Add customer insights models: CLV and campaign funnel

dbt/dbt_project.yml

@@ -29,5 +29,7 @@ models:
   marketing_analytics:
     staging:
       +materialized: view
+    intermediate:
+      +materialized: view
     marts:
       +materialized: view

dbt/models/intermediate/_intermediate.yml

@@ -0,0 +1,42 @@
+version: 2
+
+models:
+  - name: int_customer_lifetime_value
+    description: |
+      Customer-level aggregation of conversion data.
+      Grain: One row per user_id.
+    columns:
+      - name: user_id
+        description: Unique identifier for the customer
+        data_tests:
+          - not_null
+          - unique
+      - name: total_revenue
+        description: Total conversion value across all purchases
+      - name: total_orders
+        description: Number of distinct conversions
+      - name: first_purchase_date
+        description: Date of the customer's first purchase
+      - name: last_purchase_date
+        description: Date of the customer's most recent purchase
+
+  - name: int_campaign_funnel
+    description: |
+      Campaign-level funnel metrics aggregated across all dates.
+      Grain: One row per campaign_id.
+    columns:
+      - name: campaign_id
+        description: Unique identifier for the campaign
+        data_tests:
+          - not_null
+          - unique
+      - name: total_impressions
+        description: Total impressions across all dates
+      - name: total_clicks
+        description: Total clicks across all dates
+      - name: total_sessions
+        description: Total distinct sessions attributed to this campaign
+      - name: total_conversions
+        description: Total distinct conversions attributed to this campaign
+      - name: total_revenue
+        description: Total conversion revenue attributed to this campaign

dbt/models/intermediate/int_campaign_funnel.sql

@@ -0,0 +1,50 @@
+with campaigns as (
+    select * from {{ ref('stg_campaigns_daily') }}
+),
+
+campaign_totals as (
+    select
+        campaign_id,
+        sum(impressions) as total_impressions,
+        sum(clicks) as total_clicks
+
+    from campaigns
+    group by campaign_id
+),
+
+session_counts as (
+    select
+        campaign_id,
+        count(distinct session_id) as total_sessions
+
+    from {{ ref('stg_sessions') }}
+    group by campaign_id
+),
+
+conversion_counts as (
+    select
+        attributed_campaign_id as campaign_id,
+        count(distinct conversion_id) as total_conversions,
+        sum(conversion_value) as total_revenue
+
+    from {{ ref('stg_conversions') }}
+    group by attributed_campaign_id
+),
+
+final as (
+    select
+        ct.campaign_id,
+        ct.total_impressions,
+        ct.total_clicks,
+        coalesce(sc.total_sessions, 0) as total_sessions,
+        coalesce(cc.total_conversions, 0) as total_conversions,
+        coalesce(cc.total_revenue, 0) as total_revenue
+
+    from campaign_totals ct
+    left join session_counts sc
+        on ct.campaign_id = sc.campaign_id
+    left join conversion_counts cc
+        on ct.campaign_id = cc.campaign_id
+)
+
+select * from final

dbt/models/intermediate/int_customer_lifetime_value.sql

@@ -0,0 +1,17 @@
+with conversions as (
+    select * from {{ ref('stg_conversions') }}
+),
+
+customer_metrics as (
+    select
+        user_id,
+        sum(conversion_value) as total_revenue,
+        count(distinct conversion_id) as total_orders,
+        min(date) as first_purchase_date,
+        max(date) as last_purchase_date
+
+    from conversions
+    group by user_id
+)
+
+select * from customer_metrics

dbt/models/marts/_marts.yml

@@ -154,3 +154,63 @@ models:
       - name: overall_cpa
         description: Overall cost per acquisition (total spend / total conversions)
 
