Google Shopping Ads

Sell more of what you stock,
with every item still eligible to show.

In every other paid channel the ad is something you wrote. In Shopping, the ad is your product feed — generated by a system nobody in marketing owns, and judged item by item against your own website. Products get removed individually while the campaign keeps spending, and the ads dashboard never mentions it.

Built for operators who sell products — manufacturers, ecommerce businesses, and the franchise systems and private equity portfolios running retail alongside service. Single location or national catalog, small business to enterprise.

Shopping runs inside an AI-first program here. AI-assisted feed reconciliation compares every item against the live page rather than sampling, and the product data that keeps you eligible is the same structured data AI platforms and search engines read when they decide which businesses to name.

AI systems increasingly summarize product options before a shopper reaches a retailer. The product data that keeps your items eligible in Shopping is the same machine-readable detail those systems read, which means clean feed work now pays twice — once in the auction and once in how AI describes your catalog.

Two tiers
Product-level issues remove one item quietly; account-level issues stop everything at once
Your site
Google checks your feed against your own product pages, so the site is the source of truth
Documented
Every requirement is published in the product data specification, free to read
What We Reconcile Every Cycle Against the published spec
1
Feed against the page Price and availability, item by item
2
Product-level issues The removals nobody is told about
3
Account-level exposure What would stop everything
4
Required attributes Present, correct and current

If you book jobs rather than ship products this is not your channel, and we will tell you so before you ask — most home services operators belong in local services ads and paid search instead. Where you do sell products, we run Shopping for single-location retailers through to national franchise systems and private equity portfolios, and unlike most agencies we report the eligible item count beside the performance figures.

The Removal Nobody Announces

The campaign is running. Half the catalog is not.

Shopping is the only paid channel where your inventory can be quietly withdrawn from the auction one item at a time while the campaign continues to report activity. Nothing errors, spend continues on whatever survived, and the products that stopped showing are visible only if somebody goes and looks in a different system.

A receding grid of product listings where the nearest items are lit and live while scattered items further back have quietly gone dark, with the campaign still running above them
Documented
Non-compliant product data means products may be disapproved or the account suspended
Merchant Center Help — issues
Documented
Product-level issues occur when feed data does not match the information on your website
Merchant Center Help — issues
Documented
Every required and optional attribute is published in the product data specification
Merchant Center Help — specification
Observed
Most inherited accounts have never had the feed reconciled against the live site
Allegiant observation, stated as such

Here is the structure, in Google's own words. The Merchant Center issues documentation states that if your product data and website do not comply with the policies and data quality requirements, your products might get disapproved or your account could be suspended, and that these issues can prevent your products from showing on Google.

Then it splits them into two tiers, and the distinction is the whole page. Product-level issues, per that same documentation, can occur where the product data you provide does not match the product information on your website, or where the data does not follow the product data specification. Account-level issues are a different order of problem, and the documentation is blunt about the consequence: where a review finds that your products or website do not comply with Shopping policies, the account is suspended immediately.

The two tiers fail in opposite ways, which is why one of them is dangerous. An account suspension is loud — everything stops, somebody notices within the hour. A product-level disapproval is silent. One item leaves the auction, the campaign carries on spending against the rest, impressions and clicks keep arriving, and the monthly report looks entirely normal because it is reporting on what remained.

Compounding makes it worse over time. Disapprovals accumulate. A feed that loses a handful of items each month to price drift, out-of-stock states or an attribute that changed format still delivers a healthy-looking campaign a year later, running against a fraction of the catalog it was built for. Nobody made a decision to stop advertising those products. The decision was made by a system, and nobody read the notice.

The fix is reconciliation rather than optimization. Take the feed, take the live site, and compare them item by item. Where they disagree, the site is the source of truth as far as Google is concerned, so the feed is what changes. This is the same discipline we apply to charged queries on paid search and to billed placements on display — go and read what actually happened rather than what the dashboard chose to show.

Two Records, One Product

Your feed says one thing. Your website says another.

Every product exists twice — once as a row in the feed and once as a page on your site. Google reads both. Where the two disagree, the item is at risk, and the fields that drift are almost always the same two.

