I really didn't like the WhatsApp approach on this. Basically only partnering with few big vendors like twilio and then asking people to build apps via them (by basically paying for it).
Other apps like telegram allows anyone to directly interact with them
FB paid something like 20bn for Whatsapp almost 6 years ago. Something like 4bn was paid in cash. How do you make that money back if: a) users do not pay a subscription and b) ads are not printed in the app? You leverage the app's immense user base asking businesses to pay for API services so they can implement chat-bots, notifications, in-app tools, etc.
Why using res-sellers, like carriers use re-sellers to sell SMS integration? Distribution and fair-use control. First, companies like Twilio are the natural partners to sell one more channel to customers that already integrate to SMS. Second, partners will help enforce and control anti-spam efforts. If opening Whatsapp API turns it into a spam jungle like email or sms, it will certainly lose its appeal and users will start switching to the next new spam-free chat app.
They don't necessarily need to put ads in the app. They can simply transfer that personal information to their larger people profiles making their existing ad networks more valuable. It could additionally be simply a play to keep someone _else_ from getting said data. (I think your argument still makes sense, just another angle on that part).
The privacy law implications (GDPR, the new California law, etc.) would probably crush Facebook as a company if they started doing that.
They already paid over 100 million for just making false promises when they bought WhatsApp, now that the EU has sharpened their blades to take on big data farms like Google and Facebook, doing something like that would be suicide for FB. Any additional data farming would probably attract auditors and lawsuits that Facebook can't get out of so easily anymore.
Facebook would need to put a lot of work into following the GDPR (because someone using an Asian VPN is still protected, so IP based geoblocking won't work) to prevent a fine. It would be a very risky move.
I don't think Facebook is going to sell private WhatsApp information any time soon.
I'm mostly speculating but, they wouldn't be selling the data, they would be selling access to their ad network that has models built off that data used to match ads to people. If they have a high(er) conversion rate, they can charge quite the premium for this information. Similarly, if they could merely keep people in their platform (WhatsApp), nobody else has that information -- same effect.
(Also speculating) if that's more or less how it works, and an ads conglomerate like Google or Facebook is worth hundreds of billions, then even a fine of a billion dollars wouldn't be enough to deter them from this activity.
None of this is meant to stir up controversy or be negative or anything, I just think its an interesting mental exercise to consider this scenario and these motivations.
Well probably just a liquid source of funds transferred directly to an account vs. some other financial instrument. No one is going to do a deal like that in physical cash.
I registered for the WhatsApp Developers Programme a few months back and was told that 'It's not ready yet, but we'll contact you when it is'. Yesterday I got an email telling me that basically I have to partner with (and pay for) services from Twilio or couple of others if I want 'developer' type access.
It seems that independent developers will never get legal access to integrate automation or build CRM type apps using WhatsApp, which sucks considering how prevalent WhatsApp usage is.
WhatsApp founders always promised that there would be no advertising on WhatsApp. Yes I know they are gone. I confess I am not on Twitter and use Facebook once every 2 to three months. I am in about 6 WhatsApp groups and every evening I sit down to catch up on what's been happening by reading the group posts. I think the success of WhatsApp is that there literally no bots unless you sign up for one. Instead of selling adverts charging organisations to develop WhatsApp bots just might be the way to keep WhatsApp free of rubbish.
Other advantage of WhatsApp over social media is you don't know is something is viral because you can't count the likes.
Twilio's bread and butter is telephony developer relations. Why would WhatsApp get into this market when they can outsource it for free to Twilio who has the product and documentation nailed. The support however is a different story. Once the demand is high enough, WhatsApp are free to revisit. It also keeps WhatsApp low (relatively speaking) cost to run.
The type of messages that can be sent over WhatsApp via Twilio are so restricted and expensive that it was not worth integrating when we looked at this at my work.
Even with Verizon's proposed A2P charge per SMS, it is half the cost to use SMS for conversing with customers compared to WhatsApp. Said surcharge has been postponed repeatedly, meaning SMS is an order of magnitude cheaper to use than WhatsApp
I use Twilio to automate SMS messaging with users. Some of Twilio's sales people told me about this a few weeks ago, saying I'd be able to seamlessly connect with people through SMS, Whatsapp, or however they like to communicate.
I think the idea here is that if you're building a Whatsapp bot, you probably want the other types of bots that Twilio offers anyway and it's easier if the services are bundled together under a single vendor.
Whatsapp doesn't really expose an API. They require you to maintain docker containers that binds to phone numbers. It has issues. Generally I agree but their business model requires "good behavior" by the partners.
