The 4Ps of the marketing mix in Amazon Marketing Cloud: product, price, place and promotion as one query each
In short: The 4P marketing mix model splits the marketing mix into product, price, place and promotion. In Amazon Marketing Cloud each P becomes its own query: product through `tracked_asin` and `purchase_unit_price`, place through `postal_code` and `site`, promotion through `ad_product_type` and `campaign`. Unit price and revenue of a purchase you read from `purchase_unit_price` and `total_product_sales`.
The four Ps and their AMC question
The 4Ps are a thinking grid, not a measurement method: product, price, place and promotion. In Amazon Marketing Cloud each P is one query on one table. The table below names for every P the question it answers, the field that carries the answer and the table that field lives in.
The four Ps are deliberately kept apart. One query that mixes all four P in a single row returns either empty fields or numbers that do not share the same population. One query per P is the fastest way to find out whether a signal holds at all.
| P | Guiding question | AMC table | Field |
|---|---|---|---|
| Product | Which advertised product sells? | amazon_attributed_events_by_conversion_time | tracked_asin |
| Price | Which unit price sits in a purchase? | amazon_attributed_events_by_conversion_time | purchase_unit_price |
| Place | In which region do people buy? | amazon_attributed_events_by_conversion_time | postal_code |
| Promotion | Which program and campaign advertise? | sponsored_ads_traffic, dsp_impressions | ad_product_type, campaign |
Product: which ASIN sells?
For the product P Amazon provides the measured sales per advertised ASIN. The field tracked_asin holds the advertised product, total_purchases counts the purchases attributed to it and total_product_sales the revenue from them. You are therefore not reading the catalogue of your brand but the purchases Amazon attributes to an advertised product.
If you want to know which product carries your campaign, that is the first look: the ASIN with the most attributed purchases. For a comparison between programs you additionally need the field ad_product_type, because Amazon DSP appears in this table with NULL in that field.
SELECT
tracked_asin,
SUM(total_purchases) AS total_purchases,
SUM(total_product_sales) AS total_product_sales
FROM
amazon_attributed_events_by_conversion_time
GROUP BY
tracked_asin Price: which unit price sits in the purchase?
The price P is not available in Amazon Marketing Cloud as a separate table value but inside the purchase event. purchase_unit_price names the unit price of a purchase, total_product_sales the total revenue of the same event. You calculate the price per ASIN by dividing the revenue by the number of purchases.
A price comparison over several months therefore needs a grouping by tracked_asin and conversion_event_date. That way you see whether the price per ASIN stays stable over time or moves with the campaign mix. What you conclude from it remains your own assessment: Amazon delivers the measured values, not a recommendation for a target price.
SELECT
tracked_asin,
AVG(purchase_unit_price) AS average_unit_price,
SUM(purchases) AS purchases
FROM
amazon_attributed_events_by_conversion_time
GROUP BY
tracked_asin Place: in which region do people buy?
The place P has two levels, and the two are often confused. The geographic level is the field postal_code at postal-code level; Amazon applies the HIGH level as its aggregation threshold, so a query without matching filter values returns empty rows. The second level is the placement: the field site shows on which website, app or streaming platform the ad ran, supply_source where the inventory came from.
Keeping place and placement apart is the difference between a regional analysis and an inventory analysis. A regional query groups by postal_code, an inventory query by site and supply_source. Only both levels together support the statement whether a region or a placement carries the revenue.
SELECT
postal_code,
SUM(total_purchases) AS total_purchases
FROM
amazon_attributed_events_by_conversion_time
GROUP BY
postal_code Promotion: which program and campaign advertise?
The promotion P is the program mix of your advertising in AMC. Sponsored ads are in the table sponsored_ads_traffic with the field ad_product_type (values like sponsored_products) and the field campaign; Amazon DSP is in dsp_impressions with site and supply_cost. Both programs write different units: spend is in microcents, total_cost in millicents.
Anyone comparing both programs in one table has to convert the units first, otherwise euros are compared with hundred-thousandths. For a comparison of ad programs the context also matters: sponsored_ads_traffic knows the search term and the match type, dsp_impressions knows the site and the inventory source.
SELECT
ad_product_type,
campaign,
SUM(impressions) AS impressions,
SUM(clicks) AS clicks,
SUM(spend) AS spend
FROM
sponsored_ads_traffic
GROUP BY
ad_product_type,
campaign What you need to know before the first run
Every query in this guide has a limit you should know before the first run. Amazon applies aggregation thresholds: a row only carries values once a certain number of distinct users is reached, otherwise it stays empty. For the field postal_code Amazon names the level HIGH, for tracked_asin and ad_product_type the level LOW. The identifier field user_id carries the highest level VERY HIGH; it must therefore not appear in a final SELECT.
An empty result therefore usually does not mean "no revenue" but "too few users". The price P, the place P and the promotion P are one query each; only once all three return rows do you compare the Ps with each other.
How this differs from Amazon Marketing Mix Modeling
The model in this overview is a thinking grid, not a model: Amazon itself does not calculate your data along the four Ps. Amazon Marketing Mix Modeling is a separate product with three delivery paths and answers the question of budget allocation across all channels.
Both can be combined: you read the four Ps out of your event data with queries and feed the results into a marketing mix model. You do not need a second data source for that, because both evaluations rest on the same events.
Frequently asked questions
What is the difference between the 4Ps and a marketing mix model?
The 4Ps are a thinking grid for four questions about your data: product, price, place and promotion. A marketing mix model allocates budget across channels to estimate the effect of each channel. You need the 4Ps as preparation, because they clarify which numbers exist for each question at all.
Which table answers all four questions?
For product, price and place it is amazon_attributed_events_by_conversion_time, because purchases, unit price and postal code sit on the same event there. For promotion you additionally need sponsored_ads_traffic for sponsored ads and dsp_impressions for Amazon DSP, because only those tables know the programs that delivered the advertising.
Why does my query return nothing?
Most of the time it is the aggregation threshold. Amazon returns rows only once a certain number of distinct users is reached, and for postal_code the level HIGH applies. So check the time range and the filters first, then the threshold: an empty result does not prove missing revenue.
Are the costs in euros?
The cost fields carry different units: spend in sponsored ads is in microcents, total_cost in Amazon DSP in millicents, supply_cost in microcents. The field currency_name names the currency. Convert the units before you compare two programs.