Kippino, the KPI bot

This is a Slack bot that asks people about KPIs

Check out the announcement post on our tech blog.

Building

$ npm install
$ ./node_modules/typescript/bin/tsc

Running

To run the bot you'll need:

a Service Account auth JSON for the Google Spreadsheet API, follow these instructions to get one.
the spreadsheet ID: it's the alphanumeric string after https://docs.google.com/spreadsheets/d/ in the URL of your Google Spreadsheet
a token for your Slack bot: follow these instructions to get one for your bot

Then you can run the bot by passing the above with env vars:

$ AUTH_JSON=./xxx.json SLACK_TOKEN=XYZ SPREADSHEET_ID=ABC \
node build/index.js

Worksheets

At start, the bot will create the KPIs and Data worksheets if they don't exist yet.

The KPIs worksheet is where you put the KPIs you want to track. Each KPI must have the following attributes:

name: an alphanumeric unique identifier for the KPI
question: this is the text that gets used to create the question together with the period (for instance if the question field is How many customer signed up and the period is May 1st, 2016, then the bot will ask How many customer signed up on May 1st, 2016?)
owner-slack: the Slack username of the user that will be asked about this KPI (this may be different from the name that you see in the chat, you can look up the username on the Slack profile of the user)
frequency: how often to collect the value of this KPI (daily, weekly, monthly)
since: the date (YYYY-MM-DD) since you want to track this KPI - in case of weekly KPIs it must be the Monday of that week, in case of monthly KPIs it must be the 1st day of that month.
kippino-enable: set this to true or yes if you want to enable this KPI.

The Data worksheet is where Kippino puts the values collected from your team. Each data point has the following attributes:

timestamp: ISO formatted timestamp for when the value was captured
kpi: the KPI name
value: the KPI value
for: the date associated to the period that this value refers to (same convention used by the since attribute)
source: the user that provided this value

Commands

The bot accepts the following commands:

help provides a brief description of the bot and the list of commands
sync KPIs reloads the list of KPIs (this is also done every hour)
sync users reloads the list of Slack users (this is also done every hour)
list KPIs lists the configured KPIs
pending lists the users that have pending questions (i.e. the bot asked them for a KPI but they didn't respond yet)

Deployment options

Docker

You can run Kippino via Docker, there's already a pre-built docker image or you can build your own from the provided Dockerfile.

$ docker run \
  -v ./auth.json:/auth.json \
  -e AUTH_JSON=/auth.json \
  -e SLACK_TOKEN=xyz \
  -e SPREADSHEET_ID=abc \
  --name kippino measurence/kippino

Kubernetes / Google Conainer Service

You can easily deploy Kippino on a Kubernetes cluster with the following config. The authentication JSON and the Slack token should be configured as secrets.

---
apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: kippino
  labels:
    app: kippino
spec:
  strategy:
    type: Recreate
  template:
    metadata:
      name: kippino
      labels:
        app: kippino
    spec:
      volumes:
      - name: kippino-google-auth
        secret:
          secretName: "kippino-google-auth"
      containers:
      - image: "measurence/kippino:latest"
        name: kippino
        volumeMounts:
        - name: "kippino-google-auth"
          mountPath: "/kippino-google-auth/"
          readOnly: true
        env:
        - name: SLACK_TOKEN
          valueFrom:
            secretKeyRef:
              name: kippino
              key: "slack-token"
        - name: SPREADSHEET_ID
          value: "XYZ"
        - name: AUTH_JSON
          value: "/kippino-google-auth/auth.json"
        resources:
          requests:
            cpu: 0.1
            memory: "256Mi"
          limits:
            cpu: 0.1
            memory: "256Mi"

Gotchas

Dealing with user timezone and online status

Currently the bot is quite stupid and it will start asking questions as soon as a new day starts in the bot timezone.

It will also keep waiting for an answer "forever", it doesn't keep "nudging" people until they respond.

Spreadsheet locale and number formatting

Make sure that your spreadsheet is configured in the US locale since the bot uses Javascript's parseFloat to parse numbers in the US format (using . as decimal separator).

Date formatting

Dates in the spreadsheet must be formatted as YYYY-MM-DD.

For weekly KPIs, the since field must be set to the Monday of the week you want to start collecting the KPI.

cloudify/kippino