Pipedream vs n8n vs Make: the Post-Workday Migration Decision (July 2026)

The renewal notice lands, you glance at the "Pipedream has joined Workday" banner now sitting at the top of the pricing page (per Pipedream's site, fetched 2026-07-10), and the question writes itself: do I re-up, or is n8n or Make cheaper and close enough to move to? That is the real decision, and it has a different answer depending on which two or three connectors your automations actually depend on.

Here is the short version before the reasoning.

Stay on Pipedream if your workflows are code steps in Node or Python on managed infrastructure, or if you touch Workday, HR, or finance systems. Nothing else in this comparison matches that combination, and the acquisition points the roadmap toward it.
Move to self-hosted n8n if you are a solo builder or small team with Docker capacity, want unlimited executions for the price of a small VPS, and plan to drive automation from Claude or ChatGPT through MCP.
Move to Make if you want the lowest cash cost with a visual canvas, no servers to run, and a connector library that has quietly caught up to Pipedream's.

This is the acquisition-driven version of the decision. If your question is really about raw code-level control between the three, our May comparison covers that frame and this one does not repeat it.

What Workday actually changed (and what it did not)

Workday announced a definitive agreement to buy Pipedream on November 19, 2025, at Workday Rising EMEA in Barcelona (Workday newsroom). The release put the close inside Workday's fiscal Q4, ending January 31, 2026, and the "Pipedream has joined Workday" banner now on the site says it went through.

The strategic language is worth reading literally, because it tells you where the product is headed. Workday's framing is that Pipedream's connectors let its AI agents move "beyond insights" and take action across enterprise apps like Asana, HubSpot, Jira, Recurly, and Slack, with Workday's HR and finance data at the center (Workday newsroom). Gabe Monroy, Workday's SVP of Platform, put it as "not just insights, but action." Pipedream's side of the announcement is calmer: founder Tod Sacerdoti wrote that the 3,000+ connectors and the builder community come along, and that the teams "will continue to operate independently and conduct business as usual" (Pipedream blog).

Read together, that is two things for you. The connector library and the open builder community are not going away in the near term. And the investment gravity now runs toward enterprise, Workday-adjacent use cases, not the solo developer running three cron jobs. Neither is a reason to panic. Both are reasons to know exactly what you depend on before your next renewal.

The pricing, side by side (fetched July 10, 2026)

The one thing that makes this comparison hard is that none of the three bill in the same unit. Pipedream charges credits by compute time. Make charges a credit per module action. n8n charges per workflow execution and does not care how many steps are inside it. So the table below lines up plan names and headline prices, but read the "billing unit" row before you compare the numbers.

Plan tier (fetched 2026-07-10)	Pipedream	n8n	Make
Free tier	300 credits/mo, 3 active workflows	Self-hosted Community Edition: unlimited executions, $0	1,000 credits/mo, 15-min minimum run interval
Entry paid	Basic $29/mo (2,000 credits/mo, 10 active workflows)	Cloud Starter 20 EUR/mo billed annually (2,500 executions)	Core $9/mo (10,000 credits/mo)
Next tier	Higher tiers lift the credit cap, usage charges apply	Cloud Pro 50 EUR/mo billed annually (10,000 executions)	Pro $16/mo, Teams $29/mo (10,000 credits/mo)
Self-host	No	Yes, free, unlimited	No
Billing unit	Credit = 30s compute at 256MB	Workflow execution, any number of steps	Credit = one module action
Connectors	3,000+	Several hundred native, plus HTTP and custom code	3,000+ standard apps

Sources, all fetched 2026-07-10: Pipedream pricing and Pipedream credit docs; n8n pricing; Make pricing. The n8n cloud prices are billed in euros (roughly $22 and $55, exchange-rate dependent, so treat those as approximate). Pipedream's higher paid tiers are uncapped on credits but add usage charges above the included amount (Pipedream docs).

Two numbers deserve a callout. Make's Core plan at $9/mo (Make pricing) is the cheapest paid entry in the category. And the old "Pipedream has far more connectors" line is mostly gone: Make now advertises 3,000+ standard apps (Make pricing), matching Pipedream's 3,000+ (Workday newsroom). Connector count is no longer the reason to stay on Pipedream. Code steps and the Workday ecosystem are.

Pipedream: stay if the connectors and code steps are load-bearing

Pipedream's edge was never the visual builder. It is that a step can be a real function in Node, Python, Go, or Bash on managed infrastructure, sitting next to 3,000+ connectors and 10,000+ pre-built tools (Workday newsroom). If your automations lean on that (custom logic between API calls, no server to run, real language runtimes) then the acquisition does not change the value on the page today.

The candid con is the trajectory, not the product. The billing unit is compute time, which is fine for short workflows and punishing for long-running or memory-heavy ones, since credits scale with both duration and the memory you allocate (Pipedream credit docs). And the roadmap money is now pointed at Workday-scale enterprise buyers. A solo developer on the free plan's 300 credits and 3 active workflows is not the customer Workday paid for. That does not mean the free tier disappears tomorrow. It means the pressure over the next few renewals runs toward enterprise packaging, and you want to know your exit cost before that shows up in your bill.

Who should not move: a team using Pipedream to wire Workday, an HRIS, or ERP connectors into agentic workflows. That is the one place where the acquisition is a tailwind, and no other tool here has comparable depth in that ecosystem.

n8n: switch if you self-host or build with MCP

n8n is the node-graph automation tool you can run yourself. The Community Edition is free, self-hosted, and does not meter executions, so your cost is whatever the box costs. On a small VPS that is a few dollars a month for effectively unlimited runs (n8n pricing confirms cloud plans meter executions while self-host does not).

The reason to look at n8n right now is MCP. Since late April 2026, n8n's built-in MCP server can build and update workflows from a prompt, not just run existing ones, and it ships in every edition including the free Community Edition (n8n blog, April 29, 2026). You point Claude, ChatGPT, Cursor, or Windsurf at your instance, describe the workflow, and the client builds it, validates it, and fixes itself if a node errors. That is a genuinely different build loop from clicking modules on a canvas, and it is first-party rather than a third-party bridge you have to babysit. Pipedream does not yet ship an equivalent native MCP server; coverage there leans on Composio or other third-party MCP integrations.

Here is the self-host setup on a small box. This runs n8n in Docker and persists its data.

# On a fresh Ubuntu VPS (2 GB RAM is plenty to start)
docker volume create n8n_data

docker run -d --restart unless-stopped --name n8n \
  -p 5678:5678 \
  -e N8N_HOST="<YOUR_DOMAIN_OR_IP>" \
  -e WEBHOOK_URL="https://<YOUR_DOMAIN>/" \
  -e GENERIC_TIMEZONE="America/New_York" \
  -v n8n_data:/home/node/.n8n \
  docker.n8n.io/n8nio/n8n

# n8n is now on port 5678. Put a reverse proxy (Caddy or nginx) in front for TLS.

The config shape follows n8n's self-hosting docs. Once it is up, wiring an AI client to the MCP server is a small block in the client's config:

{
  "mcpServers": {
    "n8n": {
      "url": "https://<YOUR_DOMAIN>/mcp",
      "headers": { "Authorization": "Bearer <YOUR_N8N_API_KEY>" }
    }
  }
}

The drawback is the one every self-hoster signs up for: you own uptime, backups, and version upgrades. Its native integration library is also smaller than Pipedream's or Make's. The HTTP Request node and custom-code nodes cover most APIs that lack a prebuilt connector, but "covered by HTTP plus code" is more work than "there is a native module for that." If nobody on the team wants to run a server, n8n Cloud exists at 20 EUR/mo and up (n8n pricing), though at that point Make usually wins on price.

Make: switch if you want the lowest cost and a visual canvas

Make is the visual-first option, and it is the value pick. You build on a canvas of modules, branch with a router, and drop into an HTTP module or a code app only at the edges. The Free plan gives 1,000 credits a month with a 15-minute minimum interval between runs, and Core at $9/mo lifts that for 10,000 credits (Make pricing). For a team that wants a working automation this afternoon with nothing to host, that is the fastest and cheapest path in this comparison.

Make also closed the two gaps that used to send developers to Pipedream. The app library is now 3,000+ standard apps (Make pricing), on par with Pipedream. And Make ships an MCP server of its own for connecting an AI client to your scenarios (Make MCP), so the agentic story is no longer Pipedream-only.

Where it bites is the billing unit. Make charges a credit per module action, so a scenario that touches eight modules per run burns eight credits per run. At low volume that is a non-issue and the $9 Core tier swallows it (Make pricing). Push a high-fan-out scenario at scale and the per-action meter adds up faster than n8n's per-execution model, where the same eight-module run is one execution. The other candid limit: Make keeps you furthest from a real runtime. If your automation is mostly custom code between API calls, you will fight the canvas, and Pipedream or n8n fits better.

Pondero earns a commission if you start Make through our link. It is still the value pick here on the pricing that Make publishes, and the candid cons above are the reason we would not hand it to a code-heavy team.

