Hacker News new | past | comments | ask | show | jobs | submit login
How Zapier Reached $35M ARR (ryanberg.co)
243 points by gmays on Nov 15, 2018 | hide | past | favorite | 95 comments

I remember about two years ago when I reported a bug in the Hubspot Form API Zap where they weren't handling one of Hubspot's documented return codes properly. Basically their zap was only handling one success code and one failure code when actually there were two different types of success.

I documented the issue, linked Hubspot's API docs confirming it, submitted a ticket through their normal support portal and had a confirmed bug from a Zapier engineer within an hour. The fix was live before I came back from lunch about an hour later.

This is why they're successful. If I submit a bug report in Blink (usually major text rendering/CSS issues), a Google employee will look at the issue a month later, maybe. If I submit a bug report in one of their APIs, if it isn't Adwords and isn't customer impacting, the response is basically a middle finger/"we will fix never".

Zapier made a lot of my contracting work way easier and often I get the client involved and they become self-sufficient. Zapier should pay me money! kidding

Slack is also super responsive when you file a ticket and in some cases have pushed out a change in a few hours. They accept tickets and issues from anyone (not just admins).

I wouldn’t even know how to file a ticket with GDocs.

My home address has been bugged in Google Maps for months. When you enter it, Google Maps modifies it for some reason and makes it point to the wrong location.

I've sent a bunch of reports and "feedback" in about it but it all just seems to go into a black hole without even being acknowledged.

For example Google Fonts - there are several very good and popular open-source fonts that still (after years) waiting to get pulled into the service. You can just look at the tickets at https://github.com/google/fonts

This is quite suprising to me because i think google this way gets their snooping JS on lot of websites. It generates data for them. Sure they also have to host it... but then again this is just matter of accepting pull request.

Can you explain the last sentence? Sort of blew my mind..

Like if you find a bug in Google Docs ... what do you do? Scream into the pillow? Work around it? Complain on twitter?

I believe OP meant "file a bug for GDocs", not "file a bug using GDocs".

Ah, that makes sense. I was thinking of how one could file a bug using GDocs and it was an interesting idea. Thanks!

Theoretically I think it's possible to submit bugs using GDocs especially since you can essentially use Javascript to do so. The coding engine is pretty powerful. I can imagine using forms to do so.

Chrome vs. Zapier is an unfair comparison. Chrome is ~6.7 million lines of code and takes hours to build while Zapier is probably a microsevice architecture and is easier to update a component / validate an issue.

I'm not _just_ talking about Chrome. I think that you would be hard-pressed to find any Google product where you'll find a responsive support channel.

Had a major DNS bug with Google DNS, it was on their side and they were completely messing up how they were handling some requests. Took them 2 months or so to fix it, they never described what happened, never said what was broken, and it took finding resources inside the company to solve. Google doesn't functionally have support for many of their paying services, nevermind their non-paid services.

Having humans doing customer service is not something google will ever do well. It’s not in their DNA. They are all about solving big problems with fancy algorithms.

There is a reason Google keeps on making failed social networking products and a 100 messaging apps. They are far more interested in algorithms that can scale than messy little customer issues.

The problem is that you can’t deliver great customer service with algorithms just yet.

Couldn't agree with you more. Google is not a people-company.

Yet Google is very successful

At selling ads.

> Zapier is a monolithic application built on Python and Django.


So what you're saying is microservices for desktop apps, right ?

>If it isn't Adwords and isn't customer impacting

I think that explains it.

>Zapier made a lot of my contracting work way easier and often I get the client involved

Could you please elaborate more?

Also if you ever wish one missing feature in Zapier, what would it be?

> Also if you ever wish one missing feature in Zapier, what would it be?

For me it'd be a price point between free and 25/mo. I like the product and would have so many uses for it.. but I'm a person not a business and $25/mo is a bit too steep for what I need

>but I'm a person not a business and $25/mo is a bit too steep for what I need

Same. There are several things I want to do with Zapier but at the free tier I have to limit it to one if I want it to for sure work all month.

I'm not using it for business, I'm using it to make things a little easier for myself wholly unrelated to any income generating thing so it's just not worth paying 25$ a month to me so for right now I only use it to get text messages when I get email from a specific individual.

What kind of scenario is worth $12.50 but not $25 to you?

In general, it doesn’t make sense to price a business tool so that an individual can afford to under-use it.

>In general, it doesn’t make sense to price a business tool

SAAS isn't exclusively a business service and many, many, many services have multiple pricing tiers. In fact, Zapier has multiple tiers... for businesses 'starting at 20$' and for teams 'starting at 250$' a month. They scale price based on how much you use the service.

