"We believe that Pro’s price of $9.99 per month is money well spent. However, because Pro is new and will evolve in response to your feedback, we wanted to provide some payment flexibility. Our introductory set your own price subscription allows you to choose your monthly payment for the first year of Pro. This special offer expires in 4 weeks, on Wednesday, October 7th."
Interesting approach. I wonder what users will pay on average during this first year.
If they give me a full blown console where I could run lambda-like python code on top of all the existing device integrations they already have, I'd be glad to pay the full $9.99/month. But the product would have to evolve a lot versus where it is today.
I look at IFTTT and Zapier and wonder what people use them for. Don't get me wrong, I think both have a target consumer but they feel expensive for what they appear to offer. I feel like if you're investing in automation the agnostic nature of owning and operating the code is more valuable than building into a walled system that relies on Internet connectivity and a SaaS service. OpenFaaS plus some Python could get you a long way on a $5/month VPS. Or... Pennies in any of the FaaS offerings wired up with cloud native tools. Or Node-RED?
Are there any good OSS orchestration / automation frameworks out there that scratch this itch?
You basically just POST/GET url they provide with the json payload then hook it up to a notification service (whether that be their Android app, Telegram, etc.)
Came in handy when I needed to poll something on my end, which would trigger the webhook when it met certain conditions.
1. Here's a URL (custom to each user/applet). You can either GET or POST to it. No auth/headers required (the URL includes a token).
2. You can optionally submit a JSON in the format of {"value1": ..., "value2": ..., "value3": ...} and set up the applet to set those values as the notification title, body, link, or image.
Rather unfortunately, you can only specify three values in the JSON so you can only ever have three of those four properties filled in for a single notification. But you can set up multiple receiving URLs/applets such that you can send the first notification containing three of those properties and then a second notification containing the last property.
Short answer is that yes, there are other options. But they're always vastly more complex and would require actual work to use. Why bother with all that when I can just issue a GET or POST to a simple HTTP endpoint (no auth/headers required) and have IFTTT fill in the title, message, link (if the notification is tapped), and image with the values in the uploaded JSON object? The IFTTT API is truly a joy to use as a someone trying to quickly hack something together using cURL.
Re Pushbullet specifically: it no longer supports iOS but either way, it's also much more complex [0] to use.
We used Zapier at a previous startup to automate some gsuite tasks. It's really nice because it comes with monitoring and non-developers can edit/re-configure them if needed. It's expensive but headache free.
We use it to relay tweets to a discord channel webhook. The requirement was that it had to relay the tweet contents with no flair or "tweeted by" or link to the tweet, all discord bot solutions fail this requirement and i didn't want to learn how2tweeterapi.
I currently use IFTTT mostly for the webhooks feature. I have some bots on Discord posting stuff from multiple sources, and not having to host my own server to do so is nice.
I also use it to send important notifications from some of my servers to my phone. I just have to `curl` the info to IFTTT server with my API key and send the info I need.
You can do this yourself. Just ping your lambda functions as if they were webhooks.
The problem with IFTTT has always been speed for me. I had the soccer goal notifications activated once I got my hue lights years ago so that when my team scored itd flash the lights. Usually this happened five minutes after I watched it on the TV. Useless.
Hah I feel tricked. I was going to go for $1/mo and pay them $12 for a year. I don’t use ifttt but why not. The minimum is $1.99. That’s not pay what you want. I might still do it and combine what my brother or friend use it for into one account for a year, but still
A couple weeks ago, I was looking for some way to convert a Gmail email filter into an Android clock alarm (not just a custom notification) so I can be woken up if a certain email arrives. I haven't found a solution for this. I had heard of IFTTT, so I took a look at the website, but I couldn't make heads or tails of it. I signed up for a free account and I searched through the integrations available, but it didn't seem to get me what I wanted. Maybe it's just me, but it wasn't clear enough that it could solve my problem that I should invest the time to figuring out how to use it.