Two translucent panels facing each other, the feed record on one and the website product page on the other, with matching fields joined by threads and mismatched fields left unconnected
Unreconciled

The feed drifts away from the site

  • Price changed on the site, not in the feed. A promotion, a supplier increase, a manual edit nobody pushed downstream.
  • Stock ran out and the feed did not follow. The item keeps advertising and keeps disappointing.
  • Identifiers missing or malformed, so items cannot be matched to the product they actually are.
  • Images that no longer resolve after a site migration nobody connected to the feed.
  • Disapprovals accumulating in a system nobody with ad account access ever opens.
Reconciled

Both records agree, every cycle

  • Price and availability checked against the live page, not against the last export.
  • Stock state flows through automatically, with a manual check where the connection is fragile.
  • Identifiers validated against the specification before an item is submitted at all.
  • Image links tested, because a broken one removes the item rather than degrading it.
  • Issues read on a standing cycle and cleared, with the count reported whether or not it is flattering.
§
Price and availability are the two fields that drift, and they are the two Google cares most about. That is not arbitrary — they are the fields a shopper acts on. A person who clicks an advertised price and finds a different one at the destination has been misled, whether or not anybody intended it, and the platform treats that as its problem to prevent rather than yours to explain.
Four Things The Feed Has To Get Right

All published, all checkable, all owned by somebody outside marketing.

These come from Google's own specification. What makes them hard is not the rules — it is that the data originates in an inventory or ecommerce system managed by someone who has never seen a Shopping campaign.

1
Price
Must match the landing page
Checked againstYour live product page
Failure modeItem removed

Google publishes the requirements for the price attribute, and the practical rule is that what you submit has to be what a shopper finds when they arrive.

The drift is almost never deliberate. Somebody runs a promotion in the store platform, the feed regenerates on a schedule that does not align, and for a window the two records disagree — which is exactly the window a disapproval lands in.

What we do with it
  • Align the feed refresh with how often prices actually change, not with a default schedule.
  • Spot-check advertised prices against live pages as a visitor would see them.
2
Identifiers
GTIN and product identity
EstablishesWhich product this is
Failure modePoor matching or removal

The GTIN attribute is how Google establishes that your listing and a manufacturer's product are the same thing, which is what lets your item appear against the right searches.

Missing or invented identifiers are the most common inherited defect we find, usually because the field was optional in the store platform and somebody filled it with a placeholder years ago.

What we do with it
  • Validate identifiers against the specification before submission rather than after rejection.
  • Where a product genuinely has no identifier, declare that properly instead of inventing one.
3
Images
Image link
Must beReachable and compliant
Failure modeItem removed

The image link attribute has requirements about what the image may contain and whether it resolves. A Shopping ad is mostly an image, so this is not a cosmetic field.

Site migrations are where this breaks at scale. Image paths change, the feed keeps pointing at the old ones, and a large share of the catalog leaves the auction over a weekend nobody was watching.

What we do with it
  • Test image links as part of the reconciliation rather than assuming they resolve.
  • Re-run the check immediately after any site change that touches media paths.
4
The destination page
Product page availability
CheckedBy Google, directly
Failure modeItem or account issue

Google documents how to fix a product page that is unavailable, which exists as guidance because the destination is checked rather than assumed.

This is where Shopping meets ordinary site health. A page that is slow, blocked, geo-restricted or intermittently erroring is a page Google may be unable to verify, and the item goes with it.

What we do with it
  • Treat product page availability as a paid media dependency, not only a web issue.
  • Check what a crawler sees rather than what loads in your own browser session.
Where A Feed Actually Fails

Five stages, and two completely different kinds of failure.

Understanding the pipeline is what makes the difference between fixing a symptom and fixing a cause. The same visible outcome — products not showing — has origins at four different stages, and only one of them lives inside the ad account.

The pipeline, and the two failure tiers

Product data starts in an inventory or ecommerce system. It is exported or generated into a feed. The feed is submitted to a merchant account, where it is validated against the specification and against your live site. The campaign then bids using whatever survived that validation. Finally an ad shows.