With the free tier you are limited to two 'zaps', they run less frequently, you don't have access to multi-step, you have a rather finite limit on how many times they can run each month etc.

It's just bizarre there isn't a lower tier. I mean, it's even more than my squarespace hosting and squarespace has a 12$ and 18$ tier. Or evernote. Or any number of other services.

I'd happily pay 5$ a month for considerably more functionality but I'm not trying to automate an entire business and 20-25$ or more is nowhere near justifiable. Sure, if you make 6 figures maybe you're like "hey 20$ to handle some weird email situations for me and make some docs automatically, sure!" but for the majority of people it's a "well, guess I won't be giving them anything".

Not all SaaS are B2B-focused, but Saoier is. They sell tools that make sense for businesses to use. It doesn’t make sense that they should offer something to someone who barely values what they do (you most likely can afford $20/mo if you can afford $5/mo, so your issue is that you don’t think you are getting that $15/mo value from Zapier, and that’s most likely because you aren’t running a business and are trying to automate something unimportant.)

And it might still be fine for Zapier to provide a cheap tier for people who don’t value what they do, but if they offered that, too many businesses that get a lot of value from Zapier from, say, running five zaps, would downgrade.

When you are in a situation where you don’t want something very much, and a seller would lose a lot of money by trying to make you happy, you find a substitute and Zapier has plenty, especially for a developer with more time than money.

Two patterns to lowest tier customers:

1. They still submit support tickets. Sometimes more often than people on higher tiers.

2. They’re hard to convert to higher paid plans.

Also, the market for note taking and web hosting is bigger than Zapier. It’s not quite as horizontal.

In the end, sometimes it just isn’t worth it to offer a lower price.

regarding the pricing, i dont think its fair to compare Zapier (or IFTTT) services with other SaaS. The reason is there is considerable amount of investment on maintenance from Zapier side for keeping their plugins/libraries up to date with the APIs/services. If you consider evernote, once they develop it, they have to fix bug on may be the core platform plus a dozen other microservices and UI modules. But for an integration service like Zapier, they have to maintain more than 200 and sometimes close to 1000 components.

>The reason is there is considerable amount of investment on maintenance from Zapier side for keeping their plugins/libraries up to date with the APIs/services.

Which they are already doing for the companies paying them several hundreds, or thousands, of dollars a month...

That's like saying "Sorry, we won't sell you 100 lowly private individual an Impala, we only want to sell hundreds of them at a time to fleet-buyers, it costs us a lot of money to design these things!"

That’s actually how it would go if there were a lot of interested fleets and not enough individual buyers to sustain dealerships.

I totally understand that you want to use Zapier and save money, but you have to keep in mind a) you are (likely) a more expensive customer than you think you are, b) Zapier can’t serve you without self-cannibalizing. And this is an issue because you undervalue software; if Zapier is worth anything to you, it’s likely worth $20/mo but you are conditioned to not pay for something like that. This isn’t peculiar to you, but it’s something to be aware of.

> Also if you ever wish one missing feature in Zapier, what would it be?

Extracting images from emails!

If anyone from Zapier is reading, how do you folks do it? What does your support staff look like? Do you have allocated support engineers?

I'm on a team of fiveish serving a whole company of fiftyish. Sometimes we resolve issues quickly, but often it pains me how long resolutions take. I'd love to do better but it's not clear how, short of staffing an engineer towards those reactive things specifically.

I work at Tyk.io and our average response time is less than an hour. If we talking about resolution time, it is about few weeks (depending on release cycle). But overall we are so good at it, because all team members, including engineers, do the support. There is some "first line" of support, which answer basic questions, and re-directs to the related person. Additionally, since we are a remote team (same as Zapier), we have team members across all over the globe, and literally 24 hours online.

Involving engineers into support is not cheap, but it pays off with extraordinary customer loyalty.

And for sure a lot of automation, extensive internal and external FAQs, and good documentation.

New ticket: don't intervene with the user scroll in your home page. It's painful. (I'm on a mobile device)

Yeah, we are aware of it. In process of fixing.

I agree and had a similar experience with them. Plus, their suite of integrations just work.

Clickbaity title aside, Zapier is achieving $35M ARR because their product is amazing. I'm infinitely suspicious of SEO articles but I'm convinced that Zapier could have Slack'd (word-of-mouth'd) their way to where they are now. They're in a league of their own when it comes to integration/functionality/UX -- I don't even know who their bigger competitors are, if they exist.

