Goodbye IFTTT
benjamincongdon.me
benjamincongdon.me
Spotify is one that comes to mind, originally there was so many triggers on Spotify that were useful, but now there is only like 2 boring triggers. It really killed the IFTTT platform when there was less options from so many of the services.
They went from having less services, but a very deep amount of triggers/hooks to hundreds of services with very little depth. From a deep pond to a wide shallow ocean. And I'm not sure if that's IFTTT to blame, or the services themselves just doing it and IFTTT had no say. Could be a bit of both, who knows. Or maybe services made exclusive partnerships with other services like Zapier, and made their triggers exclusive to Zapier for some deal they put together.
And just to add, at the end of the day, anything automation related that I was doing for professional work was always done on Zapier because they had far better/more options. So IFTTT had always just been a personal playground, noting serious, which is all more reason to not pay for it.
Worries me that with post-smartphone technology it’s more valuable for companies to remove interoperability than to foster it.
Look at Instagram, you can’t even post a hyperlink because it’s more beneficial for them to prevent a fundamental internet feature.
I've been in this position before. We wanted to open up our APIs, encourage integration with 3rd-party tools, and make our product as open as possible.
Good intentions, but the unintended consequences are significant. We expected the openness to drive more sales, improve customer satisfaction, and generate more goodwill around our product.
In reality, less than 1% of our customers used the API at all. A portion of those who did were the most demanding customers we had, constantly complaining on social media that our API didn't support everything they wanted from day 1. Ironically, the most vocal API-using customers were more negative than positive for us. The extreme fringes of the DIY hacker communities can get very entitled and ugly.
Creating and maintaining the API was more work than we anticipated. With every new feature we had to make the decision to exclude it from the public API, or spend 50-100% more time integrating it into the next public API release cycle.
Overall, it doesn't make business sense to create an API if it will only be used by <1% of your customers. The only time an API makes sense is if the API will really, truly, genuinely be used by a significant portion of your customer base.
As techies, it's easy to forget that tools like IFTTT aren't mainstream outside of technical products. They're actually extremely obscure for the vast majority of the general public.
Because because if they were on a free plan, why did they get API access anyway - that does not sound sustainable.
I don't think decentralization is inherently a non-starter and I don't see a convincing argument why it would have to be. It's a matter of marketing and UX.
We aren’t the customers and we need companies that will treat us properly, but imho the root cause is providing everything for free - that will necessarily need to change.
It's likely due to the scale of spam and abuse with links on IG.
I'm not seeing them now, I stopped using Instagram.
Evidence: I run a SaaS and had a long-standing TODO: "Implement Zapier and IFTTT support". Well, I eventually did implement both, and then learned that in the meantime IFTTT changed its approach, started charging businesses, and the amounts are not insignificant. I had no idea if any of my customers actually wanted IFTTT, so I simply killed the integration and sticked to Zapier.
I'm guessing "hope Google calls asking for a new integration and charge them a million dollars" was the business model, and they're hoping to transition into a model where they charge users.
Absolutely. The 2000 - 2010 time range was filled with such great hope for APIs, for expanding humanity creativity.
The last decade though has been shuttering & closing & withdrawl of computing, the systems receding, getting further & further away from general usability, a retreat into the walled garden, systems effervescing from manipulability, going up, into the cloud.
I've been calling the before time the Pax Intertwingularis era, the peace of intertwingularity, interoperation, interfacing of systems. When we were all excited to build & interconnect & share, when the hope, what we all saw, was humanity getting better & better from the power to wire ourselves together in new & changing ways.
(demo video here https://youtu.be/mY82F8GMSo0)
We basically allow you to create your own custom triggers based on any change happening on a website, with value presence, absence, increase/decrease detection (deltas) etc.. and we’re compatible with both Zapier and IFTTT (and custom webhooks if you want)
disclaimer, I’m the founder
And I always suspected these services wouldn't be free forever - and what a surprise, they eventually had to figure out a way to make money.