I really, truly don’t mind paying for software, but all of these subscription services cost obscene amounts of money. $10 per month is $120 per year. $120! I might pay that much for boxed software, but it would need to do a heck of a lot more than IFTTT, and I’d expect to get much more than one year of life out of it.
Affinity Photo, for instance, cost $60, and VMware Fusion cost $80, to name two products I purchased recently. Neither needs to be bought again after a year unless an update contains a new feature I want (and probably not even then for Affinity), and both can do a lot more than IFTTT. Neither has a server component, true, but how big a cost can that possibly be for IFTTT?
> Would you run this boxed software on your machine?
Yes. Would you rather have ice cream or get punched in the face?
> Would you maintain it?
If maintaining it means scheduling updates through my host's package manager or running docker pull, and not "hey it's thursday, time for a new configuration format", then sure.
> How much is your time worth?
Enough to never host Elasticsearch, not enough to go SaaS-only.
Any service with any level of client integration will a) version their APIs, and b) keep things sufficiently stable so that clients aren't constantly breaking.
For everything but the most high profile platforms, IFTTT expects the service provider to build a custom API made for IFTTT, and pay IFTTT to include the service in their public catalog.
I currently maintain about 280 different API integrations. I have a team that does constant maintenance. Usually 1-2 API breaking changes per day is what we see after you get past the stable big ones. Even versioned APIs kick you off the old version pretty regularly. In the upcoming months we have 3 scheduled version bumps that are mandatory as they will discontinue the old version completely. Not "use the new version for more features", like completely 100% "we are not supporting v2 any longer. You won't get responses in 60 days. Good luck."
It's easy to say, "they're doing versioning wrong" but it's the reality.
Everyone does oauth wrong. Everyone does versioning wrong. Everyone adds or removes fields that were once critical to someone's workflows. Maybe 25% of our integrations are a joy to work with. Once you get to the smaller players, things start to get weird.
> I currently maintain about 280 different API integrations.
And if I was trying to self-host software that integrated with 280 different APIs, I'd probably have that problem, too.
Most people don't need 280 different APIs. Then need a half dozen? Maybe a dozen? Most of which are likely to be the "stable big ones" that you refer to.
So the rate of breakage in your experience just isn't applicable for anyone doing anything at the hobbyist level.
That's not to say it's maintenance free! But I've been running my own automation with Huginn for a year now, and for what I need, I could not justify $120 per year.
Seen a lot of ppl complaining at the price on Twitter and offering other smaller, unknown services. People rarely think that these aggregators store your access tokens to all services and they pretty much work as passwords. I would look up these services first and look at their investors - are they well known for their security standards and all. I think their price is fair if running these connections is your thing and you care about your information.
I’d rather not host it myself, but, do you think IFTTT’s server costs are a significant contributor to the price? If they are, I have to question why this idea of cloud computing is supposed to make economic sense in general.
IFTTT is just a bunch of web requests chained together. The strength of the product is its UI and the number of services it supports.
I'm not, like, a fan of IFTTT, but it's likely that you just aren't the target market for the product; it might be valuable to you, but "people like you" might be a small enough niche that the product can't be made sustainable at a price point you'll swallow. This happens a lot, right?
You should compare recurring cost with other recurring costs. Otherwise, it will always seem expensive, because you can multiply an arbitrary period of time to the price.
After inspecting my own personal finance, I’ve found that my spent on software subscription is less than 1% of my yearly spent. So, I’ve stopped worrying about it quietly accumulate. Not yet worth optimizing.
Yes, I'm comparing it to the price of a vserver/cloud instance running cronjobs or webhooks.
That's what all technical people I know use ifttt for.
And usually a 3$ one can run hundreds of these sort of automations, unless you're relying on some of these ifttt-ingrained applets.
And before anyone draws the "this takes longer"/labor costs cards.. I only started posting my blog post blurbs to Twitter/Fediverse via cronjob because the ifttt-rss-to-twitter thing was broken at some point.
