BoldieBot Daily Script

Purpose

I needed to automate the delivery of key GA4 metrics to non-analytical teams (Tech, Product, Finance) via Slack. The goal was to provide a daily snapshot of company performance without requiring access to analytics tools.

Hypothesis

By automating daily metric delivery to Slack, I believed I could increase cross-team awareness of performance data and potentially lead to more data-driven discussions.

Technical Implementation

1. GA4 API Integration

I used the Google Analytics Data API v1beta to fetch GA4 data. The core of the data fetching logic was implemented in the get_report function:

python
Copy
def get_report(client):
    property_id = 'YOUR_GA4_PROPERTY_ID'# Masked for security
    yesterday = datetime.now() - timedelta(days=1)
    last_week = yesterday - timedelta(days=7)

    general_request = RunReportRequest(
        property=f"properties/{property_id}",
        dimensions=[{"name": "date"}],
        metrics=[
            {"name": "activeUsers"},
            {"name": "sessions"},
        ],
        date_ranges=[
            {"start_date": yesterday.strftime('%Y-%m-%d'), "end_date": yesterday.strftime('%Y-%m-%d')},
            {"start_date": last_week.strftime('%Y-%m-%d'), "end_date": last_week.strftime('%Y-%m-%d')}
        ],
    )

    event_request = RunReportRequest(
        property=f"properties/{property_id}",
        dimensions=[{"name": "eventName"}, {"name": "date"}],
        metrics=[{"name": "eventCount"}],
        date_ranges=[
            {"start_date": last_week.strftime('%Y-%m-%d'), "end_date": yesterday.strftime('%Y-%m-%d')}
        ],
    )

    return client.run_report(general_request), client.run_report(event_request), yesterday, last_week

This function fetches general metrics (Users, Sessions) and event-specific data (Offer Form Submissions) for yesterday and the same day last week.

2. Data Processing

The process_data function handles the API response:

python
Copy
def process_data(general_response, event_response, yesterday, last_week):
    metrics = ['Users', 'Sessions', 'Offer Form Submitted']
    results = []

    for i, metric in enumerate(metrics[:2]):
        yesterday_value = int(general_response.rows[0].metric_values[i].value)
        last_week_value = int(general_response.rows[1].metric_values[i].value)
        results.append(calculate_change(metric, yesterday_value, last_week_value))

    yesterday_str = yesterday.strftime('%Y%m%d')
    last_week_str = last_week.strftime('%Y%m%d')

    for row in event_response.rows:
        if row.dimension_values[0].value == "Offer Form Submitted":
            date = row.dimension_values[1].value
            count = int(row.metric_values[0].value)
            if date == yesterday_str:
                yesterday_value = count
            elif date == last_week_str:
                last_week_value = count

    results.append(calculate_change('Offer Form Submitted', yesterday_value, last_week_value))
    return results

This function calculates the day-over-day change for each metric.

3. Slack Integration

I used the Slack SDK to post messages. The core posting logic:

python
Copy
def post_to_slack(message):
    slack_token = "YOUR_SLACK_BOT_TOKEN"# Masked for security
    client = WebClient(token=slack_token)

    try:
        response = client.chat_postMessage(
            channel="YOUR_SLACK_CHANNEL_ID",# Masked for security
            text=message
        )
        print(f"Message posted to Slack successfully")
    except SlackApiError as e:
        print(f"Error posting message to Slack: {e}")

Technical Challenges and Solutions

GA4 Data Discrepancies Problem: Initial script reported zero Offer Form Submissions, contradicting GA4 interface data. Solution: I implemented extensive logging in the data processing function: This revealed a date format mismatch. GA4 was returning dates in 'YYYYMMDD' format, while I was comparing with 'YYYY-MM-DD'.

python
Copy
print(f"Looking for Offer Form Submitted events on {yesterday_str} and {last_week_str}")
for row in event_response.rows:
    event_name = row.dimension_values[0].value
    date = row.dimension_values[1].value
    count = int(row.metric_values[0].value)
    if event_name == "Offer Form Submitted":
        print(f"Offer Form Submitted event found on {date}: {count}")

Error Handling Problem: Silent failures during API issues. Solution: Implemented try-except blocks with detailed error logging:

python
Copy
try:
# API call or data processing logic
except Exception as e:
    print(f"An error occurred: {str(e)}")
    sys.exit(1)

Date Handling Problem: Inconsistent date formatting across the script. Solution: Standardized date handling using strftime:

python
Copy
yesterday_str = yesterday.strftime('%Y%m%d')
last_week_str = last_week.strftime('%Y%m%d')

Automation I used cron to schedule daily execution: This runs the script daily at 9:00 AM and logs output and errors.

bash
Copy
0 9 * * * /usr/local/bin/python3 /path/to/ga4_to_slack.py >> /path/to/ga4_to_slack.log 2>&1

Setup Cron

I wanted my GA4 to Slack script to run automatically every day at 9:00 AM, so I used a tool called cron. Cron is like a scheduler for your mac that can run programs at specific times. To set it up, I opened Terminal and typed crontab -e to edit the list of scheduled tasks. I added this line: 0 9 * * * /usr/local/bin/python3 /Users/Dimosthenis/Documents/GA4toSlack/ga4_to_slack.py >> /Users/Dimosthenis/Documents/GA4toSlack/ga4_to_slack.log 2>&1. This tells the computer to run my script at 9:00 AM every day and save any messages about how it went. To test if it worked, I didn't want to wait until the next day, so I changed the time to a few minutes in the future. For example, when it was 2:42 PM, I set it to run at 2:46 PM by changing the first two numbers in the cron line. I then waited for that time to come and checked if my Slack channel got the message and if the log file showed that the script ran. After making sure it worked, I changed the time back to 9:00 AM for regular daily runs. Setting this up took some trial and error, but now our teams get fresh GA4 data in Slack every morning without anyone having to remember to run the script!

Results and Next Steps

The script now runs daily at 09:00, providing consistent KPI updates to non-analytical teams. Early feedback indicates increased data awareness across departments.

Next steps:

Add weekly/monthly trend analysis.
Add more KPIs

Marketing: Cost per acquisition, Conversion rates
Product: User retention metrics or App data (DAU, etc)

Add visualization capabilities, potentially using matplotlib to generate and attach charts to Slack messages.

Keep iterating and stay curious!