Skip to main content
Aurora CognitiveAurora Cognitive

ANALYTICS AND CRO

Decide from evidence, not from opinion

An event schema written before any tool is installed, collection that survives blockers and consent choices, and tests that run long enough to mean something.

Define my event schema

30 minutes with an engineer, no sales call

What this is

Schema first, tooling second

Most measurement problems start with a tag installed before anyone agreed what a conversion is. We define the events, their properties and their meaning in writing, collect them server side where we can, respect the consent choice on every path, and only then choose tools. On top of that sits a testing process with a stated minimum duration, plus qualitative input so you know why a page loses, not only that it did.

This is for you if

  • Two dashboards report different totals for the same week and both are defended by someone.
  • You know which page loses people but not what they were trying to do when they left.
  • Someone changes a page, the number moves, and nobody can say whether the change caused it.

This is not the right fit if your traffic is too low for a test to conclude. Splitting a few hundred sessions produces noise you can read any way you like. We would run qualitative research instead: session review, short user interviews and a heuristic pass, then change the page on that evidence.

Scope

What we build

Documented event schema

Every event named, with its trigger, its properties, its owner and the decision it supports. Anything that supports no decision is not collected.

Server side collection

Events sent from your own backend where possible, so blockers and browser changes do not quietly remove a third of your data.

Consent aware setup

Measurement that reflects the visitor's choice on every path, with the difference between consented and total traffic visible rather than hidden.

Dashboard tied to decisions

One view per recurring decision, each with the number, its definition and its source, instead of a wall of charts nobody acts on.

Testing process

A hypothesis format, one change per test, a stated minimum duration before anyone looks, and a written result whether it won or lost.

Qualitative inputs

Session review, short interviews and a heuristic pass, so every test starts from an observed problem rather than a meeting opinion.

How it goes

From arguing about numbers to deciding on them

  1. 01

    Schema definition

    One to two weeks

    We work back from the decisions you make monthly, define the events those need, and write the schema before touching a tool.

  2. 02

    Collection and consent

    Two to three weeks

    Server side collection implemented, consent handling wired through every path, and totals reconciled against known reference points.

  3. 03

    Dashboards and qualitative baseline

    One to two weeks

    Decision views built, plus a first qualitative pass to name the problems worth testing.

  4. 04

    Test cycles

    Ongoing in three to six week cycles

    One change per test, run to the stated minimum duration, result written down and applied or reverted.

Artefacts

What you receive

Written record

Event schema with owner per event

Metric definitions and their sources

Test hypotheses with minimum durations

Result log including the tests that lost

In place

Server side collection in your own stack

Consent aware measurement on every path

Decision dashboards

Data quality checks on event volume

Qualitative

Session review notes on the key paths

Interview summaries

Heuristic review of the deciding pages

Handover session recording

We state a minimum test duration rather than a promised lift, and we report losing tests as plainly as winning ones.

Questions

Questions we get about analytics and CRO

Next step

Bring the number two dashboards disagree on

Book a technical assessment

30 minutes with an engineer, no sales call