> Otherwise, it will always seem expensive, because you can multiply an arbitrary period of time to the price.
I would frame that the other way—it seems expensive because it is expensive, if you look at the long-term cost. And if you don’t look at the long-term cost, you’re fooling yourself.
You’re right that choosing one year as the time unit is arbitrary, but I actually think it’s an overly-generous choice, because as I mentioned, when I buy non-subscription software I almost always expect to use it for longer than a single year.
I’m increasingly viewing subscriptions as a cheap ploy to trick consumers into paying more. “Oh, that iPhone doesn’t cost hundreds of dollars, it’s just $20 a month for two years. You can pay $20, right?” What bothers me most about the scheme is it appears to be shockingly effective. This is literally grade-school arithmetic!
I’m not suggesting to look at the unit price, either. It’s not about whether it’s one year or one month.
I’m suggesting to put it in the context of other recurring expenses, to get a real sense of how much it affect your finance. A recurring expense has a different impact on your finance than a one-time expense, so it should be evaluated as such.
In fact, $100000 dollar upfront is less affordable than $5000 for 20 months, due to the nature of income and expenses.
The polling rate will be much quicker. Zapier offering is pretty weak at $15/Mo too. Better than ifttt but Zapier’s real power comes at the higher prices.
And what do you suggest replaces it? Developers don’t offer subscriptions because they are greedy. They offer them because it’s the best way to have a sustainable software business.
But it's not a software, it's a service that depends on a server farm, and a constantly evolving set of hundreds of third-party integrations.
It literally will break if left unmaintained if not funded properly, unlike a traditional software.
I do agree with other subscription-based models that sells software like Adobe Creative Cloud, where I'd be perfectly okay with having a non-updating version for a fixed cost.
An aside... I've never really got why technical folks would use IFTTT vs self-hosted. For non-techies I get it... but there are so many great and free tools available e.g. python, tasker/automagic/B4A, node-red, etc, etc.
I'm guessing this move to PRO will drive capable folks away.
Because I created applets/whatever’s on ifttt like 5 years ago and I haven’t touched them since, and they keep working.
God forbid I have to maintain a server, upgrade for new libraries, handle moving a server, etc etc. Write the code in the first place, debug problems, update my API keys...
It took me something like 2 minutes to set up an IFTTT action to email me whenever a keyword showed up in an RSS feed I'm interested in, and I didn't have to worry about keeping the script running or anything like that. There's no way I could have done it that fast even with Python, and I would have to maintain the damn thing.
My use of IFTTT is things like:
* When my garage door is open (Garadget) then turn on my garage indicator light (TP Link Kasa Lightbulb). That way if I leave the garage door open on accident I have a bright red light telling me that
* When the garage door opens after dusk, turn on the driveway lights
* When the doorbell rings (Ring) and it's after dusk, turn on a group of lights (Leviton)
* When the door unlocks (August Smart Lock) turn on a group of lights (Leviton)
I'd love to self host rather than use IFTTT, but is that even possible with those other tools?
Checkout Home Assistant. You can self host it, and as long as there is an integration available you should be able to do all that. Getting started with it is a bit of a pain compared to IFTTT but it's self hosted, open source, and free.
This looks like exactly what I've been looking for! Thank you for the recommendation. I started playing around with it in a VM and it's been great so far.
Yeah, this just pushed me into dropping my account. It's been slow or intermittently failing for ages as it is on the tasks I do have on there, so I wouldn't call myself predisposed to give them ANY cash to begin with. Limiting things on top of that just simplifies the decision.
I put a bunch of scripts together, accept webhooks and run scripts based on them or timed scripts after set intervals. Combined with a Telegram bot for notifications, and Ive replaced IFTTT with a messy. but more extensible solution.
Interesting move. It seems to me that IFTTT caters for a non-technicial hobby audience, because technical people and commercial outfits can replicate everything it does (and more) with a web server and some python scripts. And ten bucks a month is a lot for a hobby.
Replicating everything is Potentially a lot of time for a hobby too. Yet that is being advocated. There’s no way half the developers or techies I know would keep maintaining a self hosted alternative.
I'm guessing the free users aren't worth that much to them now they have got established. If 90% of their revenue comes from 10% of their users then this might be a convenient way to cull them.
This reminds me, I've been meaning to set up a Huginn install to test out. I don't really have a use for it at the moment, but I think more people in this thread would find it interesting. They describe it as:
> Think of it as a hackable version of IFTTT or Zapier on your own server.
I'm here for this. Zapier is super overpriced and IFTTT is hampered by its attempted "ease of use". As an individual, I need something in between, which hopefully IFTTT Pro is/will become.
Maybe we're getting a good deal with Spotify, Netflix etc with a 10$/month all-you-can-eat entertainment and it's worth more than that, but IFTTT Pro certainly doesn't feel like a 10$/month service. I registered for the 2$/month service for now, but I just did to give me some time to look for a selfhosted alternative.
65 comments
[ 3.2 ms ] story [ 141 ms ] threadInteresting approach. I wonder what users will pay on average during this first year.
If they give me a full blown console where I could run lambda-like python code on top of all the existing device integrations they already have, I'd be glad to pay the full $9.99/month. But the product would have to evolve a lot versus where it is today.
Are there any good OSS orchestration / automation frameworks out there that scratch this itch?
Incredibly useful and convenient.
Came in handy when I needed to poll something on my end, which would trigger the webhook when it met certain conditions.
1. Here's a URL (custom to each user/applet). You can either GET or POST to it. No auth/headers required (the URL includes a token).
2. You can optionally submit a JSON in the format of {"value1": ..., "value2": ..., "value3": ...} and set up the applet to set those values as the notification title, body, link, or image.
Rather unfortunately, you can only specify three values in the JSON so you can only ever have three of those four properties filled in for a single notification. But you can set up multiple receiving URLs/applets such that you can send the first notification containing three of those properties and then a second notification containing the last property.
Re Pushbullet specifically: it no longer supports iOS but either way, it's also much more complex [0] to use.
[0]: https://docs.pushbullet.com/#create-push
You can find huginn[1] & n8n[2] interesting.
https://github.com/huginn/huginn https://n8n.io/
We use it to relay tweets to a discord channel webhook. The requirement was that it had to relay the tweet contents with no flair or "tweeted by" or link to the tweet, all discord bot solutions fail this requirement and i didn't want to learn how2tweeterapi.
I also use it to send important notifications from some of my servers to my phone. I just have to `curl` the info to IFTTT server with my API key and send the info I need.
The problem with IFTTT has always been speed for me. I had the soccer goal notifications activated once I got my hue lights years ago so that when my team scored itd flash the lights. Usually this happened five minutes after I watched it on the TV. Useless.
Affinity Photo, for instance, cost $60, and VMware Fusion cost $80, to name two products I purchased recently. Neither needs to be bought again after a year unless an update contains a new feature I want (and probably not even then for Affinity), and both can do a lot more than IFTTT. Neither has a server component, true, but how big a cost can that possibly be for IFTTT?
I’m probably never using IFTTT again. Oh well.
Yes. Would you rather have ice cream or get punched in the face?
> Would you maintain it?
If maintaining it means scheduling updates through my host's package manager or running docker pull, and not "hey it's thursday, time for a new configuration format", then sure.
> How much is your time worth?
Enough to never host Elasticsearch, not enough to go SaaS-only.
