiated. The structural cost is that it never ends and it does not accrue to an asset unless your contract says the workflows are yours in your accounts. Model it as a subscription you cannot cancel without losing what it runs.
Subscription SaaS point tool
Single-purpose products like missed-call text-back or an AI receptionist. Over three years that is $7,200 to $25,200 at those rates. It does exactly what the vendor built, no more, and you never own it.
Aplos AI
$2,000 to $5,000 for a single automation, $8,000 to $25,000 for a multi-tool build, fixed and quoted in writing before work starts. $0 per month to Aplos afterward. You pay only vendor tooling, typically $20 to $150 per month direct, much of which you already pay today.
The employment figures above are modeled assumptions taken from the cost model we published in automation vs hiring. They are not measured results and your actual numbers will differ by state, benefits plan, and hiring channel. Software prices change frequently, so treat the SaaS and platform figures as illustrative of how those pricing models behave rather than as current vendor quotes. The Aplos ranges are our published fixed prices and those are exact.
Yes, and in a real set of situations you should. If you are a solo operator with one straightforward workflow, a contact form that needs to land in a spreadsheet and fire a confirmation email, Zapier or Make will do it in an afternoon for almost nothing. Paying anyone four figures to build that would be a waste of your money. DIY stops making sense when the workflow branches, when it has to handle errors without silently dropping records, when it touches software with no prebuilt connector, and when task volume pushes you into the expensive tiers. At that point you are not saving money, you are spending your own hours on unpaid platform maintenance.
When the work needs judgment, when it changes shape week to week, or when the volume is genuinely enough to keep a person busy all day. A coordinator who negotiates with vendors, calms an upset customer, and rebuilds a schedule on the fly is doing work no workflow can express. If you have enough of that to fill a role, hire, and hire well. The mistake we see is hiring a person to run a checklist that never varies, then paying that salary every year forever because nobody ever turned the checklist into a system.
You can, and there are good builders there. Two risks are worth naming. The first is quality variance: you often cannot tell a careful builder from a fast one until the automation has been running for a month. The second is key-person risk. If the workflow lives in someone else's head and someone else's account, a 2am failure means waiting on a reply from another time zone. We build in your accounts, hand over written documentation and a Loom walkthrough, and leave you able to maintain it or hire anyone else to.
Nothing, if you have a continuous stream of new work. Retainers earn their keep at companies shipping process changes every month, where a standing team that already knows your stack beats re-onboarding someone new. The problem is the common case: the retainer was sold to build three automations, the three automations were finished in month two, and month seven is still billing. Ask any retainer agency two questions before signing. Whose accounts does the automation live in, and what happens to it the month you stop paying. If the answer to the first is theirs, you are renting.
We can migrate it, rebuild it more robustly, or leave it exactly where it is. Plenty of Zaps are fine and should not be touched. The ones worth rebuilding are usually the ones that fail silently, the ones that have grown into a chain of five Zaps taped together, and the ones where task volume has made the bill uncomfortable. The free 30-minute audit call sorts yours into those buckets, and we will say so plainly when the answer is leave it alone.
Never. Fixed price, quoted in writing before work starts, and $0 per month to Aplos after delivery. You still pay the tool vendors directly, typically $20 to $150 per month depending on what the workflow runs on, and in most cases you are already paying that today.
You are not stranded, because the workflow runs in your accounts with documentation, so your own team or any competent builder can open it. Most breakage is also not mysterious: a vendor changed an API, a credential expired, someone renamed a field in your CRM. The handoff doc covers the common failures and where to look. If you would rather we make the fix, we quote it as a small fixed-price piece of work. There is no support contract you have to carry all year for the right to ask.
Most agencies bill monthly and keep the automation in their own accounts, which makes leaving expensive by design. We quote a fixed price up front, build inside your accounts, hand over documentation and a walkthrough, and stop billing. Delivery is 1 to 2 weeks for a single automation and 2 to 3 weeks for a multi-tool build, with a 30-day window where we fix anything that is not working as scoped.
There is no universal number, but the pattern is consistent. One simple workflow: build it yourself. Two or three workflows that share data, or one workflow with real branching and error handling: the DIY hours start to outweigh the build price. Five or more, or anything touching software without a prebuilt connector, and DIY has usually stopped being a cost decision and become an availability decision, because the only person who could build it is the person running the company.
Modeled, and labeled as such everywhere they appear. The employment numbers use the assumptions we published in our automation versus hiring breakdown: a $40,000 base salary, employer payroll taxes around 7.65 percent, benefits at roughly 25 to 30 percent of salary, and $3,000 to $8,000 of first-year recruiting cost, which puts year one between $55,000 and $65,000. Software pricing moves constantly, so we describe how each pricing model behaves at volume instead of quoting a vendor sticker price that will be stale. The Aplos ranges are our own published prices, and those are exact.
Software Comparisons
We write honest breakdowns of the software our clients actually run. No affiliate bias, just the real trade-offs.
Free 30-minute audit. We'll tell you what's worth automating and what it costs. No obligation.
Get Your Free Audit →