ARMR
ARMR
AmazonWalmartAdvertisingCreativeResultsAboutBlog Book a growth call
Advertising•20 November 2026•6 min read•

Retail Media Reporting Across Amazon and Walmart: Build vs. Buy

Retail Media Reporting Across Amazon and Walmart: Build vs. Buy

TL;DR: If logging into Amazon Ads, Walmart Connect, and a DTC analytics dashboard every week just to answer "what's actually working" sounds familiar, spreadsheet reconciliation is already costing more than the hours it takes. A custom reporting layer earns its cost once contribution margin by channel needs to live in one place. Standard KPIs across one or two channels still favor buying a tool.

A brand selling on Amazon and Walmart, with its own DTC site running alongside both, ends up with three separate stories about how last week went. Amazon Ads reports Total Sales attributed to an ad click. Walmart Connect reports GMV on a different attribution window. The DTC dashboard in Shopify or GA4 shows a blended ROAS built on yet another model. None of these numbers were designed to sit next to each other, and nobody at the marketplace or the ad platform is going to reconcile them for you.

For a while, a spreadsheet is enough. Someone pulls three exports every Monday, lines the columns up by hand, and the team gets a rough picture of what's working. That picture gets less reliable as the catalog, the channel count, and the ad spend all grow, and the person doing the reconciliation becomes the only one who actually understands how the numbers were stitched together.

At some point the question stops being "can we keep doing this by hand" and becomes "what replaces it." That's where the build-vs-buy decision actually starts, and it's worth being honest about scope before picking a side: most brands that go this route end up working with a team that specializes in retail software development rather than trying to bolt a cross-platform data pipeline onto an internal engineering backlog that's already busy with the storefront.

Why Amazon, Walmart, and your own site don't agree on what "sales" means

The mismatch isn't a bug, it's a byproduct of three platforms building metrics for their own use case. Amazon Ads' Total Sales attributes a purchase to an ad click within its own lookback window, the same kind of window that drove Amazon to build its own cross-channel attribution reporting, because brands kept asking how off-Amazon marketing influenced on-Amazon sales. Walmart Connect's GMV counts gross merchandise value tied to an impression or click under its own separate rules. A DTC blended ROAS, built in GA4 or a Shopify app, usually mixes a shorter last-click window with whatever the ad platform's pixel reports back. None of the three numbers measure the same event the same way, so adding them together produces a total nobody should actually trust.

PlatformPrimary metricAttribution windowWhat it misses
Amazon AdsTotal Sales per clickPlatform-defined lookback windowProduct cost, fulfillment fees, margin
Walmart ConnectGMV per impression or clickSeparate window, not aligned to Amazon'sOrganic sales lift, cross-channel halo
DTC site (GA4 / Shopify)Blended ROASShorter, last-click weightedMarketplace-only demand, platform ad fees

The weekly reconciliation tax most brands don't notice until it's too late

Manually joining three exports every week rarely shows up as a line item anywhere, which is exactly why it goes unexamined. It's a recurring, five-to-ten-hour-a-week cost buried inside someone's regular job, and it scales badly: a new SKU line or a new marketplace means rebuilding part of the spreadsheet rather than adding a row. A handful of signs mean a brand has already outgrown it.

  1. More than two people touch the reconciliation spreadsheet. Once ownership is shared, version drift and silent formula breakage become routine.
  2. A pricing or budget decision got made on numbers older than 48 hours. Platform exports lag, and a spreadsheet workflow often lags further behind that.
  3. Contribution margin by product by channel isn't visible anywhere in one place. ROAS and ACOS are everywhere; actual profitability per channel usually isn't.
  4. Launching a new marketplace or channel means rebuilding the spreadsheet from scratch. If growth keeps breaking the reporting process, the process is the constraint.

What a custom retail-media reporting layer actually replaces

This isn't a full business-intelligence platform rebuild. It's a thin ingestion and normalization layer that pulls the Amazon Ads API, the Walmart Connect API, and the DTC analytics source into one shared schema, then maps each platform's native metric into a common definition the team agrees on. What that buys back: apples-to-apples contribution margin by channel instead of three incompatible dashboards, same-day data instead of waiting on next week's platform export, and one alerting surface instead of three separate logins nobody checks consistently.

