Blog · Pricing
Real ranges for US small businesses — what moves the number up or down, what should be in the quote, and how to spot a proposal that's about to balloon.
For US small businesses with 1–50 employees, a workflow automation project in 2026 costs between $3,000 and $60,000 up-front depending on scope, plus $80–$1,200/month for tooling and support. The single biggest cost driver isn't the technology — it's how many exceptions, approvals, and integrations the workflow has. Most well-scoped SMB projects land in the $7,000–$16,000 range and pay back in 4–9 months.
These are the ranges we quote at Kivolaro and that match what we see other reputable US/LATAM agencies charge for comparable scope. They cover workflow automation, AI-assisted automation, and custom internal tools for businesses with 1–50 employees.
| Project type | Build cost | Calendar time | Monthly (SaaS + support) | Year 1 total |
|---|---|---|---|---|
| Single workflow | $3,000 – $7,000 | 2–3 weeks | $80 – $200 | $4,000 – $9,400 |
| 2–3 connected workflows | $7,000 – $16,000 | 3–5 weeks | $150 – $400 | $8,800 – $20,800 |
| Operational system (4–7 workflows + dashboard) | $14,000 – $30,000 | 5–8 weeks | $300 – $800 | $17,600 – $39,600 |
| Custom platform / internal tool | $25,000 – $60,000 | 8–14 weeks | $400 – $1,200 | $29,800 – $74,400 |
If you want to play with these numbers against your own workflow, the budget calculator uses the same model with the multipliers explained in the next section.
Four factors do 80% of the cost variance. When an agency gives you a single number without saying which of these they're assuming, that's the #1 signal the quote is going to grow.
A linear workflow (input → transform → output, no branching) is cheap. A workflow with branching logic, exception handling, and approval steps can be 2–3× the cost even if it looks similar on the surface.
Each tool that has to talk to the automation adds roughly $500–$1,500 in build cost. The price is driven by API quality — HubSpot, Stripe, Slack are cheap; bespoke ERPs and legacy on-prem systems can be 3–5× more.
✏️ Real-world honesty: 70% of "we need AI" requests we get can be solved cheaper and more reliably with traditional automation. AI is the right tool when the input is unstructured (free-text, images, PDFs with weird layouts) — not when it's a structured form with clear rules.
The dirty secret: half the cost of "automation" projects is cleaning the data the automation reads from. If your CRM has 6,000 duplicate leads, three different formats for phone numbers, and 200 records with "TBD" in the email field, that's a separate project before the automation can even run.
✏️ Rule of thumb we use: if the source data is in good shape (consistent formats, no orphaned records, clear ownership), automation cost stays in the ranges above. If it's messy, add 20–40% for a data hygiene pass.
If a quote doesn't itemize these, push back before signing. Each of them is real work that someone has to do, and "absorbing" them invariably means cutting corners on the ones you'll need most.
⚠ Red flags in a quote: no discovery phase, no tests mentioned, no documentation line item, no post-launch period, or a flat hourly rate with no scope cap (this is how projects 3× — there's no incentive to be efficient).
Zapier and Make are great until they aren't. The transition usually happens around 3+ connected workflows or 5,000+ task executions/month, whichever comes first. The math:
| Scenario | Zapier annual cost | Custom annual cost (year 1) | Winner |
|---|---|---|---|
| 1 simple workflow, 500 tasks/mo | ~$240 (Starter) | ~$5,000 build + $1,000 SaaS | Zapier |
| 3 workflows, 5,000 tasks/mo, branching | ~$2,500 (Professional) | ~$10,000 build + $1,800 SaaS | Toss-up; custom wins by yr 2 |
| 6 workflows, 25,000 tasks/mo, exceptions + approvals | ~$7,200 (Team) + per-task overage | ~$18,000 build + $3,600 SaaS | Custom (pays back yr 1) |
Where this math breaks down: if the workflow has branching logic Zapier can't express cleanly, you'll spend more in engineering hours wrestling with Zapier than building custom would have cost. We've seen "Zapier maintenance" silently eat 8 hours/week of someone's time — at $50/h loaded, that's $20,000/year in hidden cost invisible to the Zapier bill.
If you want to run the math against your specific situation, the ROI calculator and the Zapier vs Make vs n8n comparison have the underlying numbers.
✏️ Note: these are illustrative compositions of the multipliers above — not specific client engagements. They show how a single change (one more integration, one more approval step) actually moves the price.
The faster you can answer these, the closer the quote you get will be to the final invoice. Vague answers force the agency to estimate wide (or pad), and unvague answers force them to commit to a real number.
If you can answer 5+ of these clearly, you'll get a precise quote. If you can answer 3 or fewer, ask the agency to do a paid discovery first (1–2 weeks, $1,500–$3,000 credited toward the build). Don't ask for a fixed price on a workflow nobody has mapped yet — both sides will regret it.
For a single linear workflow with 1–2 integrations, no AI, and clean source data, build cost runs $3,000–$5,000 on a custom solution or $20–$50/month on a tool like Zapier or Make. Below $3,000, the work usually doesn't include discovery, tests, or documentation — which means you'll pay the difference in maintenance within 6 months.
A single workflow: 2–3 weeks. Two to three connected workflows (the most common SMB scope): 3–5 weeks. An operational system with 4–7 workflows plus a dashboard: 5–8 weeks. A custom platform or internal tool replacing a major spreadsheet: 8–14 weeks. These are calendar weeks, not engineering weeks — they assume normal back-and-forth on questions and approvals.
If the workflow being automated takes more than 4 hours per week of someone's time, almost always yes. At a loaded cost of $40/hour, 4 hours/week is $8,300/year — a $5,000 automation pays back in 7 months and keeps paying back forever. Below 4 hours/week, do the ROI math first; sometimes the cheaper move is a better template or a checklist, not an automation.
Zapier (and Make, n8n, Pipedream) are low-code platforms that string together pre-built integrations. Custom automation is code we write that does exactly what your workflow needs without the platform's constraints. Zapier wins on speed-to-first-version and on simple cases (1–2 workflows, <5,000 tasks/month). Custom wins on complex branching, large volume, tight integrations with legacy systems, and total cost of ownership past year 2.
Three reasons. (1) Different scope assumptions about what's included — discovery, tests, documentation, support. (2) Different complexity reads on the same described workflow (one agency sees a 1-week job, another sees a 4-week job once they ask about exceptions). (3) Different markup on tooling and engineering hours. The most accurate way to compare quotes is to ask each agency to itemize discovery, build, testing, docs, and support — then compare the totals.
Yes. The tables in this article and in the budget calculator are the same ranges we quote new clients. We don't have a hidden price list. The reason most agencies don't publish is that pricing varies by scope — but the variance lives inside the ranges above, not above them.
Blog
Send us the rough scope — trigger, end state, tools involved — and we'll come back with a number that lives inside these ranges, with discovery, tests, and 30-day support included.
Tell us the problem →
