Make vs Zapier for New Zealand businesses.

How do Make and Zapier compare in October 2026?

Both are hosted, no-code automation platforms, and both moved to new billing in the past year. Make now bills in credits, converted 1:1 from operations. Zapier moved AI steps, code and its SDK onto task-based pricing.

Where is each one stronger?

Make is stronger on cost at volume and on complex logic. Routers that split a scenario into branches cost nothing, and its visual editor shows exactly how data bundles flow between modules. It also exposes your scenarios to ChatGPT, Claude and Cursor through its MCP Server on every plan. See our Make guide.

Zapier is stronger on breadth and simplicity. It says it offers integrations for more than 9,000 apps, which matters when a niche industry tool only has a Zapier connector. Because filters and formatting are free, a well-built Zap spends tasks only on the steps that do real work. See our Zapier guide.

How do you test Make and Zapier on the same workflow?

Build one workflow in both: a new Shopify order adds a row to a Google Sheet, an AI step writes a short care note for the products bought, and a Gmail draft is created for a person to check and send.

1. Build it and time it. Record build time and whether the person who will own it understood each step.
2. Probe the meter. Run 20 orders with one product and 20 with six. Read the usage logs and record credits and tasks per order. This is where the two tools diverge.
3. Cause an outage. Disconnect the Gmail account for an hour while orders arrive. Check who is alerted, whether failed runs can be replayed, and whether any order was silently skipped.

What do Make and Zapier cost over a year?

Worked example: 800 orders a month, prices as listed on 19 October 2026. Assumptions: the Make trigger uses one credit per order, Zapier uses a Standard-model AI step (one task), and both write one care note per order.

• Make: trigger, Sheets, AI and Gmail = 4 credits an order, or 3,200 a month. That fits Core’s 10,000 credits at $9 a month, or $108 a year. Using your own OpenAI or Claude connection adds token charges paid to that provider.
• Zapier: Sheets, AI and Gmail = 3 tasks an order, or 2,400 a month. That is beyond Professional’s 750-task starting tier; Team starts at 2,000 tasks for US$69 a month on yearly billing, or US$828 a year.

Now write one note per product, with an average of three products an order. In Make, an iterator (1 credit) plus AI and Sheets per product (6 credits) plus the trigger and Gmail make 9 credits, or 7,200 a month, still inside 10,000. In Zapier, looping is free but the AI and Sheets steps run per product, giving 7 tasks an order, or 5,600 a month. Both bills grow, but Zapier’s grows from a higher base.

Costs that appear later:

• Running out. Make’s scenarios stop until credits are added; auto-purchase is available on Core, Pro and Teams, and unused credits expire at the end of the term. Zapier keeps running at a higher pay-per-task rate shown only in your billing settings.
• Builder AI. Make’s Maia uses credits per message after the first 70, which are free for 30 days on paid plans; Make’s own examples show about 140 to 160 credits a message.
• Currency. Zapier bills in USD; Make’s “$” is unlabelled. Budget for exchange-rate movement.
• Switching. Neither imports the other’s workflows. A written specification makes a rebuild faster; see writing a workflow specification.

Our AI cost calculator helps you run the same arithmetic on your own volumes.

Which should you choose?

Choose Make if

• Volumes are high or rising and cost per run matters.
• Scenarios branch a lot, so free routers save real money.
• Someone on the team is comfortable thinking in bundles and iterators.
• You would rather a scenario stop at the limit than run up an unplanned bill.

Choose Zapier if

• Non-technical staff will build and change the automations.
• You need a connector only Zapier offers.
• Most steps are filtering and formatting, which Zapier does not charge for.
• You want SAML single sign-on and shared workspaces for up to 25 users on Team.

Where do both fall short?

• Hidden multipliers. Loops, polling and per-item steps multiply usage in both. Re-check usage logs after every change.
• AI steps without checks. A care note or reply that is wrong one time in twenty still reaches customers if nobody reviews it.
• Personal data in run history. Both keep logs of the data that passed through. Under the Privacy Act 2020, IPP 9 says not to keep personal information longer than needed.
• Offshore processing. Make’s EU zone does not guarantee EU AI processing, and neither tool offers a New Zealand region. If that matters, compare n8n versus Make for the self-hosted route.

For workflows that touch money or customers, our AI workflow automation service builds in review steps and monitoring from the start.

Written by the Levy Street team in Wellington. Last updated 19 October 2026.