The first time I saw Zapier, I got the tinge of developer/engineer/builder envy that I get whenever I see products that just are obvious in retrospect, and seem to be very well and flexibly implemented. I heard of it from a forward-looking product manager I respected at a famous internet security firm that I worked at in the past and it absolutely blew my mind (also he gave me a short demo) because of how easy it made things for end users despite the complexity I know the must have had to deal with.

That and like every Saas company offers or refers people to zapier

They've singlehandedly turned hiring a developer to connect A to B into a business risk.

An entire area of contract work is basically irrelevant. I get to bill people to train them to do simple things for themselves, cheaper and focus on the more important work instead. And I don't lose any of those clients because everybody won and they have other work to give me.

They deserve it!

Yeah, IMO that's another part of their genius business strategy -- all the SaaS apps that don't want to write plugin ecosystems or abide by an ever-changing list of interoperability specifications and want developers to love them can just pick one protocol (in the general sense of the word 'protocol', not the HTTP1/HTTP2/HTTP3 sense) -- Zapier's.

Then they can put "great interop" on their landing page, list some companies that Zapier interops with on their page, and be done.

I admit I haven't looked too closely, but what makes Zapier so amazing compared to IFTTT?

Just to name a few:

- more small business/enterprise integrations (e.g Salesforce, Shopify, Magento)

- can run Python or Javascript as part of a workflow

- complex workflows with branching and conditionals

I can spend several hours building and deploying something with AWS Lambda and API Gateway or 15 minutes in Zapier and it just works.

IFTTT is a toy by comparison, but is also more straightforward to use - different audiences.

The way I see it, Zapier is a subset of IFTTT, but it's not the functionality that separates them, it's the approach.

For me it starts with the name -- IFTTT means absolutely nothing to the average non-technical user, it sounds like alphabet soup or maybe a standards body or something. Zapier also doesn't mean anything, but it means nothing in a easy to distinguish way from other things. The audience for IFTTT (which is obvious once you know the acronym and makes sense) is for developers where Zapier is generally for smart forward-looking non-technical to fully technical people in the right domain. IFTTT might get a crazy article or two in tech-focused publications and some fanfare from nerds but Zapier is the kind of thing that silently chugs along to be everywhere in ~5 years once product managers/leads, managers, business analysis people get a hold of it -- and the tech isn't even bad.

I don't work for Zapier, I don't push use or pay for their products, I have no connection to them except realizing when I first saw it that it was a excellent idea in a parched solution space with an above average to great implementation & UX.

[EDIT] - Thinking about this some more, Zapier is a subset of IFTTT the concept as in I could Zapier's functionality in the sentence "if X then Y", but that statement actually doesn't mean anything because you can maneuver X to be just about anything and Y to be just about anything. Zapier supports non-trigger based workflows from what I understand and so does IFTTT... so maybe they're more equivalent than anything with Zapier focusing on enterprise-level polished integrations.

I would put it the other way round. There's stuff I do with Zapier that I can't see how to do with IFTTT. In particular the path functionality and Python/Javascript embedding mean I can string lots of zaps together into very complex workflows.

IFTTT created a product (and a closed one at that, being dismissive of developers wanting to give them free labor to create integrations). Zapier created a platform and actively courted developers wanting to add integrations.

It's been a year+ since I've done any work w/ either. At the time IFTTT was 'consumer-y' and Zapier was 'business-y'. By which I mean, I could wire together meaningful workflows in Zapier across multiple SaaSs inserting my own javascript to clean things up along the way. At the time IFTTT wasn't as developer friendly.

From what I've used they approached a similar domain problem from different angles.

Zapier - B2B and web app apis, IFTTT - Home and IOT automation.

For me it was that IFTTT allows only one IF condition and one THEN step, whereas Zapier allows more simoltanous IFs & then THENs. IFTTT is free, & kind of have no limit, whereas Zapier allows only 100 executions of your zaps in a month.

I've been following Zapier from the sidelines for a while, and from early on I was fascinated about the way they managed to leverage the network effect to grow.

In retrospect it seems obvious, but they probably would't be where they are today if they hadn't built it as an open platform (for partners developing their own zaps) from the beginning.

Actually met two of the founders (Wade and Bryan) in mountain view when they were based in the YC offices (5+ years ago). Sometimes it's easy to think someone's just lucky, but in this case I think it's pure focus, talent, and a great understanding of the acquisition sources that got them here.

Looking forward to see where they'll be in another 5 years.

“created 25,000 search optimized landing pages”

It’s garbage that this approach works for SEO. Who wakes up in the morning and decides that creating 25k versions of the same web page is a good use of time, automated or not. Apparently it pays off, but I don’t want to compete on who can generate the most fluff, and it’s bogus that this behavior is rewarded.