A product-level issue removes one row and nothing else changes. The campaign continues, the budget continues, and the report is accurate about the items that remain. This is the failure that runs for months, because there is nothing to notice — the only evidence sits in a different interface that the person reviewing ad performance has no reason to open.

An account-level issue stops everything. Google's documentation states that where a review finds products or a website not complying with Shopping policies, the account is suspended immediately. That is severe, and it is also survivable precisely because it is impossible to miss. The dangerous tier is the quiet one.

Which is why our first pass on any Shopping account is not the campaign. It is the item list — how many products were submitted, how many are actually eligible to show, and what the difference between those two numbers is made of.

  • Inventory system — where the data is really authored
  • The feed — a generated artifact, on a schedule
  • Merchant account — validation against the spec and your site
  • The campaign — bids only on what survived
  • Product-level issues — quiet, cumulative, months-long
  • Account-level issues — loud, total, immediately obvious
An engineering-style elevation of the product feed pipeline from inventory system to shown ad, with the item-level and account-level failure points marked at the stages where each occurs
How We Run A Shopping Account

Count the eligible items first, then worry about bidding.

Bid strategy decides how you compete for the items that are showing. It cannot do anything at all about the items that are not, and on most inherited accounts that second group is larger than anyone has checked.

1

Before touching the campaign

Count what is eligible
  • Products submitted counted against products actually eligible to show
  • Every outstanding issue read and grouped by cause rather than by count
  • Account-level exposure identified before it becomes a suspension
  • The gap between the two numbers reported plainly, whatever it is
You find out how much of your catalog is actually advertising.
2

Fix at the source

The system, not the feed
  • Recurring issues traced back to the inventory system that authored them
  • Feed refresh aligned with how often the underlying data really changes
  • Identifiers and images validated before submission, not after rejection
  • Product page availability treated as a paid media dependency
The same item stops being disapproved every month.
3

Then run it, and keep counting

Standing reconciliation
  • Feed reconciled against the live site on a standing cycle
  • Free listings checked alongside paid placement where eligible
  • Reported revenue reconciled against orders that actually reached the business
  • Eligible item count reported every month beside the performance figures
Performance is judged against a catalog that is fully in the auction. The reconciliation habit runs through paid search too.
What We Do On A Shopping Account

Six things that happen outside the campaign interface

Almost none of this work happens where a partner expects to see it. That is precisely why it goes undone, and why an account can look well managed while a third of the catalog sits out of the auction.

Submitted counted against eligible

The two numbers are reported side by side every month. A campaign performing well against a fraction of the catalog is not a campaign performing well.

Source: Merchant Center Help — issues

Feed reconciled against the live site

Price and availability compared item by item against the page a shopper would land on, because that page is what Google checks against.

Source: Merchant Center Help — product data specification

Issues traced to their source system

A disapproval fixed in the feed returns next month. A disapproval fixed in the system that generates the feed does not, which means the useful work usually sits outside marketing.

Related: the same "fix the source, not the symptom" order applies in CRM work.

Product pages checked as a crawler sees them

Availability, speed and blocking are checked from outside a logged-in session, because a page that works for you may not be verifiable by Google.

Source: Merchant Center Help — product page unavailable

Structured data kept consistent with the feed

Product markup on the page and the feed record describe the same item. Where they disagree, you have given two answers to the same question.

Source: Google Search Central — product structured data

Free listings checked alongside paid

Where eligible, products can appear in unpaid surfaces from the same data. It is the same feed doing more work, and it is regularly left unconfigured.

Source: Merchant Center Help — free listings

Why Operators Move The Account To Us

We report the products that are not showing.

Shopping reporting almost always covers the items that ran. The useful number is the one beside it — how many products you submitted and how many are actually eligible — because that gap is the ceiling on everything else in the account.

We will also tell you when Shopping is not your channel. If you book jobs rather than ship products, Shopping has nothing to advertise, and an agency proposing it is selling a package rather than reading your business. That is most of the operators we work with, and we say so.

Shopping sits alongside paid search, display and remarketing in the wider digital program.

