We discussed implementing paid features here earlier in the year, and soon we’ll be launching the first of those features on self-hosted (and Cloud*). In the spirit of openness, we want to give you a sneak peek of what we have planned.
On self-hosted, we will launch two paid plans with the following features:
Team: sharing of workflows and credentials
Enterprise: LDAP and log streaming (plus all team features)
Self-hosted pricing will scale with usage. This is because usage is related to value, and we think it’s fair that people for whom n8n is more useful pay more, and people who only use it a little pay less.
More specifically, pricing will scale with the number of active workflows** you have. We chose this because:
It’s easier to predict how many you’ll need than with executions
Running out of them doesn’t impact anything you have running in production (unlike executions)
They’re related to value (one active workflow roughly equals one problem you’re solving)
The team plan will start at $50/month for 10 active workflows and unlimited users, with the ability to buy more active workflows if you need them.
For those of you who can’t afford the paid features, we are sorry, but the open-source community version will always be available for free, and you’ll be able to use it as you’ve always done till today.
Obviously, we’d love to put everything in the free plan or make the paid plans very cheap. But we have to make sure that we put n8n on a sustainable path — that way, we can continue to improve the free plan. If you’ve been following n8n’s development, you’ll have seen that we have made many improvements recently, and there are more to come.
I appreciate the $25 for 5 workflows tier you added under the amount in this post, but it just feels like the ratio of workflows to $ seems off. Especially compared to your cloud pricing which comes with sharing for $50/month and up to 100 active workflows (albeit with 15k executions, presumably per month). We’ve been running n8n for nearly a year and have ~11k executions, but with 7-soon-to-be-8 triggers, and a number of workflows currently on Zapier just begging to be moved to n8n as soon as we have time. Honestly I’d much rather pay you than pay them, but I don’t want to stress about keeping the number of triggers constrained or explain a ballooning budget when it was originally sold as a cost and efficiency savings.
This also seems like it would incentivize a poor practice of creating a single timed trigger and then then branching off what would normally be a large number of workflows to count as a single active workflow.
(Also, where do workflows that are meant to be triggered by n8n itself fall into this paradigm? Do they count as an active workflow or a manual/test workflow? Hopefully the latter in order to not discourage healthy design practices.)
Again I repeat/rephrase: you deserve to be paid for your work, I want to see n8n be a success, but $5/workflow/month seems like a poor metric, especially when any one single workflow may be saving less than 15 minutes work/month (but I’m lazy and would prefer automating than remembering to spend those 15 minutes).
Thanks a lot for your feedback. That is all great to know!
Totally agree, the ratio between cloud and self-hosted is totally off! As you pointed out correctly, does it have to do with us limiting those plans around executions rather than workflows. Over time we will harmonize the plans.
But what do you mean by the ratio of $25 for 5 workflows is also off? That is literally $5 for a workflow?!?! If a workflow does not save at least 10x that amount of money/time a month I would argue it is not worth having in the first place. I also assume that the most active n8n workflows will be saving much more than 10x that amount per month. And totally honestly, if people do not receive much more value than the equivalent of probably a Starbucks visit with a friend ($25), we have a really huge problem at n8n and should probably close down.
About your 15-minute example of being lazy. It for sure does not give the above-mentioned 10x but it really depends on how much you value your time and/or what you are paid. Esp. also what other things you could be doing with that time instead. But for now, just staying on the salary side of things and doing the math. As long as somebody earns more than $20/hour it is for sure still worth it. Ff somebody earns $60/hour it is 3x, plus knowing that things do not get forgotten, they are done in a reliable fashion and taking the mental load off.
And just to make it clear. The version (the community version) you and everybody else are running right now, is still free and unrestricted and will stay that way. Meaning as many workflows, triggers, users, and executions as you want. We will obviously also keep on releasing new versions with a lot of improvements moving forward. It is really just if you upgrade to the paid version which includes additional features (like collaboration). Only that version is restricted by active workflows. So it will for sure always stay very cost-efficient and nothing has to change unless you actively change to that version.
Regarding what metric we are charging for. There was actually incredibly much internal discussion around that, and a lot of time and thinking went into it. Should it be executions, workflows, active workflows, users, operations (similar to Zapier/Make), … . There are for sure reasons for each of them and also very good reasons against them. If we take executions, for example. There seems to be a very clear relationship between the number of executions and the value people receive. On the other hand, executions are not always equal. Somebody may use n8n to process logs and so needs multiple millions of them in a 3-node workflow, somebody else, on the other hand, processes some high-value enterprise deals with 50 steps and just requires 10 every month. Like that n8n becomes suddenly incredibly cheap (way too cheap) for organizations with one use case but unpayably expensive for organizations with another one. Meaning suddenly n8n could not be used anymore for those. (but also think generally still think 5 Euros for each is still very cheap)
So why else active workflows? It has some other very great advantages:
It correlates very clearly with value. No matter what people do, it is one “task” which provides value
It is very easy for users to predict how many they will need. If you have executions/operations it is often very hard to do so, esp. with operations. It often requires a lot of math and thinking to figure that out beforehand.
