This is the same Segment that started out as a single Javascript file that a couple guys threw together before rebranding right? I mean, they're so much more now (or so it seems) but feels like it was just yesterday... and I'm only 26!
Segment is a customer data platform (CDP) that helps you collect, clean, and control your customer data. Here’s more information [1] from Peter Reinhardt, one of the co-founders.
It's basically a really complex distributed pub/sub tool. It ingests data from sources, say Shopify and Heap Analytics, and can then send that data to 1 or more other supported platforms along with some user-defined transformations. Like a really fancy IFTT.
Genius move for Twilio, honestly. Segment can supply the central hub to tie in all channel spokes Twilio already powers (especially after the SendGrid acquisition from a couple years ago).
I can't help thinking Segment had much more room to run though...everyone I know who uses them loves them, and they were still only just scratching the surface of the addressable market.
Agreed really good fit. I bet a large % of Segment's customers were using Twilio endpoints. Really nice combination of a data hub / last mile across multiple channels...fully expect they will continue moving up the stack towards a CRM / Marketing automation solution.
Segment published a report on tool usage. "SMS and Push" was @ 12% and was not in the top-10. Even in the "SMS and Push", there was Braze/Iterable etc but no Twillio.
This is not surprising. The personalized SMS/push notification use cases (for which you need Segment data) are commonly implemented via marketing platforms (like Braze etc) as opposed to directly via Twillio.
I think this was the other way around. Twillio is powering a lot of these new age marketing automation tools. Instead of just being at the infrastructure layer, it wants to go up the stack and own the marketing applications. Sendgrid and now Segment acqusitions are part of the same strategy.
They did the same thing for the contact center (call center) software space. Lot of the cloud contact center companies were using Twillio and now they launched their own contact center product. That has been one of the key drivers in their 3x increase in valuation in the last 1 yr.
It's a bit like Amazon with it's branded meatspace products or AWS properties which can be directly sourced/inspired by by their own down-stream / customer data. They don't even need to look into traffic, just look at your best/top paying customers, and copy them
Sigh!
At least Twilio buys Segment, not just copies it I guess..
Who’s a likely candidate for CRM acquisition? Seems like the space is a bit crowded and Salesforce is the big dog and lots of second tier but who’s a good fit for Twilio?
There’s also Vero and Iterable that I’ve used. Iterable was pretty terrible and we ended up replacing it with in house tooling, Vero has been really great and we’re investing more in our Vero setup.
The other case is they didn’t want to deal with going through an IPO, got a great offer for their business and a major liquidity event for their employees.
Going public, despite profits made, really sucks the soul out of a company.
Twilio and Segment’s visions align pretty well so I think that played a role as well. Outside of golden handcuffs, a large portion of their employees are liquid immediately as opposed to an IPO with a 6-12 month lockup for employees plus the rise/fall of a stock price during that time.
All in all, I think they made the right choice to sell instead of going for an IPO.
If they could show strong growth continuing they'd have made more money with an IPO than an acquisition. At least that's what the comment is suggesting.
It's kinda nuts. They have some videos on startup school. They basically just threw spaghetti for months building random products for markets they barely understood until they landed on analytics and it just snowballed. This isn't even a brilliant idea. They entered a crowded market with a solution that was just a bit better than everyone else and now have a $3B exit.
I am currently working on a self-hosted analytics platform, and even though the space seems crowded, there are only a few good players. The thing is everyone started creating an "analytics platform", only that they just do it for money, thinking it's really easy to implement and don't care about the actual product, user experience or quality in general.
My point is, yes, the analytics market is crowded, but it's mostly crowded by low-effort players or archaic products. That being said, someone reading this will probably think it's a good idea to enter the market, thus leading to another below-average product.
Currently analytics is a fraud, analytics is useless, it is what you do about that data which is important. Most of analytics/ab testing software are just giving data (which is sometimes not even accurate) to customers, so that the people in the analytics team have a job to do. But the reality is that only a very small fraction of those teams know how to use the data to drive more revenue. So if you want to work in analytics there is this HUGE market which today is not addressed by anyone, its how to make more revenue from a website.