+  - name: customer_lifetime_value
+    description: |
+      Customer lifetime value with acquisition channel and tier segmentation.
+      Answers: "How much is each customer worth and where did they come from?"
+
+      Grain: One row per user_id
+    columns:
+      - name: user_id
+        description: Unique identifier for the customer
+        data_tests:
+          - not_null
+          - unique
+      - name: total_revenue
+        description: Total conversion value across all purchases
+      - name: total_orders
+        description: Number of distinct purchases
+      - name: first_purchase_date
+        description: Date of the customer's first purchase
+      - name: last_purchase_date
+        description: Date of the customer's most recent purchase
+      - name: acquisition_channel
+        description: Channel of the customer's earliest first-touch attribution touchpoint
+      - name: ltv_tier
+        description: "Customer segment: high ($500+), medium ($100-499), low (under $100)"
+
+  - name: campaign_funnel_analysis
+    description: |
+      Campaign funnel with stage-to-stage conversion rates and efficiency score.
+      Answers: "Where are we losing people in the funnel, per campaign?"
+
+      Grain: One row per campaign_id
+    columns:
+      - name: campaign_id
+        description: Unique identifier for the campaign
+        data_tests:
+          - not_null
+          - unique
+      - name: campaign_name
+        description: Name of the campaign
+      - name: channel
+        description: Marketing channel
+      - name: total_impressions
+        description: Total impressions across all dates
+      - name: total_clicks
+        description: Total clicks across all dates
+      - name: total_sessions
+        description: Total sessions attributed to this campaign
+      - name: total_conversions
+        description: Total conversions attributed to this campaign
+      - name: total_revenue
+        description: Total revenue from conversions
+      - name: impression_to_click_rate
+        description: Click-through rate (clicks / impressions)
+      - name: click_to_session_rate
+        description: Rate of clicks that became sessions
+      - name: session_to_conversion_rate
+        description: Rate of sessions that converted
+      - name: efficiency_score
+        description: Overall funnel throughput (conversions / impressions) for ranking campaigns
+

dbt/models/marts/campaign_funnel_analysis.sql

@@ -0,0 +1,54 @@
+with funnel as (
+    select * from {{ ref('int_campaign_funnel') }}
+),
+
+campaign_details as (
+    select distinct
+        campaign_id,
+        campaign_name,
+        channel
+
+    from {{ ref('stg_campaigns_daily') }}
+),
+
+final as (
+    select
+        f.campaign_id,
+        cd.campaign_name,
+        cd.channel,
+
+        -- Funnel counts
+        f.total_impressions,
+        f.total_clicks,
+        f.total_sessions,
+        f.total_conversions,
+        f.total_revenue,
+
+        -- Stage-to-stage conversion rates
+        case
+            when f.total_impressions > 0 then (f.total_clicks::float / f.total_impressions::float)
+            else 0
+        end as impression_to_click_rate,
+
+        case
+            when f.total_clicks > 0 then (f.total_sessions::float / f.total_clicks::float)
+            else 0
+        end as click_to_session_rate,
+
+        case
+            when f.total_sessions > 0 then (f.total_conversions::float / f.total_sessions::float)
+            else 0
+        end as session_to_conversion_rate,
+
+        -- Overall efficiency score
+        case
+            when f.total_impressions > 0 then (f.total_conversions::float / f.total_impressions::float)
+            else 0
+        end as efficiency_score
+
+    from funnel f
+    left join campaign_details cd
+        on f.campaign_id = cd.campaign_id
+)
+
+select * from final

dbt/models/marts/customer_lifetime_value.sql

@@ -0,0 +1,44 @@
+with customer_metrics as (
+    select * from {{ ref('int_customer_lifetime_value') }}
+),
+
+first_touch_ranked as (
+    select
+        user_id,
+        channel as acquisition_channel,
+        row_number() over (partition by user_id order by timestamp asc) as rn
+
+    from {{ ref('stg_attribution_touchpoints') }}
+    where touchpoint_position = 1
+),
+
+first_touch as (
+    select
+        user_id,
+        acquisition_channel
+
+    from first_touch_ranked
+    where rn = 1
+),
+
+final as (
+    select
+        cm.user_id,
+        cm.total_revenue,
+        cm.total_orders,
+        cm.first_purchase_date,
+        cm.last_purchase_date,
+        coalesce(ft.acquisition_channel, 'unknown') as acquisition_channel,
+
+        case
+            when cm.total_revenue >= 500 then 'high'
+            when cm.total_revenue >= 100 then 'medium'
+            else 'low'
+        end as ltv_tier
+
+    from customer_metrics cm
+    left join first_touch ft
+        on cm.user_id = ft.user_id
+)
+
+select * from final