Any service with any level of client integration will a) version their APIs, and b) keep things sufficiently stable so that clients aren't constantly breaking.
https://platform.ifttt.com/docs/api_reference
I currently maintain about 280 different API integrations. I have a team that does constant maintenance. Usually 1-2 API breaking changes per day is what we see after you get past the stable big ones. Even versioned APIs kick you off the old version pretty regularly. In the upcoming months we have 3 scheduled version bumps that are mandatory as they will discontinue the old version completely. Not "use the new version for more features", like completely 100% "we are not supporting v2 any longer. You won't get responses in 60 days. Good luck."
It's easy to say, "they're doing versioning wrong" but it's the reality.
Everyone does oauth wrong. Everyone does versioning wrong. Everyone adds or removes fields that were once critical to someone's workflows. Maybe 25% of our integrations are a joy to work with. Once you get to the smaller players, things start to get weird.
One thing I notice is how some do the constant refresh token and most don’t.
And if I was trying to self-host software that integrated with 280 different APIs, I'd probably have that problem, too.
Most people don't need 280 different APIs. Then need a half dozen? Maybe a dozen? Most of which are likely to be the "stable big ones" that you refer to.
So the rate of breakage in your experience just isn't applicable for anyone doing anything at the hobbyist level.
That's not to say it's maintenance free! But I've been running my own automation with Huginn for a year now, and for what I need, I could not justify $120 per year.
IFTTT is just a bunch of web requests chained together. The strength of the product is its UI and the number of services it supports.
After inspecting my own personal finance, I’ve found that my spent on software subscription is less than 1% of my yearly spent. So, I’ve stopped worrying about it quietly accumulate. Not yet worth optimizing.
That's what all technical people I know use ifttt for.
And usually a 3$ one can run hundreds of these sort of automations, unless you're relying on some of these ifttt-ingrained applets.
And before anyone draws the "this takes longer"/labor costs cards.. I only started posting my blog post blurbs to Twitter/Fediverse via cronjob because the ifttt-rss-to-twitter thing was broken at some point.
I would frame that the other way—it seems expensive because it is expensive, if you look at the long-term cost. And if you don’t look at the long-term cost, you’re fooling yourself.
You’re right that choosing one year as the time unit is arbitrary, but I actually think it’s an overly-generous choice, because as I mentioned, when I buy non-subscription software I almost always expect to use it for longer than a single year.
I’m increasingly viewing subscriptions as a cheap ploy to trick consumers into paying more. “Oh, that iPhone doesn’t cost hundreds of dollars, it’s just $20 a month for two years. You can pay $20, right?” What bothers me most about the scheme is it appears to be shockingly effective. This is literally grade-school arithmetic!
I’m suggesting to put it in the context of other recurring expenses, to get a real sense of how much it affect your finance. A recurring expense has a different impact on your finance than a one-time expense, so it should be evaluated as such.
In fact, $100000 dollar upfront is less affordable than $5000 for 20 months, due to the nature of income and expenses.
What does faster mean?
My biggest frustration is that my Maker Webhooks[0] requests take hours (4 hours on average maybe?) to execute with Wyze integration.
[0]https://ifttt.com/maker_webhooks
IFTTT has rarely worked properly for me with synchronizations running terribly behind or not at all.
why would I pay for that, with a challenging interface and just single step actions? Zapier does just as much if not more for $15 a month.
polling rate is an hour normally, now with pro you can get it at 5 minutes
If the developer is actually continuing development, new releases that are worth paying for will happen.
It literally will break if left unmaintained if not funded properly, unlike a traditional software.
I do agree with other subscription-based models that sells software like Adobe Creative Cloud, where I'd be perfectly okay with having a non-updating version for a fixed cost.
I'm guessing this move to PRO will drive capable folks away.
God forbid I have to maintain a server, upgrade for new libraries, handle moving a server, etc etc. Write the code in the first place, debug problems, update my API keys...
I'd love to self host rather than use IFTTT, but is that even possible with those other tools?
https://www.home-assistant.io/integrations/
> Think of it as a hackable version of IFTTT or Zapier on your own server.
https://github.com/huginn/huginn
The 3-applet limit is weak, though.