Segment is not even doing analytics, it has created its own market because no players in analytics has any interest if you want to make more money on your website. As website owners spend their time changing analytics tool, why not make a tool to make changes tool easier ? And that's Segment, they are even more useless than an analytics tool but well yeah nobody see it
> Currently analytics is a fraud, analytics is useless, it is what you do about that data which is important
This is like saying food is a scam and useless and that it's only important when you eat it.
I do agree that just tracking analytics, without doing anything with the data, is useless, just like storing food in the basement and letting it rot.
Disclaimer: I have never used segment before, nor I had the need for a similar tool.
Segment is targeted towards enterprises, their price is too high for the average business, plus most of those businesses probably don't even use any analytics tools (or maybe just Google Analytics and look at visitors number only), so the integration offering doesn't really make sense for them.
> As website owners spend their time changing analytics tool, why not make a tool to make changes tool easier?
As far as I know, changing analytics tools for the average business involves just changing a JS snippet on their site, nothing complicated.
As a personal note, my bigger goal is to make self-hosting and integration of web applications (in general) a lot easier for the average user, so anyone can spin up a self-maintained server running their favorite application with one click. So it's more about allowing anyone to use whatever tool they want, wherever they want, instead of just focusing on allowing them to select between a very limited selection of tools.
> This is like saying food is a scam and useless and that it's only important when you eat it.
Yes but everyone knows it for food, for analytics people expect it to solve their problem, which it will not do by itself
> As far as I know, changing analytics tools for the average business involves just changing a JS snippet on their site, nothing complicated.
Well not really, first except Session replay tools and everything recorded tool (ContentSquare, Heap) (and even for them it is far for being the case), changing analytics tool is a huge step, you need to recreate all your events, teach your users, install on your website (for an enterprise website, installing an analytics js tag is a huge thing)
> my bigger goal is to make self-hosting and integration of web applications (in general) a lot easier for the average user
To me self hosting and average user is not compatible especially in analytics where the average user barely knows how to use a computer.
Apart from that, it is not really clear to me what you are trying to achieve
> Apart from that, it is not really clear to me what you are trying to achieve
> To me self hosting and average user is not compatible especially in analytics where the average user barely knows how to use a computer.
This is exactly my point and my goal. First of all, when I mean average user I referrer only to those people that actually have a business/website to use those tools for, not to the average internet user. So, if they already have a website, they probably know at least how to purchase or even edit one (through code, WordPress or a visual editor).
My goal is that for those people, who are already technical enough to have their own website, "self-hosting" of their desired tool is just as simple as using 3rd party tools. It should not be harder to use self-hosted userTrack than it is to use Google Analytics. This is almost possible now, but the server/cloud providers are still working on providing APIs and services for making this more accessible.
Even now, you can go to certain GitHub repositories and click the "Deploy to Heroku" button, and with one click you have your own server running your desired application. The problem is that this is mostly limited to some hosting providers and to a very few applications. This should be possible to do, in an easy, secure and maintainable way for a wide variety of applications and hosting providers.
I reluctantly but wholly agree with everything polote wrote:
Currently analytics is a fraud, analytics is useless, it is what you do about that data which is important. Most of analytics/ab testing software are just giving data (which is sometimes not even accurate) to customers, so that the people in the analytics team have a job to do. But the reality is that only a very small fraction of those teams know how to use the data to drive more revenue. So if you want to work in analytics there is this HUGE market which today is not addressed by anyone, its how to make more revenue from a website.
*
I've searched high and low for analytics software that will let me
A) Measure the key things and come up with 80/20 Pateo Principle insights
B) After implementing changes, help measure the impact
Yet all analytics software seems designed to just capture stuff
That's all they do - 'capture stuff'
There are no provisions to leverage that data to actually make more money, or make your product better, or anything of substance
I think the main root cause for this is that computers and AI are not smart enough yet to know "what's better" for your business/website.
Each business is unique, offers its own (almost unique) products or services, has a (sometimes hard to understand) user segment on the market, and each visitor is unique too. It is very hard to make some generic software that automatically gives useful tips on how to improve conversions, it can not tell "hey, reduce your price by 10% and start advertising towards cat owners instead of dog owners".
What analytics software does is exactly the point B you mentioned: allow you to measure impact of changes and take (yourself) decisions accordingly. You can decide to A/B test two different pricing strategies and see (based on analytics) which one converts better, but the software won't tell you what the A and B variants of the pricing strategy should be. You can test hypothesis and take decisions based on the resulting data.
> There are no provisions to leverage that data to actually make more money, or make your product better, or anything of substance
I think this is due to, as mentioned by polote, the misunderstanding of the point of analytics: its purpose is collect and display data in such way that it can be analyzed (by a human or an external software) which can then take, based on this analysis and hypothesis testing, decisions that will lead to higher conversions.
It's the same with people saying ads don't work and are useless: if you just create a poor ad, with wrong targeting, that leads to a shtity landing page trying to sell a bad product, you probably won't give a positive ROI and then blame it on the ad platform because it doesn't tell you why things are not working, but there are so many factors outside their control, it would be impossible for them to give an answer. You can only get an answer by looking in detail at the data of each step and understanding WHY something happens or doesn't happen. This WHY is very hard to tell even by experienced humans, so it would be very hard to have a computer automatically answer this WHY for us.
There's a hierarchy of needs. Segment and a lot of related services are geared towards data collection and instrumentation. That part is relatively easy and has a very mature market. The actual valuable part is to gain actionable insights and that remains stubbornly difficult. From what I've seen, it's the kind of work that still requires an experienced marketer with good instincts.
The original product was really just a javascript utility allowing you to instrument your site and funnel data to other analytics platforms like GA or Adobe. The industry was coalescing on another approach (data layer) which still exists and probably has wider adoption. Segment used their position as a middleman to add more and more value as they matured and acquired customers.
I follow(ed) Segment’s GitHub branches and always felt they adopted the latest and best front end tooling. I usually let them being the testing ground at a real company in between the stuff they developed themselves.
This goes from using Metalsmith early for their single page web sites, some coding practices (mono vs micro), some CSS frameworks and design patterns and other library choices when there always seemed like a barrage of new choices (like css in JS ) and React.
Anyway I much appreciate their efforts in this area as inspiration and real world examples published on OSS sites. Even if it was simple stuff like their marketing websites.
They clearly had respectable and internally influential talent working in this area.
It’s been a while since I’ve explored their GitHub repos for fun but the I still respect the stuff I learned.
Most of that is Ian Storm Taylor. This is what happens when software is treated as a first class citizen and the co founder(s) are technical. It's a specific class of companies like Stripe and Square. The business itself is treated as a software product, internal influences external.
I believe he/she means that the way you practice is the way you play.
If you put your best effort into the things that "do not matter" (e.g. hobby, open source work) then the discipline will show in the external displays of your work (products and services).
A fantastic outcome for a fantastic team and product. I've worked closely with both Twilio and Segment for a number of years as the internal driver behind large enterprise deployments.
I first encountered Segment when they were a young company and would frequently chat directly with Ian if I needed support or needed feedback. He was always gracious and humble, despite no doubt being ridiculously busy.
Twilio has gone from strength to strength - they have an agility that is only possible from a company that is willing to deeply listen to the market and stay one step ahead. Very excited to see where this leads!
Last round's valuation seems to be $1.5B, so it looks like the acquisition price is 2x that so everyone hopefully walks off happy. I hope Accel, GV, etc. didn't have a crazy liquidation preference.
Liquidation preference usually just bases itself on the amount of money invested -- If Crunchbase is right, they've "only" raised like $280M, so even with 3x liquidation preference, they're well clear of the $850M number that would trigger any of the annoying rights that strip stock from common sharedholders. Everyone will have done well on this deal.
I'm not sure that it's Jeff Lawson as much as George Hu. 13 year SFDC veteran and credited by Benioff for many successful initiatives in his cloud book.
Separately, when George joined Twilio, they didn't really have a sales team. Started working on contract terms, raising prices and a host of other things. He's very systematic.
Hopefully this means Segment dropping their prices - $120p/m for 10k users is out of the range of a lot of potential customers, and changing around an established analytics setup later on is always a pain.
You have to justify the valuation somehow. Of course, whether that validation is truly justified is another matter and I guess the market will decide in due course.
Offtopic: what's the best self-host tool for Intranet analytics? Would be best if can track user keystroke/mouse without predefined events, like what heap.io do.
Wow, congrats to the team @ Segment. This is very well deserved - they created this category. Also, looks that Twillio is investing big time in owning the marketing stack.
Really happy for the team and especially for the founders. They proved to be very technical and assembled a truly impressive engineering team, culminating in a huge exit.
The founders deserve every penny of that $3.2b and deserve the very nice lifestyle that will afford them. A great story that will inspire others including myself.
Really happy for the team. Huge fan of Segment from the start, and Twilio is slowly getting itself into the middle of everything. (I realized only recently that they've owned Authy for years following the recent rebranding.)
I find it interesting that Segment (and thus the entire downstream ecosystem) has gone this far while staying strictly oriented around users and accounts, without first-class support for other entities -- namely a notion of workspace/team/contract/opportunity/etc. "Groups" was sort of a false start that didn't get a new direction. There's still a ways to go connecting all the dots between how companies structure their data and how all third-party products are prepared (or not) to mirror that structure.
I feel the same way but Salesforce has become such a complicated beast that I can only assume a simple no code solution like Zapier looks appealing as an acquisition target.
A few reasons that all connect to each other. The biggest, most foundational one is that spammers are highly sophisticated criminal organizations, and will constantly be trying to figure out how to leverage your product for spam.
If they are successful, it is extremely costly for you (primarily in terms of IP reputation, which is very difficult to maintain) and cheap for them.
If they are unsuccessful, well, they'll keep trying and stopping them continues to be costly for you and they only need to get lucky once every so often for it to be worthwhile for them to continue.
Exacerbating all this, it's extremely commoditized, so even fairly large customers aren't going to make you that much money, especially given all the hassles of maintaining it.
I sort of disagree that it's commoditised, since the most popular email API would have the moat of being included in every spam filter whitelist. By choosing Sendgrid, paying them for the service and getting their IP ranges you guarantee that you get the greatest chance of getting your email delivered. That commands a premium.
Yeah you can go for Amazon SES or Mailgun or something, but they are used to spam too, Amazon SES to an even greater extent.
You are also guaranteed to get whatever the latest email innovation of email delivery they come up with as one of the early adopters. If they decide to deprecate SMTP and build something more secure, private, easier to authenticate and identify, you get that for "free" without having to change your code.
I see Sendgrid as a relationship maintainer between developer and Gmail/Outlook/Yahoo/long tail of other email providers. You pay them and they make sure your emails show up in those inboxes.
Sendgrid has issues with hacked accounts, which is probably their biggest struggle right now, but even then they have things like reputation. A hacked account would very quickly dip their reputation and be excluded from delivery anyway, without affecting the platform at large. They push 2FA tho, and I think some sort of automatic api key rotation system would be a nice remedy too.
Ultimately, Sendgrid maintains risk on per-account basis. Email providers understand that too, even though SMTP as a protocol and spam filters don't necessarily reflect that. Sendgrid as a platform is never under threat, individual Sendgrid accounts are.
Was a segment user around two years ago. Felt like a software that is made at exactly the moment when it should have, for a really hungry audience. Their success never surprised me. They also had a capable alternative- mParticle. Not sure what are they upto.
Segment (and any other SaaS that wants to own the event firehose) has pricing issues as the customer scales. As a customer you are charged on data volume[0], but your data grows at the rate = (Users x Engagement X Features X Data Sources). That rate can easily be the square of your revenue growth.
This pricing issue makes it easy for Segment's fast-growing customers to justify an in house replacement. This puts huge down-pressure on their "revenue retention" rate and makes it hard to raise more VC or go public.
[0] - Segment technically prices on user volume but in reality it's a mix of users + objects depending on what your sources are
This goes against my intuition in several ways, so I'm curious if there are examples to discuss.
- Scaled deployments have Segment connecting many systems run by many teams, so the cost of coordinating a replacement can go up just as fast as Segment's.
- High growth companies have fires everywhere, including other infrastructure, so it's not obvious that Segment makes it to the top of the list.
- I expect the bulk of Segment revenue is not from companies with "high growth" by HN standards but scaled enterprises with high complexity and lower growth. These can dominate net rev retention even if there are smaller companies growing faster.
Again, no data to support the above, just a different narrative. Anyone have facts?
So what’s the end state solution? The tools change and/or mature slower than a fast company will grow. Where are you going from here?
Are you eventually running your own cluster? Or, are you paying per Gb per second how GCP bills with decoupled compute / memory / storage? Or maybe with a little bit more overhead with the orchestration cost on top of consumption with how azure bills for ADF.
Just from reading blog posts, Spotify’s move was interesting, regarding their annual “songs you liked” at the holidays, where they sort of ran a hybrid.
Sorry to digress but what is the end state. Is there just a number where it makes sense to hire people and build and/or shift vendors? And if yes, what vendor is “after” stich? Astronomer maybe. Sorry if that’s basic.
Solution seems obvious, volume discounts in one shape or another.
SaaS companies pricing is often geared towards startups that "can't afford" to control costs, i.e. developer time spent on minimizing cost has a higher monetary and opportunity cost than ignoring those costs in the short term, even if they're piling up at $50/mo for every auxiliary service.
But that strategy does not account for those same startups after they grow enough to take a breath and start optimizing costs.
The end state is you DIY. Segment isn’t doing anything secret/magic, it’s just well executed and super convenient.
The “how” question has changed every 3 years in the last 10 that I’ve done data work. That trend is driven primarily by data scale and partially that analytics and ML access patterns have evolved a lot.
We are building Grouparoo (https://www.grouparoo.com) because we think you should be able to do this yourself and own/control the data - without the incentive structure being messed up from the cost pressure.
after a certain sales price, there are very few potential buyers, and unclear if the usual top acquirers would want to pay the premium (google, ms, salesforce, ...). likewise, always risk of things going sideways wrt ipo. $1-3B exit in this situation makes a lot of sense.
more interesting to me is twilio seems to be such an operationally awesome company, going through each phase of growth so well
RudderStack (https://rudderstack.com/) founder here. We have a different pricing vector (event-volume vs monthly-users/MTUs of Segment) so it's not directly comparable. Having said that, RudderStack can be much cheaper in two scenarios
1) Low # of events per users. For example, if you have a lot of anonymous users who don't do many events and leave. In this case, you may have a lot of MTUs but low number of events. True for a lot of consumer businesses.
2) You have a huge scale (10s to 100s of millions of events per day). In that scenario, we support deploying enterprise
RudderStack inside your VPC but still managed by us. In this case, we don't have to incur infra/data-transfer cost (and charge 80% margin) so can be much much cheaper. We have a couple of customers in this model. Some choose this for privacy/security reasons.
In normal scenarios too we can be 50% or more cheaper. Unlike Segment, we don't store any data (which is a major cost) so our COGS are quite lower. We use your warehouse for all the CDP features.
Which means your platform can't replay historical data to new sources, which offers a LOT of value for both onboarding new platforms as well as testing tools before signing a contract (since it lessens the cost of getting actual data into a tool to see if it's a good fit).
> Segment is an American customer data infrastructure company based in San Francisco, California. Its software allows enterprises to combine the collection of data from their customers, whether through a mobile app, by e-mail or in a store. Segment was founded in 2011 and accepted into Y Combinator in the same year.
So nobody feels comfortable talking about what they do.
> Segment | Customer Data Platform (CDP)
> segment.com
> Segment collects user events from your web & mobile apps and provides a complete data toolkit to every team in your company. Marketing Single view of the customer.