If the plan limits are reached nothing bad happens (no interruptions, which is important for us) to your active workflows. With executions/operations they suddenly stop working. Literally, just one user has to configure one of their workflows wrong, which then triggers a lot and all others will then stop working once the plan limit got reached. With active workflows, on the other hand, is the worst thing that can happen that a user can not activate a new workflow.
You are also right that the way we charge will potentially cause people to make bad workflow-building decisions esp. if they want to avoid paying more. People always become very creative if they want to save money. That, however, does not work out the most time well for either side. They then maybe avoided paying us $25 more per month but spent hours every month supporting badly built workflows that become on top unreliable. That is sadly the case no matter what metric we use.
What do you mean with “Also, where do workflows that are meant to be triggered by n8n itself fall into this paradigm”?
If we would not charge for the ones that are meant to be triggered by n8n (which sounds like a valid definition of an active workflow) we would not have anything we would charge for. So those are for sure included. What is not included, are not active workflows that get executed manually.
i beg you do not charge by iterations , i will shup up shop and find something else, its the whole reason I’ve moved away from services like that to n8n, as you say sometimes i will do an API look up that has 1000 results that then need an API call for each result, and that gets very costly when automating and means i end up stopping and goes backwards in terms of maturity. Im sure this is a main reason why people are moving away from those type of charged services.
I’m not a company, I’m a lone wolf doing things for my blog, but if i get to the point of enterprise level, any service that is done by iterations is immediately struck off my tender RFP list due to previous experience.
The community version also has unlimited users but we did not list it because we thought it could be confusing as people may think these users could collaborate. We probably have to rethink that.
Generally, is as mentioned above, nothing gonna change with the community version. You can still create as many users as you want, they are just not able to collaborate (share credentials or workflows, which they never could).
Sorry, I didn’t mean the math was wrong, just $5/ea feels wrong. $60/year in perpetuity for an action that I had to build and troubleshoot myself (I know you did a lot of the heavy lifting, but just going with my gut) and that I’m hosting on my own hardware.
I sat down and looked at what I’m paying on Zapier again, and the comparison isn’t as bad as I initially thought, simply because the community edition for n8n remains free and to get the collaboration features for Zapier you need the $300/month tier there (which admittedly would power about 250 workflows for us, but the floor is high).
But honestly, what this means is that we remain on the community edition for the foreseeable future instead of moving to a paid solution, or if we do, likely fragment our workflows between ones that require collaboration vs ones that do not. (I was kind of hoping for an excuse to move to a low-cost paid plan, but this isn’t it, yet.)
My question about whether workflows that are triggered by n8n itself is better phrased as: “what about workflows that are only meant to be triggered by the Execute Workflow node”. These kinds of workflows tend to not be independent workflows, but instead reused snips of workflows that may be used by multiple other workflows (that themselves had their own triggers).
Workflow 1 is triggered by a webhook, does a few things, then triggers workflow 3, waits for a response, and then does some other things.
Workflow 2 is triggered by a Google Sheet update, does a few things, triggers workflow 3, waits for a response, then does some other things.
Workflow 3 takes the data inbound to it by workflows 1 or 2, processes it in a standardized way and sends an email.
Workflow 4 is the workflow that gets triggered if any of the above workflows have an error, using n8n’s inbuilt error settings.
I assume this is a “2 active workflows” scenario, but wanted to be sure.
I’m using both n8n.cloud, self hosted (via elest.io) and the maOS Desktop app (for local development/testing).
I’m just wondering why the self hosted team plan starts at $ 25/mo for 5 active workflows, whereas the cloud start plan is $ 20/mo for 20 active workflows. In other words, in the self hosted plan, you pay $ 5 per active workflow plus the additional cost of self-hosting, whereas for cloud you (only) pay $ 1 per active workflow.
thank you @jan for sharing some background information and what is happening at n8n. I followed the whole discussion and I was wondering, if you considered before posting, that people might break down the costs per workflow. If I look at the self hosted version, I’d pay not only the costs per workflow but also my hosting costs. Why would you charge more for that one than your own cloud version?
Moreover, did you consider having premium apps/APIs like Zapier? I don’t like the idea at all of having premium apps but I’d rather pay 3 euros more for an API that I need than having my own hosting and paying even more per workflow.