If you don't want to go commercial like HomeSeer there are many other open source projects like HomeAssistant that are still relatively friendly and provide great functionality. The RasbperryPi has really been a game changer offering cheap local processing for very cheap. I don't mind using cloud for remote access, but I would never want all my logic running remotely in the cloud. Been burned too many times by services either changing with zero or little notice, or as I stated earlier, internet issues.
With projects like HomeBridge, gluing different standards and universes together is getting easier which also negates a lot of the necessity for things like IFTTT if you are willing to trade some time for convenience of just subbing to a service like it.
Choice is always good :)
Example: Making a smart light flash red every-time President Trump tweeted was just a silly goof to setup to show friends. Not something I'm actually going to pay for. haha
Just to play devil's advocate, have you ever been introduced to either a job opportunity or investment opportunity via your friends, who see that you're able to build stuff? If so, you could potentially place a monetary value on the benefit you derived from doing that.
Especially developers, they'd gladly trade their free time for other benefits like more control or openness like the OP is doing here:
> It's just that this was the nudge I needed to move my automations to infrastructure that I have more control over, which has other benefits outside the scope of this article.
On a tangent, I think it's important to understand "why" is that. Not because most people are stupid, or they don't know the concept of opportunity costs, as popular explanations go. It's because for most individuals, time is illiquid. You can't just turn every 10 minutes you save on some chore into 1/6 of your hourly rate. So, for most people, that $2/month is a pure loss, even if it saves more than 10 minutes a month, so it needs to be (or at least feel) truly worth it.
It's no wonder consumer software, especially productivity software, is mostly free nowadays even when they cost a lot to develop. It's actually pretty hard for consumers to justify paying for any software. The consumer software developers that succeed usually shift the software cost way from the consumer somehow, either by creatively re-arranging prices (eg. baking the price into bundled hardware) or charging someone else (eg. ads). IFTTT has a tough gig and I wish them the best.
I don't agree.
If the service is free, you can just use it. The moment you have to pull your wallet, you start doing cost benefit analysis. "Do I really need to use this?" The answer is often no, even though some benefit could be potentially worth more than $2. Potentially. Often, it is difficult to calculate a monetary value.
Not only that, but this is a case of "I have altered the deal, pray I don't alter it further". Had they started with some limitations in place, and then charged to expand them, this probably wouldn't have caused such backlash. People don't really like things taken away from them, even if they originally paid nothing. They tend to prefer deriving some sort of "benefit" out of a deal.
As for developers... I'm another one who ditched IFFT. Not just because they want to charge me, but because they started charging manufacturers a lot too, which caused many of them to drop IFFT integrations (even though these same manufacturers kept many others, like Alexa). Having things stop working, specially if it's related to home automation, can be a bigger hassle than setting them up "properly" in the first place.
So now I have Home Assistant and Node Red. With Hass.io, all you have to do is download the image, put on a SD card, stick it into a Raspberry Pi and you have Home Assistant. A couple of clicks and it installs Node Red already integrated and ready to go.
Does it cost more than $2? Certainly. In both labor and parts. In return, I can integrate almost all my devices (I'm missing a couple that don't have APIs at all, but they didn't work with IFFT anyway). I can do things that IFFT can't even dream of. I don't have to worry about price increases, EULA changes, keeping payment information up to date, cloud or internet outages. If the Pi goes up in smoke, I can throw it away and move the SD card over (or restore a snapshot).
Yes, it did cost a couple of hours of work. But since I had to do something about it anyway, might as well bit the bullet and trade off the time I'd be watching a movie to do this instead. I'd argue that the resulting system is worth way more than $2 a month.
Now they have that reason, and so they wrote this article.
Anecdotally, here is the first shortcut I wanted to create when Shortcuts appeared on my iPhone: switch off Slack notifications at 6pm on Fridays, switch them back on every Monday at 6am. This is still not doable as of iOS 14.
I really don't get the hype around fancy automation (stuff you can see on website like MacStories for example) if simple toggles I need to improve my life are not there yet.
FWIW, you can do this in Slack itself (on iOS, see You > Notifications > Notification Schedule).
iOS 13 added time- and location-triggered Shortcuts (see Automation, but Slack doesn't expose this functionality to Shortcuts, and (as you note) Apple hasn't yet exposed iOS notification controls to Shortcuts.
