Google Allo

» see allo.google.com as it lived, on the wayback machine

~ the last words ~

Google Allo was announced at Google I/O on May 18 2016 and shipped on September 21 of the same year. The pitch was that the new Google Assistant would live inside the chat thread next to your friends. You could @-mention it to ask a question or have it draft a reply for you. It could also pull up restaurant suggestions in a thread with someone you were planning to meet. Allo was supposed to be the launch vehicle for Assistant in messaging the way the Google Home speaker was the launch vehicle for it in the kitchen. The encryption story was Signal-protocol-class but only in an opt-in incognito mode, which was already a controversy on launch day.

The early download numbers looked respectable. Five million installs in the first week, ten million by mid-December 2016. Then the curve flattened hard. By February 2017 Allo had dropped out of the Play Store top 500 free apps. WhatsApp had over a billion monthly users in the same window. Facebook Messenger was at a similar scale. The product Google had built was a fourth or fifth option in every market where the social graph already lived in someone else's app, and there was no good reason for anyone to migrate just to talk to a chatbot.

Internally Google had also already started building the RCS-Universal-Profile push, which would put the carrier-class messaging features Allo offered into the default Messages app on Android. That made Allo the redundant project. In April 2018 Google publicly paused Allo development and moved most of the team to Messages. The shutdown announcement came in December 2018 with a March 14 2019 service-end date. Users got a couple of months to export chat history before the lights went out.

i never installed allo. by 2016 whatsapp had already won the messaging market in turkey by such a margin that nothing google shipped would have moved a single contact off it. i remember reading the launch coverage on android central and the engadget review of the assistant integration. it sounded interesting and had no chance. google has shut down enough messengers since gtalk that this one was easy to skip without remorse.

~ the record ~

~ the assistant integration was the whole pitch ~

The reason for Allo to exist was the in-thread Google Assistant. You could @google in a conversation and pull up flight info or restaurant suggestions without leaving the chat. The assistant could also propose responses based on the message you had just received, including reactions to images. This was four years before ChatGPT and felt like the right direction. The execution problem was that the rest of the messenger around the assistant was thinner than what people already had. Voice messages and a web client both came late or never. By the time Allo had basic feature parity with WhatsApp, the user trial window was closed.

~ the encryption controversy ~

Allo shipped with end-to-end encryption only in incognito mode. The default mode stored messages on Google's servers so the assistant could read them and learn from your conversation. Edward Snowden tweeted on launch day that nobody should use it. The Verge ran a piece pointing out that the security story was strictly worse than what Allo's own team had originally promised. Google's defense was that the assistant needed plaintext access to function. Both things were true and the resulting product was a privacy story that satisfied no one.

~ google's messaging graveyard ~

By 2026 Allo sits in a long line of Google messengers that did not work. Google Talk launched in 2005 and was eventually rolled into Hangouts, which then split into Chat and Meet. Allo and Duo arrived as the next attempt and ended up folded into Messages and Meet themselves. The pattern is that Google ships a new messenger every few years and rolls the engineering into the next attempt after the network effects of the incumbent refuse to budge. Allo's contribution to this cycle was the assistant integration, which moved into Messages once the rest of the app was abandoned.

The deeper problem is structural rather than product. Messengers are won by network effect, not by features. WhatsApp won the global market in 2010-2014 by being free and cross-platform when SMS still cost money in most countries. Once the graph was on WhatsApp no amount of better technology was going to move it. Allo arrived at the moment when the war was already over, and Google has not really tried to relaunch from scratch since.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »