Our customer engagement and analytics stack at Aito.ai
Introducing the six most essential building blocks that drive our customer-facing operations and insights.
When chatting with other founders, we often compare our “stacks”. I know it sounds weird. But let me assure you the dodginess of this post is limited to that first sentence. Next, I’ll introduce six tools we use to steer the ship.
There are tons of excellent services available these days that make a startup founder’s life easy. And I only know a fraction of them. So here’s what we at Aito.ai use to drive our customer engagement and analytics. How does it compare to yours?
Segment
We use Segment to move the data around and get it to where it matters. Essentially, each of our online properties is a source (website, Console, tutorials, API docs, …) and feed the event data onwards to other relevant platforms below. Like this:
Segment has been significant in allowing us to deploy new tools quickly and without touching the code many times. For example, we had a trial testing onsite chat platform (now discontinued) and that all work by enabling the destination in Segment.
BigQuery
While all our production systems run solely in AWS, we use GCP BigQuery as an analytics data warehouse. All raw data from Segment sources goes to Big Query, and we run (some) of our periodic and ad-hoc analytics from there.
This area is in flux right now, as today the data is still too scattered around, making it sometimes a pain to produce necessary insights efficiently. Some stuff is in our production databases, some in AWS Cloudwatch metrics… Absolutely an area we are looking to develop next.
SQLBot
How could you not like someone who puts this in their FAQ?
SQLBot is a tiny tool for one purpose only: bringing SQL query results visible in Slack. But don’t let its “smallness” fool you. We use SQLBot every day, multiple times a day. There was a dashboarding tool in the past (Klipfolio), but our people rarely went there other than me making an investor report.
Then we changed things. We have two types of queries that SQLBot brings available to everyone in Slack: scheduled and keyword-based.
Scheduled queries are things that we look at every day, and SQLBot pushes them to chosen Slack channels on a schedule: how many new customers came in yesterday, how does customer activity look like yesterday, this week, this month and so on.
Keyword-based queries allow anyone to trigger queries with Slack commands. For example, the Slack command below returns the latest 20 new Aito instances created with their user details.
/sqlbot iI am looking to develop further these further. Slack command words in SQLBot allow variables, which means that we can bring the relevant data to our team in Slack instead of sending them out. For example, I envision a Slack command like this producing the instance details of a given user and their recent activity:
/sqlbot owner tommi@aito.aiGoogle Analytics
We use GA for all the “basic” analytics tracking traffic sources, page performance, and you name it. We are also curiously following what happens in the EU with Google Analytics after Austria’s decision and have relatively frequent chats on actions and alternatives. Ideas welcome!
Appcues
Now we get to the exciting part! We recently deployed Appcues to run our customer onboarding flows on our Console. Initially, we built 19 flows that demonstrate the Console’s features to the users. I must admit that getting all to work smoothly with our single-page app was way more work than we envisioned.
There are still tons of features in Appcues that we are not using, and for example, creating an onboarding “checklist” for users is in our plans.
Encharge
Last but not least, we have Encharge. It’s where the most important things happen — our customer messaging. We have built both our customer marketing messaging flows and the transactional emails from the system on Encharge. We were looking at Hubspot as an alternative but decided to go with a fellow European startup with super friendly founders who are always there to help.
In short, most of the important actions on our platform trigger a webhook to Encharge, which then deals with messaging flows. Here’s a super simple example that sends users a message when their Aito Evaluation is ready and notifies our Slack, too. (A completed evaluation is one of the best indicators that we have a likely paying user, so we track them extra closely.)
We are merely at the beginning with Encharge. Next, we plan to look at segmenting registered users based on their behaviour to provide users with better and much more targeted content and help.
That’s it! Do you use similar tools? Is there a tool that you love that would fit our stack? Ideas on how to improve? I’m available at Aito’s Slack community for great chats!
