6 Phases of Analytics

Slide38

6 Phases of Analytics

  1. LAST CLICK
    What was the last thing the customer did before they clicked?
  2. OVERLAP
    There can be multiple clicks by a single person.
  3. TRIGGERS
    Triggers sequences of events that fire off according to actions by customers (think IF/THEN).
  4. SEQUENCES
    Sequences (spiral) — more and more sequences create a spiral effect.
  5. LIFT TESTING
    How to get from point A to B. Try another path to find out how else to get to B. Test to find out a better conversion rate, etc.
  6. MEDIA MIX MODELING
    Media mix modeling – Different channels impact different cities. What type of media creates more impact?

Watch each video segment to learn more about each phase.

Last Click – The most common, but flawed, model that gives 100% credit to the last thing the customer did before conversion. Penalizes the “assist” or word of mouth power in other parts of the funnel. By default, you’re doing this.

Attribution Lesson 2 (04:57)

An Overlap – is when people are part of two audiences. They could like pizza AND chicken wings. An Affinity Grid shows the set of interests and behaviors that describe our target customer. Actions can overlap, too, if they have been on the site in the last 7 days and are a follower. This lets us calculate propensities to convert.

Attribution Lesson 3 (06:03)

Triggers – unlike overlaps, require order of operations. IF/THEN statements (Event A happens and then Event B). An overlap helps us initially determine when two targets are related, but a trigger tells us which target comes first. Branded search is triggered by something upstream, for example.

Attribution Lesson 4 (02:09)

Sequences – are triggers stitched together. The most complex of funnels can be disassembled into atomic building blocks of triggers. The most powerful sequences go cross-channel from social to email, from email to site, from app to in-store visit, or other combinations.

Attribution Lesson 5 (03:48)

Lift Testing – measures the incremental impact of a marketing action. This is not the same as split testing, which is finding “winners” among creative variants. Lift testing requires a holdout audience (the control of a test and control experiment) so we can measure true incremental impact. We can randomize by many possible variables.

Attribution Lesson 6 (11:39)

Media Mix Modeling – is the most sophisticated and final stage of analytics. It allows us to see how all channels work together to drive conversions in the customer journey. All prior phases are prerequisites.

Attribution Lesson 7 (08:15)