Baseline touchpoint parameters
A touchpoint is one standardized advertising contact category. One catalog creates both the daily performance rows and the keys inside ordered paths, preventing unmatched names.
identifier
Meaning. An internal English name used by path configurations.
Example. sponsored_products_top_of_search is easier to audit than referring to “touchpoint number six.”
Why it makes sense. Stable names prevent a reordered configuration list from silently changing paths.
Confidence. High.
ad_product
Meaning. Broad Amazon advertising product family.
Example. AMAZON_DSP means Amazon Demand-Side Platform, while SPONSORED_PRODUCTS identifies Sponsored Products.
Why it makes sense. Product family is the first segment of normalizedTouchpoint.
Confidence. High as a finalized-schema field.
ad_type
Meaning. Sponsored-ad format category used as the second normalized-key segment.
Example. PRODUCT_AD describes the configured Sponsored Products rows.
Relationship. It must be null for Amazon Demand-Side Platform rows, which use inventory_type in the same key position.
Confidence. High as a structural rule.
creative_type
Meaning. Creative format such as IMAGE or VIDEO.
Example. Two Sponsored Brands touchpoints can share placement but differ by image versus video.
Why it makes sense. Creative format can change attention and response, so it should not be hidden inside a generic channel label.
Confidence. High for the field; configured effects remain assumptions.
inventory_type
Meaning. Amazon Demand-Side Platform inventory category.
Example. DISPLAY and ONLINE_VIDEO distinguish two demand-side-platform touchpoints.
Relationship. It must be null for Sponsored Ads and replaces ad_type in the second normalized-key segment.
Confidence. High as a structural rule.
placement
Meaning. Where a sponsored advertisement appears.
Example. TOP_OF_SEARCH, REST_OF_SEARCH, and PRODUCT_PAGE are separate placement values.
Why it makes sense. A top-of-search impression can have different competition and purchase intent from a product-page impression.
Confidence. High for the distinction; simulated rates are low confidence.
normalizedTouchpoint
Meaning. The finalized four-segment key derived from raw touchpoint fields:
Example. SPONSORED_PRODUCTS:PRODUCT_AD:TOP_OF_SEARCH:UNSPECIFIED.
Why it makes sense. A single key lets path entries match performance rows. Raw nulls remain null; UNSPECIFIED appears only inside this normalized key.
Relationship. Every touchpoint_identifiers entry resolves to one of these keys.
Confidence. High; finalized project invariant.
base_impressions
Meaning. Neutral daily impression count before marketplace scale, seasonality, and noise.
Example. 47000 for top-of-search Sponsored Products means 47,000 expected impressions on a neutral day in a reference marketplace.
Why it makes sense. This supplies the opportunity count from which clicks can occur.
Relationship. Scaled by traffic_multiplier, weekly_traffic_multipliers, and performance-volume noise.
Confidence. Low; scenario size.
click_through_rate
Meaning. Conditional click probability given an impression,
Example. 0.029 means an expected 29 clicks per 1,000 impressions.
Why it makes sense. The rate connects upper-funnel exposure to measurable engagement.
Relationship. Clicks become the trial count for platform_conversion_rate.
Confidence. High for the definition; low for configured values.
platform_conversion_rate
Meaning. Diagnostic purchase probability given a click,
Example. 0.052 means an expected 5.2 diagnostic purchases per 100 clicks.
Why it makes sense. It creates Amazon Ads performance diagnostics separately from Amazon Marketing Cloud path outcomes.
Important. These purchases are never added to path purchase_count.
Confidence. High for the conditional-rate meaning; low for configured values.
cost_per_click
Meaning. Reference-currency advertising charge for each simulated click.
Example. 2.2 means 2.2 reference-currency units per click before price_multiplier.
Why it makes sense. Sponsored advertising commonly uses click-based billing.
Relationship. Exactly one of this field and cost_per_thousand_impressions must be non-null.
Confidence. Medium for billing method; low for configured price.
cost_per_thousand_impressions
Meaning. Reference-currency advertising charge for 1,000 simulated impressions.
Example. 14.0 means 14 units per 1,000 impressions before currency conversion.
Why it makes sense. Demand-side-platform inventory can be purchased on an impression basis.
Relationship. Mutually exclusive with cost_per_click.
Confidence. Medium for billing method; low for configured price.
conversion_log_odds_effect
Meaning. The touchpoint's additive effect in the path conversion response.
Example. 0.91 multiplies conversion odds by:
It does not add 91 percentage points.
Why it makes sense. Log-odds effects combine additively while the logistic function keeps the final probability valid.
Relationship. Combined with baseline_conversion_log_odds and adjacent_synergy_log_odds.
Limitation. These effects are simulator assumptions, not causal estimates from Amazon.
Confidence. Medium for the method; low for each value.