WhatsApp (and thus Facebook) has three ways to monetize the service: (1) Charge everyone a small amount; (2) Charge a few users -- businesses mostly -- a large amount; (3) Sell and monetize your data with ads.
WhatsApp's founders were inclined to go with #1. Mark Zuckerberg appears to prefer #2 and #3.
Facebook (same company) also allows anyone to build a bot. Maybe they wanted to monetize Whatsapp this way, but thought it would be less controversial if it's through an intermediary.
This is nice but quite expensive. Up to 0.09$ per message inbound or outbound!
Pricing for Template messages
Outbound template messages to WhatsApp destinations have two different components:
Edit: Here's the price breakdown. [0]
> Twilio charges a flat fee of $0.005 per message for all WhatsApp messaging. This applies to all incoming messages, outbound Template and Session messages.
> WhatsApp also charges a per-message fee to send outbound Template messages. This fee may vary, depending on the destination country or region.
The second fee varies wildly up to 0.09$ in Germany! [1]
Hey all, I'm the engineering manager for teams at Twilio responsible for WhatsApp integration, among other channels. We're more than happy to answer any technical or general questions -- fire away with replies. Will keep an eye on it today.
The WhatsApp Commerce Policy states that businesses may not transact in the sale of real, virtual, or fake currency. Keyword: sale.
Does that discard the creation of remittance apps? That's not selling, that's just sending money between parties. I was thinking on sending money around the world using WhatsApp, is that allowed?
It seems there is no support for letting a user share their location over WhatsApp through the API at this moment - I was wondering why that is and if you're expecting that that will change soon? For the rest I've experimented with the system and it's been working quite well, thanks for the work!
I'm about to release a bot for Whatsapp for NanaGram (https://nanagram.co). Amped!
The user experience of SMS-first products is novel and fun. There's something magical about snapping a photo and just texting it to a unique NanaGram number powered by Twilio. It takes just a couple seconds. I'm excited to deliver the same experience through Whatsapp to customers around the globe and people on wifi.
Building NanaGram SMS-first has been a challenge. There have been all kinds of gotchas:
1) AT&T started blocking any number permanently across its network last January if it contains a hyperlink or email address, even if the customer interacts with the number. I had to create separate messages just for the 12.5% AT&T users.
2) Twilio only works when you have cell service. If you're indoors with wifi only, the text won't go through.
3) Images are downsized in an unpredictable way by the carriers. If you're sending 1 or a couple images, they're pretty good quality. If you send ~5+ photos, they get downsized pretty far and luckily they're still good enough for 4x6 photo printing. Certain carriers like cricket are super aggressive with downsizing.
4) When Apple users with iCloud enabled share more than 1 image, it generates a sharing album which has been a beast to support. Then there's the added challenge of Apple's recent move to HEIC/HEIF images.
5) Twilio doesn't support group threads (WhatsApp doesn't yet, either)
... I could go on and on.
Getting approval from Whatsapp took a while. I first applied about a year ago and didn't hear anything, then applied again and was approved a few weeks ago. I'm guessing the approval layer is to ensure only high-quality bots make the cut. They also have an approval layer for all of your messages sent outside a 24-hour window of the last user interaction. The approval for that took a few days.
Integration with WhatsApp has been surprisingly easy. I basically just had to change "+1{number}" to "whatsapp+1{number}"
> 1) AT&T started blocking any number permanently across its network last January if it contains a hyperlink or email address, even if the customer interacts with the number. I had to create separate messages just for the 12.5% AT&T users.
I don't follow - they blocked any number which replied with a hyperlink or email?
If you send a message containing a hyperlink or email address from a Twilio number to an AT&T number, AT&T will block that Twilio number across its network without any notification. This happens even if you have an ongoing thread going with the customer. Even if after the block happens the customer pings the bot with "START" or any message, AT&T silently ignores it and continues to block the number. The customer can continue to send messages inbound to the number (and I believe those get received by Twilio but can't quite remember) but any outbound responses are blocked by AT&T from making their way to the customer, even though they're marked delivered by Twilio. This was the case as of Feb 2019. It could've changed since then and I haven't checked recently.
I bet that AT&T can detect "non-real-user" phone numbers, so their hyperlink filter only acts on those numbers.
I have been using Google Voice as my secondary phone number for the past 5 years, and some services manage to detect that it isn't a real phone number and refuse to work with it. More specifically, it usually happens when I try to use 2FA over SMS (note: i know that 2FA over SMS is not considered secure, but some of the services I use only offer this as their 2FA, and it is better than me not using 2FA at all). It doesn't just fail to deliver the message, the service straight up tells me to enter a "real" phone number.
