Skip to main content

[Beta]: Sessions in the Attribution Deep Dive

Learn what sessions, new visitor sessions and first order sessions mean in the Attribution Deep Dive, and how each visit is credited to a channel. Includes why session numbers differ from your attributed orders in Klar and from GA4.

Written by Marc Garbella

Sessions bring the top of your funnel into the same view as your spend and your revenue. You can now see, for every channel, campaign, or ad: how many people visited, how many of them were there for the first time, how many looked at products, and how many added something to their cart.

This article explains what each metric means and how to read it.

What is a session?

A session is one continuous visit to your store by one visitor.

It starts when a visitor lands on your site and ends after 20 minutes of inactivity. Everything the visitor does in that window — every page they view, every product they look at, every add to cart, and an order if they place one — belongs to the same session.

A few consequences worth knowing:

  • One person can have many sessions. Someone who browses in the morning and comes back in the evening has two sessions.

  • A long pause splits a visit in two. If a visitor leaves a tab open, walks away for an hour and comes back, that becomes a second session.

  • Moving around your site does not start a new session. Clicking from an ad to a collection page to a product page to checkout is one session, not four.

  • A session is counted on the day it started. A visit that begins at 23:50 and ends at 00:20 counts on the first day.

How Klar recognises a visitor

Klar's pixel gives each visitor a first-party identifier, so the same person is recognised when they come back. Where Klar can tell that two devices or browsers belong to the same person — for example someone who browses on mobile and then buys on desktop — those visits are grouped under one visitor.

This works well, but it is not perfect, and it is the reason "new visitor" should be read as "new as far as we can tell":

  • Someone who clears their cookies, switches browsers, or uses a private window may be seen as a new visitor again.

  • A visitor is "new" relative to the period Klar has been tracking your store. In the first weeks after going live, almost every visitor looks new.

Which channel a session is credited to

Every session is credited to exactly one channel — the most valuable marketing touch that happened during that visit.

When several touches happen in one visit, Klar keeps the most meaningful one, in this order of priority:

  1. Paid channels (Meta, Google paid search, TikTok, influencer, display, …)

  2. Branded paid search

  3. Owned channels (email, SMS/WhatsApp, organic search, organic social, partnerships)

  4. Referral

  5. Direct and unknown

So if a visitor clicks a Meta ad, then clicks an internal link that looks like a referral, then reloads the page directly, the whole session is credited to the Meta ad, not split across the three.

If a session has no marketing touch at all — a direct visit, or a visitor whose source we cannot identify — Klar looks back at that visitor's history and credits the session to the last meaningful channel that brought them to you, within the last 90 days. This is why direct traffic in the Deep Dive is smaller than you may be used to elsewhere: a returning customer typing in your URL after first arriving through a Meta ad is credited to Meta, not to direct.

The metrics

Traffic

Metric

What it counts

Sessions

All visits credited to this channel, campaign, ad, landing page or product.

New visitor sessions

Visits from people on their first-ever session with your store.

Returning visitor sessions

Visits from people who have been on your site before. Sessions minus new visitor sessions.

New visitor sessions are your best read on how much genuinely new audience a channel is reaching, as opposed to re-engaging people who already know you. A channel with a high share of returning visitor sessions is doing retention work, even when it looks like acquisition.

Engagement

Metric

What it counts

Product views

Product pages viewed across these sessions. One session can contribute several.

First product views

Views of the first product a visitor looked at in a session — the product that pulled them in.

Add to carts

Add-to-cart actions in these sessions.

Unique product views

Visitors who viewed a product, counted once each.

Unique add to carts

Visitors who added to cart, counted once each.

Use the totals when you care about volume of activity, and the unique versions when you care about how many people did something.

Reading the numbers correctly

Session metrics won't match your attributed orders in Klar

Session metrics and Klar's attributed orders are built for different jobs, and they will not tie out. Both are correct; they answer different questions.

  • Order sessions credit the whole order to the one channel that drove that visit. It tells you what was happening on your site.

  • Klar attributed orders consider the customer's full journey across all touches and channels, incorporating everything Klar knows about incrementality. It tells you which channels actually earned the revenue.

Use session metrics to understand behaviour on site — traffic quality, engagement, funnel drop-off. Use attributed orders and revenue to decide where to put budget.

Session metrics also only cover visits our pixel can see, so they will run below your total store order count.

Session metrics won't match GA4 or your shop platform

Every analytics tool defines sessions slightly differently — different inactivity windows, different rules for what starts a new session, different ways of recognising returning visitors. Comparing absolute session counts across tools is not a useful exercise. Comparing trends and relative differences between channels inside one tool is.


Questions this helps you answer

  • Which campaigns bring genuinely new audience, and which mostly re-engage people who already know us?

  • Where do visitors from each channel drop off — do they not view products, not add to cart, or not check out?

  • Which landing pages convert the traffic we send them, and which waste it?

  • Which channels bring first-time customers rather than repeat buyers?

Did this answer your question?