How to export Amazon Ads data to BigQuery (and a faster alternative)

Admaxxer is a marketing analytics and attribution platform for DTC brands and SaaS companies. This guide is the straight answer for ecommerce teams running Amazon Ads (Sponsored Products, Sponsored Brands, Sponsored Display): how do I get my Amazon Ads data into BigQuery? The goal underneath is almost always the same — join ad spend to sales, build dashboards, and stop manually pulling reports out of the Amazon Ads console.

We'll cover the legitimate routes honestly — the native report download, the Amazon Ads reporting API, and generic third-party connectors — citing Amazon's and Google's official documentation. Then we'll show you the route most teams actually want: skip the pipeline and get attribution plus live dashboards without a warehouse to build and maintain.

This is an ECOM/DTC guide (orders, sales, AOV, ROAS/ACOS). If you run a SaaS product the export mechanics are the same — though note that Amazon Ads is overwhelmingly an ecommerce channel, so the SaaS case is rarer here; deltas are flagged inline where they apply.

The honest truth about "Amazon Ads to BigQuery"

There is no native, first-party connector that pipes Amazon Ads data into BigQuery on its own. Google's BigQuery Data Transfer Service ships managed connectors for a set of sources — including a native Google Ads transfer — but Amazon Ads is not one of its native sources. So getting Amazon Ads data into BigQuery is always one of three things: download reports by hand, pull them through Amazon's reporting API and load them, or pay a third-party connector to do it for you.

(Separately, Amazon offers Amazon Marketing Cloud — a clean-room product for advanced analytics — but that is a different tool with its own access model, not a BigQuery export connector. For most "get my Amazon Ads numbers into BigQuery" needs, the three routes below are what apply.)

Route 1: Native report download from the Amazon Ads console (CSV)

The Amazon Ads console lets you generate and download reports — by campaign, ad group, targeting, search term, and more — across spend, impressions, clicks, sales, and ACOS, then export them.

When this is the right call: a one-off analysis, a monthly export, or a quick check. You download the file and load it into a BigQuery table manually.

The catch: it's manual and static. BigQuery doesn't know the file exists until you upload it, there's no auto-refresh, and Amazon's reports come in several types you have to assemble yourself. For a recurring dashboard, this is a chore that quietly lapses.

Route 2: Amazon's reporting API + your own pipeline

For automated loads, Amazon exposes performance data through its reporting API. The current version is documented in the Amazon Ads reporting v3 overview, with a step-by-step get-started guide. The model is asynchronous: you request a report for a given ad product, report type, columns, and date range; Amazon generates it; you poll for completion and then download the result.

The work looks like this:

1. Register for API access and complete Amazon's Login-with-Amazon authorization to get a token for the advertising account, per the Amazon Ads API overview.
2. Request a report for each ad product and report type you need (Sponsored Products, Sponsored Brands, Sponsored Display each have their own report configurations).
3. Poll for completion, then download the generated report file when it's ready.
4. Load it into BigQuery — via load jobs or streaming inserts, per Google's BigQuery documentation.
5. Schedule, retry, and monitor the whole loop, and backfill missed runs, so the warehouse stays current.

When this is the right call: you have a data team, you want full control over the schema, and Amazon is one of several sources flowing into a warehouse you already run.

The catch: you now own an asynchronous, multi-report-type pipeline. Tokens expire, Amazon versions its reporting (note the existence of a v2-to-v3 migration in Amazon's own docs — APIs move), and a silent failure leaves your dashboards stale. The build is straightforward; the upkeep is the real cost.

SaaS note: rare for Amazon Ads, but if it applies, the only change is joining the spend to subscriptions/MRR downstream rather than product sales.

Route 3: A generic third-party connector

Several off-the-shelf connectors (described generically — evaluate each on its current capabilities and pricing) will pull Amazon Ads data on a schedule and load it into BigQuery for you, sparing you the asynchronous report dance.

When this is the right call: you want automation without building it and you're comfortable adding a paid connector.

The catch: a connector lands raw Amazon Ads rows. It won't, by itself, reconcile ad spend against total sales, separate ad-attributed from organic sales, or give you a blended cross-channel view. The attribution work — the hard part — still sits with you.

The metric trap: ACOS in a warehouse still isn't the full story

There's a specific reason raw Amazon Ads rows in BigQuery mislead more often than other channels: Amazon's headline efficiency metric, ACOS (advertising cost of sales), only counts ad-attributed sales. It tells you the return on the orders the ads got credit for — not the effect of that ad spend on your total sales, including the organic and brand-search sales that advertising helped lift.

That gap is why experienced Amazon advertisers also watch TACoS (total advertising cost of sales — ad spend against total revenue). A warehouse full of ad-attributed report rows can show a healthy ACOS while your total business tells a different story, in either direction:

• A campaign with a "bad" ACOS may be defending branded search or seeding new-to-brand customers whose later organic orders never show up in the ad report.
• A campaign with a "great" ACOS may simply be harvesting sales you'd have won organically anyway — spend that looks efficient but isn't incremental.

Raw rows in BigQuery don't resolve this for you. To judge whether Amazon spend is actually growing the business, you have to bring total sales alongside ad spend and look at the blended picture — which is precisely the join a single-source export leaves you to build, and precisely what an analytics platform that connects spend and sales does for you. The honest takeaway: exporting the ACOS rows is easy; turning them into a real spend decision is the work.

Or skip the pipeline entirely

Ask what you were really after. Teams rarely want Amazon rows in a warehouse for their own sake — they want to know what Amazon Ads spend returned in sales, see it on a live dashboard, and weigh it fairly against Meta, Google, TikTok, and Pinterest. The warehouse was a detour.

Admaxxer does that job with no pipeline. Connect Amazon once on the Amazon Ads integration page (and your store on the Shopify integration page if you also sell off-Amazon), and your spend and sales line up automatically.

What you get instead of raw rows:

• Attribution, not just data. Amazon spend joined to sales, so you see ROAS and blended MER directly — not impressions and ACOS you still have to model.
• A live dashboard that refreshes itself, instead of reports you re-download and re-upload.
• A fair cross-channel view — Amazon alongside Meta, Google, TikTok, and Pinterest on one screen, which a single-source BigQuery export can never give you.
• Margin-aware decisions. Pair it with your average order value and AOV so you're steering on contribution, not just top-line ACOS.

If a raw Amazon warehouse genuinely serves a separate data-science need, keep one of the routes above for it — but to measure and steer spend, connecting once beats owning a pipeline. See pricing for what's included, browse the documentation, or read how we frame marketing acquisition.

Choosing your route

• One-off number? Native report download from the Amazon Ads console (Route 1).
• Have a data team and a warehouse? Amazon's reporting API (Route 2) or a connector (Route 3) to land raw rows.
• Want to actually measure and steer Amazon spend against sales, fast, without owning a pipeline? Connect it to Admaxxer and skip the warehouse.

Most teams realize they were building a pipeline to reach an answer a connect-once tool delivers directly. Choose the route that matches your real goal, not the one the search box implied.

FAQs