For example, when I get to work, i wanted it to log the time, all it did was showed me a notification that I had to click (!) before it did the task I asked it to. Silly.
One feature I'm using more these days is the "share" popup. For example, one shortcut [1] i hacked up accepts an amazon url, extracts the product id and opens that product on camelcamelcamel. So now I can be in the amazon app, hit the share button, click this shortcut and see the product price history. pretty useful. Without this shortcut it was pretty annoying. CamelCamelCamel folks could use something like this to increase their usage :)
[1]: https://www.icloud.com/shortcuts/fc526325b9eb4ebf99bce8da65a...
In iOS 14 you can run more triggered shortcuts without needing user confirmation — I would see if this is possible for you now.
> Shortcuts isn't really automation. Its just what it says - shortcuts.
No, it's automation, in the quite literal sense. You ran into trouble running shortcuts based on triggers, but that doesn't change the fact that it's essentially a scripting language that allows you to automate tasks that would otherwise be tedious.
A good starting point would seem to be what OP wanted - enable Shortcuts access to every toggle in the Settings app. Let me programmatically read and update notification, widget, screen, background, accessibility, Siri, battery, privacy, etc. settings that I can only access through the Settings app. Yes, I know I can do this for WiFi or Bluetooth settings, but there should be much, much more. And I should be able to consent to Shortcuts doing things in the background without my explicit permission, much like I can set up cron jobs on my home PC.
> And I should be able to consent to Shortcuts doing things in the background without my explicit permission, much like I can set up cron jobs on my home PC.
Again, this is now the case for many triggers (not all, I don't think) on iOS 14.
My guess though, is that they've already got this working, they're just going to wait until iOS 15 to drop it. The integration of Shortcuts into iOS has been slow but steady ever since they were acquired, and I think eventually it will do everything we want.
That said, it seems they did handle it in a much classier way than Wink who really pissed people off when they moved to a subscription model. Likewise, that was a good nudge for a lot of people to move off the "easy" platform and figure out something better.
I wonder why companies don't use a boil-the-frog approach to try to avoid this "nudge". Like, if they started off by charging, say, $1 a year I'm sure everyone would stay and just go "OK". Then start increasing the price slowly until you find the sweet spot.
There's a psychology-caused pricing dead zone if the price is too low. Users assume the price of something implies its value. If the value is lower than the mental effort to decide whether or not to purchase it, most won't. In the user's mind, somethat that only costs $1 a year can't be worth the effort to decide whether to spend $1 a year on it.
Maybe $5 a year could work as a price point, and given the gulf between Pay Nothing vs Pay Something, I doubt $5 a year would've lowered the conversion enough to make $1 a year transactions worth it. Don't do credit card transactions for a dollar or less unless you're counting on it converting into a higher amount a month later.
Because, just like in real life, the frog will jump off way before it reaches the boiling point.
TBH I never understood all the hype about IFTT; most of its usefulness can actually be achieved by the services themselves, once they choose to care about integration. The early focus on usability had already gone overboard years ago, when their redesign basically made it mobile-only. It always felt like a typical SV-hype-driven phenomenon, mostly based on short-lived home-automation gizmos, and I'm surprised they're still around - although this sort of move might well indicate they are circling the drain.
But their spam killed me. They started sending a new message asking me to subscribe what seemed like every day. I eventually clicked the unsubscribe link which seemed to reduce the flood but I still got a handful more. So now all of their mail goes directly to my spam bin.
I don't need to be giving money to a company that spams like that.
And I personally also think their free tier should be a little bit better. Currently it's extremely limited and I barely find any use of it any more.
They're constantly removing services too. I used to use it to do one thing, which was to automatically tweet any songs I like last.fm. About 2-3 years ago, they simply removed all the Last.fm triggers.
I'm glad to see they're still doing well, and I'd still recommend them to non-technical friends but huginn is a lot more enjoyable to use and I found it really simple on a day-to-day basis.
So still the same concepts of inputs -> outputs, but you would self-host on AWS Lambda / Azure Functions / etc. It would also have more flexibility to add custom code if needed to transform data too.