I've noticed this as well. Services like Twilio provide an endpoint to check the carrier. For my Google Voice number, they return "Google (Grand Central) BWI - Bandwidth.com - SVR."
No, they don't. I haven't tested this in a while though.
Edit: Oh, I think you may mean a support response. Twilio helped me dig into the issue to great lengths. Ultimately, it was out of their hands. I think it's a lazy decision on AT&T's choice. Seems like a very rudimentary spam filter. You'd think they'd at least whitelist numbers customers are actively texting, especially if they don't contain keywords like "stop" or "spam."
Chatbots on Whatsapp are kind of hard from a UX point of view because the platform don't support rich elements like buttons or webviews (unlike Facebook messenger). I hope Whatsapp adds these kind of features soon, otherwise users will get frustrated when interacting with bots and never use it to talk to businesses again. And no, AI/NLP is not the answer to this because users are lazy and prefer to tap on buttons than to write 20 characters...
I know this because I'm the creator of Botonic (a React-based conversational framework) and we've been building conversational apps on Whatsapp and other platforms for years. By the way, if anyone is interested in access to the Whatsapp Business API beta feel free to drop me a line at eric@botonic.io
“... rich elements like buttons or webviews (unlike Facebook messenger). I hope Whatsapp adds these kind of features soon”. I on the other hand hope they never add these kind of features. If I wish to converse with bots, spam, ads etc. I could always reinstall fb messenger or enjoy the pop ups on every other website. I know I’m being naive but I hope whatsapp is left alone by fb.
Simplicity of features != UX Simplicity
It's way more simple to tap on a button than to write complex sentences. We've studied this, when users have both options available (buttons and input text), 80% go for buttons.
wow this is by Miguel Grinberg who is the author of some of the best Flask tutorials and books around. Nice to see Twilio is attracting such high quality writers for their blog
Yes, this is true. I will be involved with the Twilio blog as part of my duties, by writing original articles and also by working with other contributors to get their posts published.
oh nice! Long time fan of your work and I people get us confused sometimes because PacktPub chose the title 'Flask by Example' for a book I wrote with them which is too similar to your blog posts.
I nearly wrote an article for Twilio but didn't complete the process. But have written similar posts e.g. [0] and work with other writers who also do this form of instructional technical writing. Would love to chat sometime if you have time - contact info in my profile
I am divided about this. In my home country WhatsApp is the Internet (along with Facebook and Twitter). Mobile service providers sell WhatsApp data bundles which are orders of magnitude more affordable than "normal" data bundles. So we all communicate via WhatsApp. We use WhatsApp groups and it works well. News, videos, pictures everything I get by WhatsApp from contacts. IMHO it works well because there is not advertising and no bots spamming us. Some of the groups are noisy but you can mute.
I am divided because I like WhatsApp as it is right now. But we are not paying for it and the money has to come from somewhere. Using third party providers like Twilio is probably one way to pay for it. I am divided because having to use Twilio is likely to freeze out developers from my home country. C'est la vie.
"With location sharing data, 60 billion messages sent per day and access to users' entire contact lists, Facebook has access to a ton of personal information – all uploaded and saved on its servers."
Telegram bot writing is silly easy. If you use the python module you can basically just write the whole thing as a set of functions in a file (with decorators that watch telegram for triggers) and one function that automatically polls and does all the responding at the end.
I managed to bang out a few when I had a day off a few months back. Very satisfying.
Writing a Telegram bot is easy yes but writing a good one isn't. There are some very tricky things involved which you basically can not know and wont find in the Documentation. It forces you to run into problems first and then look for solutions. Especially if your bot gets decent amount of traffic something in the range of 1msg/s and above.
The people at https://t.me/BotDevelopment (unofficial) are your best chance for support because Telegram (bot) support is a joke.
I ran it off of a $5/month 'droplet' from Digital Ocean. As they are the only hosting provider I have ever used I can neither endorse nor critically review the service. I have, however, had no issues with them and find their various documentations floating around to be very useful and highly accessible.
Definitely want to agree with the point about documentation. I pretty much always end up using it if I need to set up docker on a new machine, or do other web app related ops tasks.
In the 90's I built an IRC chatbot, at the time users were naive enough to think it was a real person, so I kept adding conversation lines.
Over time it grew into thousands of lines and some people came back to the channel just to chat with it. My friends and I had some good laughs, good times heh.
Other apps like telegram allows anyone to directly interact with them