Server-side multi-touch attribution for Python. Track customer journeys, attribute conversions, know which channels drive revenue.

Unlike client-side analytics, mbuzz runs server-side — no ad-blocker blind spots, no iOS tracking gaps, no browser privacy loss. Captures 30–40% more touchpoints than client-side tools.

pip install mbuzz

Requires Python 3.9+.

import mbuzz

mbuzz.init(api_key="sk_live_...")

Call init() once on app boot. For Flask apps, see Flask Integration below.

Track steps in the customer journey:

mbuzz.event("page_view", url="/pricing")
mbuzz.event("add_to_cart", product_id="SKU-123", price=49.99)
mbuzz.event("checkout_started", cart_total=99.99)

# Group events into funnels for focused analysis
mbuzz.event("signup_start", funnel="signup", source="homepage")
mbuzz.event("signup_complete", funnel="signup")

Record revenue-generating outcomes:

mbuzz.conversion(
    "purchase",
    revenue=99.99,
    funnel="purchase",   # optional: group into funnel
    order_id=order.id,
)

Link visitors to known users (enables cross-device attribution):

# On signup or login
mbuzz.identify(
    user.id,
    traits={
        "email": user.email,
        "name": user.name,
    },
)

Group related events into funnels for focused conversion analysis in your dashboard.

# Signup funnel
mbuzz.event("pricing_view", funnel="signup")
mbuzz.event("signup_start", funnel="signup")
mbuzz.event("signup_complete", funnel="signup")

# Purchase funnel
mbuzz.event("add_to_cart", funnel="purchase")
mbuzz.event("checkout_started", funnel="purchase")
mbuzz.conversion("purchase", funnel="purchase", revenue=99.99)

Why funnels?

• Separate signup flow from purchase flow
• Analyze each conversion path independently
• Filter dashboard to specific customer journeys

mbuzz ships with a Flask middleware that handles visitor cookies, session resolution, and request context automatically.

from flask import Flask
import mbuzz
from mbuzz.middleware.flask import init_app

app = Flask(__name__)

mbuzz.init(api_key="sk_live_...")
init_app(app)

After init_app, every request carries a visitor_id in context. Track events from anywhere in your request lifecycle:

@app.route("/pricing")
def pricing():
    mbuzz.event("page_view", page="/pricing")
    return render_template("pricing.html")

Access the current visitor or user from context:

mbuzz.visitor_id()   # Current visitor ID (from cookie)
mbuzz.user_id()      # Current user ID (if identify() was called)

# settings.py
MIDDLEWARE = [
    # ...
    "mbuzz.middleware.django.MbuzzMiddleware",
]

Initialise the SDK once at startup — in your AppConfig.ready() or settings.py:

import mbuzz

mbuzz.init(api_key="sk_live_...")

Install with pip install mbuzz[django].

Works with any Starlette app; FastAPI is Starlette underneath.

from fastapi import FastAPI
import mbuzz
from mbuzz.middleware.fastapi import MbuzzMiddleware

app = FastAPI()

mbuzz.init(api_key="sk_live_...")
app.add_middleware(MbuzzMiddleware)

Install with pip install mbuzz[fastapi].

The core mbuzz.event, mbuzz.conversion, and mbuzz.identify functions are framework-agnostic. You can call them from any Python web app or background job.

Add this to your base template or nothing is tracked. Since 0.10.0 a page response never sets the visitor cookie: only POST /_mbuzz/session mints, and the snippet is what calls it.

A page response can be stored by a full-page cache (Cloudflare, Varnish, nginx, a CDN) and replayed to every visitor. A Set-Cookie sitting in that cache hands everyone the first visitor's id, so unrelated people merge into one journey — corruption rather than loss, since every row exists and is simply attributed to the wrong person. Minting only on a response no cache stores is the only way to prevent it.

The same endpoint solves the original problem too: a cached page never enters your application, so the middleware cannot run — but this one request always reaches it.

The middleware answers POST /_mbuzz/session on all three frameworks. There is nothing extra to mount. Call it once per page, from your base template:

<script>
  fetch('/_mbuzz/session', {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({ url: location.href, referrer: document.referrer || '' }),
    credentials: 'same-origin',
    keepalive: true
  }).catch(function () {});
</script>

Two things to keep as they are:

• Inline the script, don't load it as a file. Asset optimisers delay external scripts until the visitor first interacts. A visitor who lands and converts without clicking anything first would never be established.
• credentials: 'same-origin' is required, or the cookie never comes back.

The visitor id is never created or read in JavaScript. It stays HttpOnly and server-set, which is what preserves its full two-year life — a cookie written by document.cookie is capped at 7 days under Safari's ITP, and 24 hours after an ad click.

For tracking from Celery, RQ, or other background workers, pass visitor_id explicitly:

# Capture at request time
@app.route("/checkout")
def checkout():
    visitor_id = mbuzz.visitor_id()
    process_order.delay(order.id, visitor_id)
    return redirect("/thank-you")

# Use in the job
@celery.task
def process_order(order_id, visitor_id):
    order = Order.get(order_id)
    mbuzz.conversion(
        "purchase",
        visitor_id=visitor_id,
        revenue=order.total,
    )

mbuzz.init(
    api_key="sk_live_...",       # Required
    enabled=True,                # Default: True — toggle without code changes
    debug=False,                 # Default: False — enables verbose logging
    timeout=5.0,                 # Default: 5.0s — API request timeout
    skip_paths=["/health"],      # Optional — paths to skip tracking
    skip_extensions=[".css"],    # Optional — file extensions to skip
)

mbuzz never raises exceptions. All methods fail silently and log errors in debug mode — your application flow is never interrupted by tracking failures.

Client-side tracking loses 30–40% of marketing data to ad blockers, iOS tracking protections, and 7-day cookie caps. Server-side attribution runs inside your application, where requests flow through your own domain as first-party data — so you see the full customer journey.

Full explainer: mbuzz.co/articles/server-side-vs-client-side-tracking

mbuzz runs 8 attribution models side-by-side out of the box — first-touch, last-touch, linear, time-decay, position-based, Markov, Shapley, and data-driven — so you can see how much each model disagrees about which channel deserves credit.

• Python 3.9+
• No required dependencies for the core SDK
• flask, django or starlette only required for that framework's middleware

• Getting started
• Dashboard
• Changelog

PRs welcome — especially framework adapters (Django, FastAPI, Starlette) and async support.

MIT License