When buying an off-the-shelf tool is still the right call

If the team only needs standard KPIs, ROAS, ACOS, and blended CAC, across one or two channels, and there's no proprietary cost or margin feed from an ERP, 3PL, or wholesale system that needs to flow into the numbers, buying is still the better answer. The same applies when there's no need for an external-stakeholder-facing portal and no automation trigger tied to the reporting output. A Pacvue-style or Flywheel-style aggregator, or a BI tool layered on top of native platform exports, covers that case well, and building a custom layer to replace it would be solving a problem that's already solved.

Is a hybrid approach the realistic middle ground for most mid-market brands?

For most brands doing somewhere between five million and fifty million dollars in annual GMV, yes, a hybrid is the realistic default, not full custom build and not pure off-the-shelf. The pattern that works: keep a platform or BI tool handling the standard eighty percent (the common KPIs every brand tracks the same way), and build the custom twenty percent that's actually unique, usually a normalized margin-by-channel view, a vendor or stakeholder-facing portal, or an automated budget-reallocation trigger tied to the numbers. That split keeps the build small and defensible instead of trying to replace an entire reporting stack on day one.

What it costs and how long it takes to build the custom layer

There's no honest single number here without knowing the scope. A narrow ingestion-plus-normalization build covering two or three channels and one dashboard is realistically a matter of weeks of engineering time, not months. Costs and timelines scale mainly with two variables: how many source APIs need to be integrated, and how much historical data needs to be backfilled rather than just captured going forward. A brand adding a fourth or fifth channel, or wanting a year of historical trend data rebuilt from scratch, should expect that range to stretch toward a multi-month build rather than a few-week one.

How do you decide which metrics the custom layer needs to normalize first?

Start with whichever metric currently causes the most disagreement in budget meetings, not whichever metric happens to be easiest to pull via API. For most brands running this comparison, that's contribution margin by product by channel: it's the number that should drive where the next marketing dollar goes, and it's usually the one missing from every native platform dashboard. Normalizing the easy metrics first and leaving the contentious one for later just delays the point where the reporting layer actually earns its keep.

Frequently asked questions

Do I need a data engineer to build a custom retail-media reporting layer?

For the first version, usually one engineer who can work with REST APIs and a database schema is enough. The job is ingestion and normalization, pulling data from the Amazon Ads API, the Walmart Connect API, and a DTC analytics source into one shared schema, not building a full data platform. Teams that try to staff it like a data-engineering department from day one tend to over-scope v1 before anyone has proven the reporting layer is worth maintaining.

Can I start with just Amazon and Walmart, then add my DTC site later?

Yes, and it's the lower-risk order to build in. Amazon and Walmart both expose structured ad and sales APIs, so normalizing those two first gives a working cross-marketplace view without touching the DTC analytics stack. Adding the DTC source later is an additional ingestion job against an existing schema, not a redesign of the reporting layer.

What's the difference between a reporting layer and a full BI platform?

A reporting layer is narrow on purpose: it ingests a handful of known sources, normalizes a specific set of metrics, and feeds one or two dashboards or exports. A full BI platform is built to answer questions nobody has asked yet, with ad hoc querying, broad data-source support, and a much larger surface to maintain. Most brands only need the former; reaching for the latter is how a two-week project turns into a two-quarter one.

How do I pick which off-the-shelf tool to buy if I'm not ready to build?

Start from the metric the team actually needs, not the tool with the longest feature list. If standard ROAS and ACOS across Amazon and Walmart cover what's needed, a lighter aggregator is enough. If the deciding factor is contribution margin by product by channel, confirm the tool can ingest a cost or margin feed from outside the ad platforms before buying, since that's the exact capability most off-the-shelf reporting tools lack.

What happens to the custom layer when Amazon or Walmart changes their ad API?

The ingestion job for that specific platform needs an update; the rest of the schema and the downstream dashboard usually don't. This is the ongoing maintenance cost that comes with owning a custom build rather than a subscription tool, and it's a real, recurring line item, not a one-time cost that's done once version one ships.

Is contribution margin by channel really more useful than ROAS?

ROAS tells you whether an ad was efficient relative to its own spend. Contribution margin by channel tells you whether a channel is actually profitable once product cost, fulfillment, and platform fees are factored in, which is the number that should drive a budget-reallocation decision. A channel can show a strong ROAS and still be the wrong place to put the next marginal dollar once margin is in the picture.

For more on where the build-vs-buy line sits for marketplace data in general, see our breakdown on building an Amazon analytics dashboard, and for a different tool category facing the same decision, multi-channel order management software.

Working on this yourself? If any of the above sounds like your account, the fastest next step is a free audit. Tell us about your brand and we will show you where to start.