Which one, by who you are

The single-line verdicts, by the four Pipedream users most likely to be reading this.

You are	Pick	Why
Solo developer, 5 or fewer workflows, no compliance rules	n8n self-hosted	A $6/mo droplet beats every metered plan and runs unlimited executions.
Small team, visual-first, budget under $20/mo	Make Core or Pro	Lowest cash cost, nothing to host, 3,000+ apps.
Team using Pipedream for Workday, HRIS, or ERP connectors	Stay on Pipedream	No comparable ecosystem depth, and the roadmap now favors you.
AI-agent builder driving automation from Claude or ChatGPT	n8n, Make close behind	n8n's native MCP server builds workflows from a prompt in every edition; Make's MCP server is the runner-up.

The one that flips: a solo developer who refuses to run a server should skip self-hosted n8n and land on Make Core, not n8n Cloud, because at that point Make is cheaper for the same "someone else hosts it" deal.

What migrating actually costs

Moving off Pipedream is not a one-click export. You audit active workflows, list the connector types each one uses, check n8n or Make for a native equivalent, rebuild triggers, and re-enter every credential by hand. The effort scales with connector variety and custom-code volume, not with workflow count alone.

A worked estimate, so the number is concrete. Example: a team runs 12 active Pipedream workflows. Nine are webhook-in, transform, HTTP-out with one or two connectors each. Three are heavier, with custom Node steps and four to five connectors apiece. A connector audit turns up 14 distinct apps, 11 with native n8n or Make equivalents and 3 that need an HTTP node. Budget 30 to 60 minutes to rebuild and test each simple workflow, and 2 to 4 hours for each complex one. That lands around 12 to 20 hours, two to three focused days for one engineer plus credential setup. Under 10 workflows that are mostly HTTP and webhook, a weekend is realistic. Past 15 to 20 with heavy custom code, plan for one to two engineer-weeks.

You can build the connector inventory quickly. If you have your workflow definitions exported as JSON, tally the app connectors like this:

# Example: count distinct connector apps across exported Pipedream workflows
cat exported-workflows/*.json \
  | grep -oE '"app": *"[a-z0-9_]+"' \
  | sort | uniq -c | sort -rn

#    17 "app": "http"
#     6 "app": "slack"
#     4 "app": "google_sheets"
#     3 "app": "postgresql"
#     2 "app": "stripe"

That tally is your migration map. Every app with a native target on the destination is a fast rebuild; every one that only maps to an HTTP call is where the hours go.

One more reference point, framed as the vendor documents it. n8n's own example of the MCP build loop: the input is a plain-English request, "create a workflow that once a day at 7am emails me today's forecast using my Gmail account," and the expected output is a validated, ready-to-run workflow built inside the instance in a few minutes (n8n blog, April 29, 2026). For a team recreating a stack of simple scheduled workflows, that prompt-to-workflow path is the fastest rebuild route on offer.

The migration checklist

If you have decided to move, this is the order that keeps a cutover from breaking on a Monday.

Export and list. Pull your active workflow list from Pipedream and record each one's trigger, connectors, and any custom code. The grep tally above turns exports into a connector map.
Audit against the destination catalog. Check each connector against n8n's integration catalog or Make's app library. Flag the ones with no native module; those become HTTP calls and cost the most time.
Trial the destination. Spin up a free n8n Community Edition instance, or start a Make Free workspace, and rebuild your two most representative workflows before committing.
Self-hosting? Stand up the box. A DigitalOcean droplet runs n8n Community Edition comfortably for a small team: a 1 GB box is $6/mo and a 2 GB box is $12/mo (droplet pricing). Cheap, unmetered executions, and you own the data. Use the Docker setup above.
Parallel-run for a week. Keep the Pipedream workflows live while the rebuilt ones run alongside them. Compare outputs before you turn anything off.
Cut over and decommission. Once a week of parallel output matches, disable the Pipedream workflows and let the renewal lapse.

The pick

For a solo developer, n8n self-hosted on a cheap droplet is the pick: unlimited executions, a native MCP build loop, and a bill that rounds to a coffee. For a small visual-first team that wants the lowest cash cost and no server, Make Core at $9/mo (Make pricing) is the pick, and it is the value winner in the category now that its app library matches Pipedream's. For an AI-agent builder, n8n edges Make on the native MCP server, with Make a close second. And if you run Workday, HRIS, or ERP integrations through Pipedream, stay put: the acquisition is a tailwind for you and a reason to move for almost nobody else. Run the connector audit before your next renewal either way, because the cheapest migration is the one you scoped before the invoice arrived.