They mention that these pages were created for every combination of “connect service X to service Y”, which matches twhat people are actually looking for, so I would argue this is a great use of 25K landing pages that are adding value and being rewarded by google for doing so.

It is fantastic. I pay for Zapier and sometimes I end up on these pages when searching simply because I was unaware of a new integration. The pages do a fantastic job of showing people a solution they are searching for.

It's funny. This happened to me yesterday. Not a new integration but I wanted to know if I could get data from a forum into a Google sheet once a month.

Zapier landing page: Sure, just use a webhook trigger to Google sheets.

It's not 25k versions of the same web page. It's a page for each combination of app that they service.

"Connect Trello to Gmail" is a totally different thing from "Connect Stripe to Salesforce".

It's not "totally different." It's connect "API X to API Y." Extrapolating every single permutation of this concept to outrank the competition on a specific search phrase in Google is a hack, and a failure in the broader vision of web search. These "25 thousand pages" are created for computer search algorithms first, and humans second. It's a dumb game we're forced to play to stay competitive.

They are completely different to the person making the buying decision, and they are only completely different from a technical perspective if you’re an architecture astronaut in low earth orbit that all you see of an API is boxes on a diagram.

no, they're explicitly created for human searchers. Searching for "connect gmail to workflowy" or whatever combination should bring you to Zapier (or another company that does the same thing), right? These landing pages facilitate that.

Just to add to that, when you're creating a Zapier plugin, you need to add human readable descriptions of your triggers and actions - so the actual content about linking X to Y or X to Z actually becomes readable and usable as a help page.

The template is the same, but the content is different. How is it that different from LinkedIn ranking for an individual's name or a news article or forum discussion?

I don't know about others but I favour something like StackOverflow because it's consistent and predictable. I know it will load quickly, things will be in the spot I expect, and I won't have to jump through modals to get to the content. They'd have more than 25,000 informative and templated pages ranking and it's a positive thing.

Say you do business anywhere in the state of California. To another human you would say, "we serve the entire state of California." But to outrank your Google competition you must create a landing page for every single city in California you do business in. This is not the same thing as a different forum post or different profile pages. This is auto-generated redundant fluff. It bring no value other than outranking the competition on a specific search phrase, and possibly misleading the reader to believe you're more local or focused than you actually are. I get that it's the rules of the Google game we're all playing, but I don't buy that it's somehow positive to the web user's experience.

But if you do business everywhere in the state of California, how is it wrong to say that you also do business in the cities of Los Angeles and San Francisco?

It's not a lie; it's an exaggerated truth.

It's what most "plumbing" style companies do, by making a separate page for each permutation. It's easy to generate with some stock data for each option and is very good at matching the queries that users actually type.

It only works if the company supports it obviously, but it's a nice win for them and completely legitimate in terms of both search engines and users finding what they need.


I found Zapier many years ago because I was searching for a way to connect two specific products together. They made a page to advertise their general purpose tool as a way to solve my specific need. It was effective, and at a time when there was no other way.

Not sure why you see this as such a negative thing.

I love Zapier and we use it more and more for various bits of glue, but I haven’t found a robust development workflow for it. As soon as you begin editing a Zap, it turns off, which is a sort of “code it live on prod” I haven’t done since early PHP days :)

We don’t use it for truly production-critical things, but at some point integrations like this for CRM, support systems, etc need to be working and can’t just turn off for 30 minutes as we hand-copy a new idea into it.

How to have a “test endpoint” for a given Zap, or to promote/copy a change from one Zap to another — has anyone worked out a best practice for this?

A draft/publish workflow for updating existing zaps would be amazing, particularly for those that use webhooks (easily half of my zaps). Or something more specific, like using a single webhook to trigger more than one zap.

Can't you duplicate the zap, edit the dupe, then replace the original?

Sort of, but things like webhooks aren’t straightforward, and if you want to (for example) receive or send events with a test area of your third party sites, there’s not a clean way to toggle those.

(Note: I work at Zapier.) Thanks for the kind words, aaronharnly! You're right, we don't have a generalized way to do that sort of thing right now, but if you're triggering from a webhook, one option might be something like:

- duplicate your prod zap

- adjust the url to send a webhook to both your prod zap, and your (duplicate) dev zap (with just a single request): https://zapier.com/help/webhooks/#triggering-multiple-webhoo...

- once your changes are ready, "promote" it by turning the new zap on, and the old one off

There could be other workarounds for doing similar things, kinda depends on the overall structure of that particular Zap. Feel free to give us a shout anytime, and we could take a closer look at your specific setup! https://zapier.com/app/contact-us

Thank you!