(disclaimer: I am the author)
Those projects already exist, you are probably better off contributiing to existing ones rather than creating a new one.
Probably but there's so many IFTTT alternatives around (including the referenced node-red). If you're more tech savvy, you can either stand something up yourself, or you can use something like glitch (https://glitch.com/) to create some quick webhook adapters.
For example, you can't do anything all with Wyze hardware directly
Yes, I get that doing other things more advanced like IoT integration and whatever else IFTTT has moved into bring up challenges etc, but that basic thing at the heart of what the If This Then That concept is rooted in is still useful and them asking for a few bucks isn't that bad. I've still got some things running out there to send an email, post a tweet etc on updates. But doing some things more regularly than their tweet/check limits has forced me to move elsewhere on my own recently. All depends on the use.
And yeah, totally miss Yahoo Pipes but from the time when I used that, IFTTT actually improved slightly just enough to cover what I was using Pipes for haha. (and yeah there's pipes.digital)
ActionsFlow recently shared around here using github actions is interesting but I didn't feel like migrating a bunch of stuff over to it just to test
If Time is 5:00p If TempOutside < 70 and Send Alert to "Open Windows"
It sounds like they now support multiple IF's but I can't use that feature in my three free, so I will never know if it would have worked for me. Getting a notice to open my windows when it's 70 outside isn't worth the price of admission but I suspect I could get hooked and find other things that were more useful. Just never happened for me.
They have a ways to go in terms of improving UX, but I think that is coming.
Yes totally agree. Still a lot to do and improve.
Even something as non-programmer as taking notes - what if I want some notes automatically added/removed/modified depending on what happens elsewhere in the world? I need a programming language!
Almost no problem is too trivial to not want a programming language attached to it which begs the question - why don't we already have it? I hope the answer is quite simply that we've been so focused on new hardware, that software is lagging behind. There are more sinister possibilities but I'd err on the side of nobody having made it a priority because most people think small by default.
The only downside to plain, old code is that there isn't a pretty UI for it.
I run a small service [1] and I've integrated it with Zapier but there's absolutely no way that I'm gonna pay $199/year to support IFTTT.
I don't value my time too highly, but I certainly do value it enough to not want to put in the grunt work needed to work with the APIs [0] that my smart thermostat provides. While as with IFTTT, setting the temperature given a particular trigger is a couple button clicks.
[0]: https://www.ecobee.com/home/developer/api/documentation/v1/o...
There are a few integrations I use frequently, which I could do the integration myself, but it'd take a good weekend of my time to do them properly.
I'll rather pay the $24/year, since I value my free time more than that.
In other words a programmable ecosystem like IFTT?
It also runs on a free-tier Heroku Dyno.
If I was ITTT, I would ask him not to slam the door on his way out.
They are offering a life-long 80% discount for early registrants, with one month's notice. That is quite generous. To twist that into a "pressure sales tactic" is one of the most HN things I've heard.
To me, being asked, "Hey are you gonna buy this? You've only got a month, come on, buy it already!" on a daily basis qualifies as a pressure sales tactic, no matter how good the deal may be.
I have an email that says "Upgrade to Pro before your Applets are archived" and then one for 12 days remaining, one for 9 days remaining... 6, 5, 4, 3, 2 and finally 1 day remaining.
Imagine if this was the sales page for the newest iPhone with a "discount for existing customers" and only one month to use the discount. The entire internet would be up in arms.
Pressure doesn't mean that their product does not have any worth. In fact, I subscribed to it. It just means that they used pressure to close the sale.
If that's true, no way I'm paying $2/month for that. I'm moving to Autocode.
Edit: OK I see I can still edit RSS applets after they're created. Still doesn't seem worth keeping the service.
For those who don't know - we have a full App Automation platform that runs on serverless tech. We automatically handle auth to your favourite APIs and easily handle webhook setup and signing. Apps can be published open source and shared with the community. We're easy for non-developers to get started with (folks more on the business operations side can get started via our Apps [1]) and everything you work with is ultimately code.
