The measurement mental model

The measurement mental model

If you understand one thing about Adobe Analytics, understand this mental model. Everything else is details on top of it.

The core idea: database + schema

Think of Adobe Analytics like a database:

A report suite is a database. Every time someone uses your website, one row (hit) is inserted. Your measurement plan defines the schema: which columns (props/eVars) you collect, what each means.

The three levels of aggregation

Analytics summarizes data at three levels:

Level 1: Hits (atomic)

A hit is one action: one page load, one button click, one form submission.

Hit 1: Alice loads homepage
Hit 2: Alice clicks to product page
Hit 3: Alice adds item to cart (event fires)

Level 2: Visits (sessions)

A visit groups consecutive hits from one user, timed out by 30 minutes of inactivity.

Visit 1 (Monday):
  Hit 1: Homepage
  Hit 2: Product
  Hit 3: Cart

[48 hours pass]

Visit 2 (Wednesday):
  Hit 1: Search results

Same person, two visits.

Level 3: Visitors (people)

All a person's visits and hits are grouped under one visitor ID (ECID or user ID).

Visitor: Alice
  Visit 1: 3 hits
  Visit 2: 1 hit
  Visit 3: 5 hits

Variables as function parameters

Think of props and eVars like function parameters in code:

def track_user_action(page, form_type, visitor_segment):
  # page = prop (hit-scoped)
  # form_type = eVar with visit scope
  # visitor_segment = eVar with visitor scope
  pass

Props = request fields (hit-level)

Every hit can have different prop values. They're not remembered for the next hit.

Hit 1: prop5 = "homepage"
Hit 2: prop5 = "product-page"
Hit 3: prop5 = "cart"

Props answer: "What page are we on right now?"

eVars = session variables (visit or visitor scope)

Set once, persists for the lifetime of a visit or visitor.

eVar1 = "marketing_campaign_xyz" // set on hit 1
Hit 1: eVar1 = "marketing_campaign_xyz"
Hit 2: eVar1 = "marketing_campaign_xyz" (remembered)
Hit 3: eVar1 = "marketing_campaign_xyz" (still remembered)
[30 minutes pass, visit times out]
Hit 4 (new visit): eVar1 = empty (reset)

eVars answer: "What was the context for this session?"

Events = counters

Fires when something happens. Think of it like counting.

// Contact form submitted
sendEvent({ events: "event1" })

// Later, in reporting:
// "event1 fired 1,247 times"

Events answer: "How many times did X happen?"

Reporting = dimension + metric

Every report is: "[what you count] by [what you group by]"

Page Views by Page Name

| Page Name   | Page Views |
|-------------|-----------|
| Homepage    | 5,200     |
| Product     | 3,100     |
| Checkout    | 1,800     |

• Dimension (left): Page Name (what you group by)
• Metric (column): Page Views (what you count)

Segments = WHERE clauses

A segment is a filter. In SQL:

SELECT * FROM analytics
WHERE page_name = 'checkout'
  AND device_type = 'mobile'

In Adobe Analytics:

Segment: "Mobile checkout visitors"
  - Page Name = Checkout
  - Device Type = Mobile

Segments let you slice and compare.

Example Super mapping

Example Super (our running example) is a superannuation website:

---

The mental model: Report suite = database. Hits = rows. Props/eVars = columns. Events = counters. Segments = filters.

Once you get this, everything else is just detail on top. Next: Analytics Fundamentals