Sorry for the late reply, but was a few days out on vacation, and I knew writing this answer will take longer.
@EricSGS Maybe there is also a big misconception about n8n. When I started to build it, my goal was never to build the cheapest solution (or, to the very extreme the “free solution” as it is maybe seen by some), my goal was to build the most powerful and best solution! I did not literally spend thousands of hours to save other people or me $50 per month on a Zapier/Make/Workato/tray… bill (it is for sure nice, but honestly nothing that excites me), rather did I do it because things were simply not possible in other solutions. So there are for sure some use cases (for example, very many small workflows that do not run often) where Zapier and others turn out to be cheaper, maybe even much cheaper. That is OK, and if people move for that reason to another solution, I can live with that, but obviously would prefer if they would not. What I, however, really hate, is if people move because something can not be done in n8n but in the other solutions.
Currently, we serve for sure also a lot of users that are mainly looking for a free solution. n8n is working very well for those people, and it should. What is, however, important to understand, is that those are not the people we target with our proprietary features. The right version for them is the free community version. The Team plan is rather for organizations that run important production workloads and have multiple people that want to collaborate. The Enterprise plan is probably self-explanatory.
I can understand that not everybody wants to pay or wants to pay less. That is totally OK, but those people should not concentrate on, or be sad about the very few features that are proprietary and they can not use for free. They should be rather happy about them as those few features are the ones that will allow them to keep on using n8n for free. Because with them we can become a sustainable company. Only if we are able to build a profitable company, will n8n stay around and not just improve the product (free & paid), we can also keep on giving support, improve docs,… and everything else.
About the point of having to build, troubleshoot and host workflows. I honestly fail to see how paying for n8n is different to any other software product out there. When people pay for Photoshop, they do not do that because it automatically edits the images, nor do people pay for Word because it writes a letter for them, they pay for the convenience the tools offer. Same with n8n. Nothing that can not be done with n8n can not also be done by a custom script. But writing, maintaining, hosting, … those scripts are also not fun and take a lot of time and resources (if people are even able to do that). That is where n8n offers the convenience and what they pay for, at least if they choose to. If $5 per month per workflow is too much or too little really depends on the use case. But as mentioned above, do I still believe that if it is not worth $5 per month, maybe it is not worth having that workflow.
Sure running multiple instances of n8n, one paid and one free is one of the multiple things that can be done to circumvent having to pay or having to pay too much. But as also mentioned above, will that likely simply improve your workload. If you see your time as free, and possible downtime or other problems are nothing you worry about, that could make sense for you personally.
With the trigger nodes. The “Execute Workflow Trigger” node does not count.
It has to do with 1. not having updated the cloud plans yet (but we will) and 2. the cloud plans having an additional limit on the number of executions.
Welcome to the community @Pascal
Seems to be the same point as @dickhoning made, and so did answer above.
Regarding paying per app. Yes thought about it before, but honestly do not like it. Do not see any clear advantage to the per-workflow pricing, which I think is much simpler. The last thing I want is that people are excited to build a new workflow and then get asked to pay more every time they add a new node. Also want to point out that other companies do not seem to charge only for that, they rather charge for steps + premium apps (which are only available in higher plans). That is something I could potentially also imagine in the future, at least for enterprise apps.
@jfcaringi Thanks, and also happy new year! Honestly do not see a problem there. Everything you will need for your personal and hobby stuff you should get in the free community edition. After all, will you be working on them alone, and the collaboration feature is not required. Also, other features like LDAP, audit logging, … are things that are not required for personal projects.
@jan Oops, I did not realise the maximum executions on your cloud solution, as this (currently) is no limitation to me. What I really like about your cloud solution, is the easy of setting up credentials to the likes of Microsoft. This alone makes it worthwhile also having a cloud version. You may want to mention this in your cloud pricing page as well … if it’s not already there
I understand that it’s impossible to have a price model for every possible use case, and appreciate that you’re trying your best to suit as many as possible.
And if I may recap; you’re trying to build the best workflow tool, and a sustainable business, keeping it free for individuals and small businesses, and offer paid (value added) features for workgroups and organisations. And if that’s the case; you’re doing a damn good job!
Also is there any limitation in the self-hosted community plan for the Number of active workflows? or number of active Running workflows?
Also in the first post of this thread, I noticed you maintained the cost of the active workflows to 50$ per 10 active workflows. so to clarify that, I need to know when you are talking about active workflows, you mean the active running workflows or just the number of the workflows that the status of them is active ( even no instance of the them is running )
Also how do you cost the workflows that does not active ( for examples the ones that trigger manually )