We're definitely not a "consumer solution" and if you're looking for something self-hosted we're happy to suggest alternatives. We run in the cloud but have a pretty sophisticated system that supports version controls, rollbacks, CI (dev vs. prod) and a whole ton of other neat features.
We have a pretty permissive free tier for developers, but what we really focus on bringing to the table is professional & team tooling around automation which is what we charge for. We don't charge by # of automations, steps, or anything like that -- just access to organization-level tooling that allows you to more easily collaborate with a broader team.
Happy to answer questions if you have any! (We just launched officially on July 7th, so if you haven't heard of us yet -- that's why.)
My only issue is that most services do not have enough triggers or abilities to prove useful for the workflows I currently want. For instance, I would like to append to a Dropbox Paper document when I star a github repo, possibly filtered by some logic. Guess what? Even though both APIs would allow this, IFTTT has neither the trigger nor the action I need. This is just one example, but I frequently run across this problem, which is a major inhibitor to me using it more. I don't have a ton of "smart home" devices which is where IFTTT shines the most IMO.
I'm staying on for now to see if I get enough value at $2/mo with the triggers and actions currently available. If I'm not getting much out of it a year from now I'll probably switch fully over to a more programmable solution.
I did sign up for the $1.99/month plan though.
I had no idea what Duplos was so I had to Google it, but I can't imagine that for a complete non-programmer, the Scratch UI would be more clear than littler sticking giant building blocks together without issue. It sounds like the author simply fell out of the target market for IFTTT, rather than IFTTT becoming _worse_.
FWIW, that Scratch UI screenshot was so hard to mentally parse. I would not expect a non-programmer to look at that and feel more secure in their understanding of the product at all.
For folks looking for a lightweight webhook processor, "faasd" is a reimagined version of OpenFaaS which runs on a single host without K8s. I would imagine it being used for "just a few functions" and automations etc. https://github.com/openfaas/faasd
On top of that, I've got RSS feeds for some subreddits I run that benefit from IFTTT. I also have some RSS feeds automatically go to Pocket so I can read them on my Kobo.
IFTTT isn't perfect... and I absolutely hate the interface, but for the simple tasks I need, it does them well. And that is easily worth two bucks per month.
With functionality like that, effectively for free (cost of device not withstanding), how can IFTTT compete?
edit: And now I've discovered I can create trello cards this way!
Whilst Shortcuts is great, there's a lot it can't do - it can't accept webhooks to post content to your Day One journals[1], for example, or read RSS feeds and add those to your Pocket library or add your Netatmo temperatures to a Google Docs spreadsheet or ...
[1] Because Day One still doesn't have an API.
Currently there are no other alternatives to IFTTT for all the triggers and actions I have across these products and so I just signed up for IFTTT Pro for now.
It’s puzzling that they didn’t sync up the nerf + pro option.
I wrote about my experience: http://blog.pinballmap.com/2019/12/05/iftttwtf/
My use case is: check an RSS for new items, and tweet those new items.
IFTTT is great for quickly moving data around, but I wish there was a zapier/google apps scripts/lambda functions way to quickly hack apart and reassemble data for easier consumption by external services.
I guess the gamble is will more than 20% of those who would pay $2 pay $10.
$10/mo is not much for a Western programmer. If they did their market research I could see them coming out on top.
While I think this is true, it's also frustrating that desktop is the only way to edit filter code. So I find myself wanting to use both interfaces for different purposes.
Still, I have some zaps from zapier that I'm going to move over to n8n. It's good enough.
I think one of the missing pieces on DIY/user serverless is that there's not a lot of connectivity. Systems like node-red have integration points for all kinds of systems, whereas serverless tech tends to have a more concentrate developer-oriented focus for it's connectivity, taking events off queues and via http endpoints.
We're only just starting to bring more focus onto the messaging bus underpinning serverless. Knative Eventing, for example, is one of the contenders, which defines a pluggable architecture for what events are, where they come from, how to write more. That way you could start extending the system, having events come in for your light switch, or your weather sensor, or your car's location or what not, adapting whatever protocols or systems you might have.
I'd move to zapier but they want $20/month which is crazy for a hobby.
If you were using the service for free for all these years, you were a cost to them. Not a benefit.