There's a great episode of the Indie Hackers podcast with Wade from Zapier: https://radiopublic.com/indiehackers/ep/s1!7cb82

Was tempted by their product but I had to cancel my account as their IMAP integration was quite lackluster. Their support was prompt and courteous though.

When setting up a basic script to save attachments to a Dropbox, two basic thinks it could not do:

- Mark an e-mail as read

- Treat attachments individually (they all got dumped in a ZIP)

Zapier told me their gMail integration is better. But IMAP is the open standard.

Zapier is pretty cool, but every time I mess around with it, I end up being held back by only being able to use two-step apps. $20 a month is too much for just messing around with some personal stuff unfortunately. I's love a $5 personal plan that would allow me to do slightly more than the free plan, like conditionals and more than two steps.

Any guesses as to when they will be acquired and by whom? I'm wondering to myself if the founders are interested in this... I'd like to see the product grow, so I hope not!

>Any guesses as to when they will be acquired and by whom?

Given they are entirely remote and have a pretty unique culture, I doubt they get acquired unless it's done like Amazon did with Zappos (basically bought them then said "keep doing whatever you're doing, we'll leave you alone").

This is a great example of pitching your product from the "Jobs to be Done" perspective. More users (in total) will search "create Trello card from message in Gmail" or similar than generic platform integration.

Zapier is a killer app for non technical marketers.

It’s a killer app for technical people that have better things to do than build custom integrations for APIs.

Would you mind elaborating?

It lets you connect data via clean APIs without needing to code. So new website signups can automatically go to mailchimp, and zapier can text high priority leads to your best salesperson, and the lead can feed into a bi tool, etc etc.

I read Indie Hackers, and it seems every non-technical founder on that site uses Zapier, and Stripe.

For everyone talking about how great the product is, here is my challenge to you:

Zapier achieved $35M ARR because they sell $35M worth of customers every year and have a product that can back that sales channel up.

1. Their SEO is on-point (search <Insert SaaS 1> integration with <Insert SaaS 2> and you'll undoubtedly get a 1st page SERP from them

2. Their customer service is great (from what I've read) and appears to be very international friendly.

3. They have good freemium hooks that get you to upgrade and clearly have A/B tested this for conversion purposes.

TL;DR - Product alone doesn't get you $35M ARR.

I want to add that their attention towards developers is also amazing. I developed an integration for the product of the company I work for, and they were just great; the review process was painless and their APIs are very clean. My only complaint is that the docs can be somewhat lacking

This might be a silly question, but what the hell is a $35M ARR? The definition of ARR (Accounting rate of return) that I'm familiar with is as a percentage

Not a silly question. ARR is the sound a pirate makes when he finds $35M in buried treasure.

This is the best answer.

Annual recurring revenue.

I'm curious - given that they support hundreds of integrations across so many platforms, all with their own product roadmaps etc, how does Zapier actually develop and maintain these integrations?

What happens if Google or Trello or Salesforce changes their APIs? How does Zapier keep up?

We maintain the ones we built originally, but new ones are typically maintained by the app/service in question. At some point, the question of maintaining and updating "X,XXX" integrations with a comparatively small team does arise, so the idea of the partner owning and being responsible for the integration begins to make sense.

Thanks - so the integrations could also be maintained by the partners. That makes a lot of sense.

As a salesforce user, the price from essentials to premium is $50 a month, time a years worth, times the number of employees. My $240/year zap handles all the tasks, or as close to it, that premium offers. I'm saving THOUSANDS of dollars and my SF rep is hating me right now.

Plus Zapier support has been phenomenal, albeit slow as fuck using email, they are persistent and knowledgeable.

It will become harder and harder to have organic growth (vs artificial paid for) due to these crazy SEO tactics. If you are not linking already, next time you write a blog or article, make sure to link to all your sources and other interesting content for further reading, as it will become harder and harder to find stuff via search engines.

Zapier seems cool, but it doesn’t seem like it’s target is for “regular people” at $240 a year.

It's for business, $240 is nothing to most of them, it's small enough to get expensed.

I had a look at it in a previous workplace. For the volumes of data that we were looking at it came to a lot more than that. I Built an equivalent app to join two services in a couple of days.

Yeah, that makes sense.

Zapier makes more sense for many smaller tasks. A common example I'm familiar with is to connect the output of a Google form with a crm search and create a ticket to follow-up based on some filters. This is something that you can set up on Zapier in 10 minutes, and you wouldn't want the hassle of actually maintaining your own backend for something like that.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact