The decline of Twilio's position has been as pretty clear for the last 18 months now, and has been a topic of conversation longer than that. Twilio never had the margins or control of their environment to nearly the degree you need to have in order to maintain software monopoly levels of growth over time.
Most of Twilio's business has been built on reselling network access purchased via SMS consolidators. These are companies who, decades ago, got their hardware installed inside the networks of the major phone carriers. This allows them direct access to send/receive SMSs. Twilio never really tried to own the network layer and these companies continued to demand higher and higher tolls for access. Short codes are a very good example of this. Twilio spent a lot of time and money trying to sew up access to those short codes.
On top of all that, high volume customers would move directly to these consolidators. Sometimes Twilio would keep some portion of the business if the sender used a round robin model to distribute sends, but often they didn't. OpenMarket and a handful of others were the goto providers at scale.
Add on to this the overall decline of the utility of SMS. Even as SMS volumes increased, they have not increased at the same rate as messaging overall. ie: other channels like iMessage and WhatsApp continue to pull volumes.
Cue Twilio's attempt to go up market. Flex and the purchase of Sendgrid were the best examples of this. Could CPaaS work for Twilio as a business model? I think we've seen so many fits and starts with the Flex product line and the integration of Sendgrid that it seemed like perhaps buying their way in to that market wasn't panning out.
Finally, as seen in the last year, the culture just seems to have gone off the rails at some point. [1] I don't know Jeff, but my respect for him is sky high. I've read all his writing over the years and he's been hugely influential for an entire generation of founders.
Unfortunately this seems like a bit of a hasty departure. I hope it isn't, but the choice of replacement CEO screams caretaker-CEO rather than shrewd strategic move.
Another commenter mentioned Stripe [2] and how they get a lot of credit for advancing DevTools. I agree that Twilio did it first and I think paved the way. Jeff Lawson and Twilio deserve more credit. "Ask your developer" was not a small thing at all.
I also think Stripe very likely is in a very similar position, but they are earlier in their cycle and they've avoided the scrutiny of being a public company. Credit card processors are gatekeepers and fully dominate in their markets. Stripe serves a purpose for them and dramatically improves access to their networks via great tooling and abstractions, but there is still a fixed cost toll to pay and what that toll is cannot be static for interminable time. Is Stripe in the "Commerce" market like Twilio is in the "Communications" market, or are they in the "payments" market, like Twilio is in the "SMS" market? (or somesuch)
In the early days of Twilio, carriers laughed at what we were doing. It didn't help that most of the people at Twilio pre-2014 didn't speak much "telecom." But to your point, it was tough to rely on these carriers who thought Twilio would be gone in a few years so they didn't put much stock in ensuring the infrastructure was redundant and robust.
I was on several calls with providers begging them to put us on a more robust network. Especially in years when there was an election. In those early years, we had one customer doing more minutes than our entire customer base. In fact, at an early Twiliocon (pre-Signal) John and Jeff had to help build out an entirely separate environment for them because it was dragging everyone else down.
My point is that the early days were definitely the "Draw the Owl" days. We were figuring out things on the fly. Carriers were dubious and it was tough to get things working consistently. However, by 2015 those same carriers were developing APIs of their own.
Twilio changed the game for the telecommunications industry. Two years before joining Twilio I was putting in large-scale Avaya systems. The fact that I wrote four lines of code to make a phone ring blew my mind. Twilio and Jeff created something special. It's sad to see his run come to an end, but every company needs to adapt to the changing times. This is just Twilio's time to adapt.
> In those early years, we had one customer doing more minutes than our entire customer base. In fact, at an early Twiliocon (pre-Signal) John and Jeff had to help build out an entirely separate environment for them because it was dragging everyone else down.
As an early large customer, I wonder if that was me... sorry for messing up everyone else ;) I do remember some issues at times, but nothing we didn't see with other voice and SMS providers. Well, other providers didn't have an online log browser, so I never had trouble with browsing logs their logs.
Twiml was much easier to use than the other 'programatic voice' apis at the time, and I never had time to figure out how to run calls on sip directly. As I recall, it didn't take much time to setup voice calls, and then boom. It was helpful that quality was good and support was helpful when there were destinations with poor results.
Fwiw I think a key difference in stripe vs twilio in your example is that stripe has 2 levers that twilio doesn't have good analogs to:
- stripe can be a "financial home" for many of its customers and build higher-margin pockets of software functionality, using easy payments as an on-ramp. See: stripe billing, radar, sigma, connect, etc.
- stripe is more likely to be able to "swap out" the credit card rails than twilio would be able to "swap out" sms. If the financial ecosystem evolves to make eg debit cards or any other kind of payment easier, stripe can simply add them as payment methods, as they already do for non-CC payment types (most notable outside of the US). So stripe is already at the abstraction level of "move money from customer to me" instead of "run their CC".
I think those are good points and valid, but Twilio definitly had equivalent arguments in their own product.
Encumbents manage scale very very well and historically they tend to come to a market late. While Stripe is rushing up the value chain, the issuers are just slowly lumbering around.
Billing, Radar, Sigma, Connect are all very well built products which drive usage of their core revenue generator. I suspect the revenue mix at Stripe is still largely payment processing. Voice minutes and SMSs on a carriers network, or authorizations and settlements on an issuer network.
I am not saying Stripe is doomed, far from it, but it's a non trival problem to solve over the long term and the default state is attrition. I think Twilio is a very good example of how difficult this can be.
> On top of all that, high volume customers would move directly to these consolidators.
This is pretty key to understanding Twilio's prospects as a business. In my experience, they are fantastically easy to integrate at first.
But they come with a cost that blows up your financial model if your product is any kind of high-volume use case, or if SMS is factored in as a direct cost and you're talking to investors about unit economics. At some point you have to get off of them and onto a cheaper-per-SMS solution, even if it costs more dev hours to maintain.
Most of Twilio's business has been built on reselling network access purchased via SMS consolidators. These are companies who, decades ago, got their hardware installed inside the networks of the major phone carriers. This allows them direct access to send/receive SMSs. Twilio never really tried to own the network layer and these companies continued to demand higher and higher tolls for access. Short codes are a very good example of this. Twilio spent a lot of time and money trying to sew up access to those short codes.
On top of all that, high volume customers would move directly to these consolidators. Sometimes Twilio would keep some portion of the business if the sender used a round robin model to distribute sends, but often they didn't. OpenMarket and a handful of others were the goto providers at scale.
Add on to this the overall decline of the utility of SMS. Even as SMS volumes increased, they have not increased at the same rate as messaging overall. ie: other channels like iMessage and WhatsApp continue to pull volumes.
Cue Twilio's attempt to go up market. Flex and the purchase of Sendgrid were the best examples of this. Could CPaaS work for Twilio as a business model? I think we've seen so many fits and starts with the Flex product line and the integration of Sendgrid that it seemed like perhaps buying their way in to that market wasn't panning out.
Finally, as seen in the last year, the culture just seems to have gone off the rails at some point. [1] I don't know Jeff, but my respect for him is sky high. I've read all his writing over the years and he's been hugely influential for an entire generation of founders.
Unfortunately this seems like a bit of a hasty departure. I hope it isn't, but the choice of replacement CEO screams caretaker-CEO rather than shrewd strategic move.
Another commenter mentioned Stripe [2] and how they get a lot of credit for advancing DevTools. I agree that Twilio did it first and I think paved the way. Jeff Lawson and Twilio deserve more credit. "Ask your developer" was not a small thing at all.
I also think Stripe very likely is in a very similar position, but they are earlier in their cycle and they've avoided the scrutiny of being a public company. Credit card processors are gatekeepers and fully dominate in their markets. Stripe serves a purpose for them and dramatically improves access to their networks via great tooling and abstractions, but there is still a fixed cost toll to pay and what that toll is cannot be static for interminable time. Is Stripe in the "Commerce" market like Twilio is in the "Communications" market, or are they in the "payments" market, like Twilio is in the "SMS" market? (or somesuch)
1: https://news.ycombinator.com/item?id=36382361 2: https://news.ycombinator.com/item?id=38913134