n8n and Make Automations for Business: What They Cost, How They Differ, and 5 Workflows for Israeli Small Businesses

What n8n and Make actually are

Both tools do the same basic thing. You draw a process on a canvas: something happens in one app, and a series of steps follows in others. A form is submitted, the lead goes into the CRM, a WhatsApp message goes out, a row is added to a sheet. In Make the process is called a scenario and each step is a module; in n8n it is a workflow made of nodes. Neither requires code for the common cases, and both let you drop into code or raw HTTP requests when an app has no ready-made connector.

Where they differ is in who runs them and how you pay. Make is a cloud service with a directory of over 3,000 apps. n8n is available as a hosted cloud service or as software you install on your own server, and its integrations directory lists more than 2,000 entries (its homepage cites "over 500 integrations"; the directory also counts entries its homepage does not). The count matters less than whether the four or five apps you actually use are there and maintained.

How the pricing works (checked 30 September 2026)

The billing unit is the real difference. Make charges credits: in its own words, each module action "counts as one credit", and some AI features use more. n8n charges per execution — one full run of a workflow, however many steps it has. A six-step scenario that runs 1,000 times a month costs about 6,000 Make credits, but only 1,000 n8n executions. For short, frequent workflows the two come out close; for long workflows n8n's model is usually cheaper.

• Make Free: $0, up to 1,000 credits a month, runs at most every 15 minutes, 2 active scenarios.
• Make paid plans on its pricing page: Core $9, Pro $16 and Teams $29 a month, each priced for 10,000 credits a month.
• n8n Cloud, billed annually: Starter €20 a month for 2,500 executions, Pro €50 for 10,000.
• n8n Business: €667 a month billed annually for 40,000 executions, self-hosted only.
• n8n Community Edition: free, self-hosted — you pay for the server and your own time.
• For reference, Zapier: 100 tasks a month free; Professional from $19.99 a month billed annually for 750 tasks.

Self-hosted or cloud?

Self-hosting n8n removes the per-execution bill, but it does not make automation free. You take over updates, backups, security and uptime, and a stuck server at 2am is your problem. The licence also has a limit worth reading: n8n's Sustainable Use License lets you use the software "only for your own internal business purposes or for non-commercial or personal use". Running your own business's workflows is fine; reselling n8n as a hosted service is not.

Location is the other question. n8n says the data on its cloud is hosted in the European Union. Make lets you choose a US or an EU data centre when you create your organization, and its help centre notes you cannot change it afterwards — so choose the EU region from day one if your customers' personal data will pass through it.

Workflow 1: a WhatsApp or website lead into the CRM, with a follow-up

The most common job, and usually the one that pays back first. Both tools connect to the official WhatsApp Business Cloud API: n8n has a built-in WhatsApp Business Cloud node and Make has a WhatsApp Business Cloud app it maintains itself. Meta's rules shape the design: when a customer writes to you, a 24-hour customer service window opens in which you can reply freely; outside the window you can only send an approved template, which Meta has charged per delivered message since 1 July 2025. Replies inside the window are no longer free either: from 1 October 2026 Meta charges per message for service replies too, at the same per-market rates as utility and authentication messages. Budget for every message the workflow sends.

• Trigger: a new WhatsApp message or a website form submission.
• Look the phone number up in the CRM; create the contact only if it is new.
• Tag the source and the service asked about, and assign an owner.
• Reply inside the 24-hour window with a short confirmation and the next step.
• If nobody has answered within a set time, alert the owner; after a few days, send an approved follow-up template.

Workflow 2: an invoice or receipt through an Israeli invoicing API

Israeli invoicing systems have real APIs. Morning (formerly Green Invoice) documents a REST API where you create a document with POST /documents, with a sandbox for testing. iCount has a Make app, verified by Make and maintained by iCount's developer, with a "Creates a document" action. Morning's Make app is community-built, and Make states it does not maintain it — so for Morning I prefer calling the API directly with n8n's HTTP Request node or Make's HTTP module, where nothing sits between you and the vendor.

• Trigger: a payment approved by the gateway, or a deal marked "won" in the CRM.
• Check that no document already exists for this order, so a retry never issues a second one.
• Create the receipt or tax invoice through the API, in the sandbox first.
• Save the document number back on the order, and send the PDF to the customer.
• For business customers above the Israel Invoices threshold (₪5,000 before VAT since June 2026), confirm the system requested an allocation number.

Workflow 3: a daily report in Google Sheets

Every morning, pull yesterday's orders, leads and ad spend into one sheet, and send yourself a three-line summary. It sounds trivial and it is the report most owners never get round to building. One technical limit to design around: the Google Sheets API allows 300 read and 300 write requests per minute per project, and 60 per minute per user. Write a day's rows in one batch, not one request per row, or a busy day will start returning errors.

• Trigger: a schedule, every morning at a fixed time.
• Read yesterday's numbers from the store, the CRM and the ad accounts.
• Append them to the sheet in a single batch.
• Send a short summary by email or WhatsApp, with anything unusual flagged at the top.

Workflow 4: appointment reminders

No-shows cost clinics, studios and consultants real money, and a reminder the day before is cheap to automate. Both tools have Google Calendar connectors; n8n has Calendly and Cal.com trigger nodes, and Make has a Calendly app it maintains. Because a reminder is usually sent outside WhatsApp's 24-hour window, it has to be an approved template — plan the wording once and get it approved before you build the rest.

• Trigger: a new booking, or a daily scan of tomorrow's calendar.
• Skip internal meetings and bookings without a phone number.
• Send the approved reminder template with the time, the address and a way to reschedule.
• If the customer replies that they cannot come, alert whoever manages the calendar.

Workflow 5: review requests

Happy customers rarely leave reviews unless someone asks at the right moment. The automation is simple; the judgement is in who does not get asked.

• Trigger: an order marked delivered, or a service marked complete.
• Wait a few days, so the customer has actually used what they bought.
• Skip anyone with an open complaint, a refund or a return.
• Send one short message with a direct link to your review page — once, with no reminders.

Limits, and when not to automate

Anthropic's engineers give advice about AI systems that applies just as well here: "find the simplest solution possible, and only increasing complexity when needed". An automation built on a process that changes every week will break every week. Do not automate a task you do twice a month, a decision that needs judgement each time, or a process nobody owns. And budget for maintenance: apps change their APIs, tokens expire, and a workflow that fails silently is worse than no workflow at all.

• Every workflow gets an error path that notifies a named person.
• Every write step is safe to repeat, so a retry never doubles an invoice or a message.
• The accounts and API keys are in the business's name, not the freelancer's.

Privacy: the basics under Amendment 13

Automations move personal data between systems, so privacy law applies to them. Amendment 13 to Israel's Protection of Privacy Law has been in force since 14 August 2025. It gives the Privacy Protection Authority the power to impose financial sanctions that can reach millions of shekels and requires some organizations to appoint a privacy protection officer; the Data Security Regulations of 2017 still apply alongside it. This is not legal advice, but the practical habits are the same for almost every small business:

• Write down which apps each workflow sends personal data to — every one of them is a processor.
• Send only the fields the next step needs; ID numbers and card details never go into an automation.
• Limit who can open the workflows and their run history — the history stores the data too.
• Set how long run logs are kept, and delete them on schedule.