25
Years buying and building media for operators
US & CA
Partners served across both countries
Credentials
Google Partner
Verified
Semrush Certified Agency
Verified
Certified CallRail Agency
Verified
Inc. Power Partner
2024 and 2025
50PROS Top 10 Global
Awarded
BBB A+ Accredited
Accredited
What A Product Record Is Made Of

Every attribute is a reason to show, or a reason to be removed.

A feed row is not one thing. It is a stack of attributes, each with published requirements, and any one of them can take the item out of the auction on its own.

A single product record shown as a cube separated into its required attribute layers, with each supplied attribute in place and one layer floating apart because it disagrees with the website
Attribute How it is usually treated Against the specification What it actually decides
Price Whatever the export produced Must agree with the live page a shopper lands on, at the moment they land
Availability Updated when someone remembers Drives whether the item should be in the auction at all right now
Identifier Optional field, often a placeholder Establishes which product this is, which decides what searches it matches
Image link Set once at launch Must resolve and comply — a Shopping ad is mostly the image
Landing page Assumed to work Checked directly by Google, so site health is a paid media dependency
The disapproval notice Lives in a system nobody opens The only place the removal is recorded — and it is not the ads dashboard

The last row is why this page exists. Every other line can be fixed once somebody knows about it. That one is about whether anybody knows at all, and on most inherited accounts the answer has been no for a long time.

What We Decline To Sell

Four Shopping line items you can stop paying for.

One is work on the wrong half of the problem. One is a number reported without its denominator. One is a channel sold to a business that cannot use it. The fourth is a fix that has to be bought again every month.

Bid optimization on an unreconciled feed. Refining how you compete for the items that are showing, while a share of the catalog sits outside the auction, is careful work on the smaller half of the problem. Reconcile first; the bidding matters afterwards and not before.

Performance reported without the eligible item count. Return on ad spend is a ratio, and it is perfectly capable of looking healthy while most of your products are not advertising at all. A Shopping report without the submitted-versus-eligible figure beside it is missing its denominator.

Shopping sold to a service business. This channel advertises products with prices, images and stock. If you book jobs rather than ship items, there is nothing for it to do, and an agency proposing it has not read your business. For most home services operators the answer is local services ads or paid search instead.

Feed fixes that never touch the source. Correcting a disapproved item in the feed makes it eligible until the next export overwrites the correction. Recurring monthly feed cleanup is a subscription to a problem, and the actual repair usually belongs in the system that generates the data.

The pattern beneath all four: the work that is visible in the ad account is not the work that decides the outcome.

Evidence note. Platform requirements described on this page are drawn from Google's own published Merchant Center and Search Central documentation, linked at the point of use and read live on the review date in the byline. Currency caveat: the product data specification, Shopping policies and Merchant Center interface are revised continuously and without notice, and requirements differ by country and program — read the linked page against your own account before acting on it. Google's documentation is vendor documentation, authoritative for how Google's own systems behave and not independent research; we treat it that way. Statements about what we observe in inherited accounts are Allegiant observations and are labeled as such in the text. We describe what the documentation states about disapproval and suspension and do not predict what will happen to any particular account. No return on ad spend, conversion rate, click-through rate, impression share, cost-per-click or disapproval-rate figures appear anywhere on this page, and no pricing is quoted, because those figures are either vendor-panel estimates or specific to accounts that are not yours.
Questions Operators Ask

Google Shopping, answered against the documentation

Every answer below links to Google's own published documentation, so you can check it without taking our word for it.

Merchant Center requirements change continuously and differ by country and program. Read the linked page against your own account before acting on anything here.
Why are only some of our products showing?+

Almost always product-level issues nobody has read. Google's Merchant Center issues documentation states that where product data and website do not comply with the policies and data quality requirements, products might get disapproved, and that product-level issues can occur where the data you provide does not match the information on your website. The campaign keeps running on whatever remains, so nothing looks broken. Count products submitted against products eligible — the gap is the answer. Same shape of silent failure as on website maintenance.

What is the difference between an item issue and an account issue?+

Scale and visibility. A product-level issue removes one item and leaves everything else running, which is why it can persist for months unnoticed. An account-level issue is different in kind — Google's documentation states that where a review reveals products or a website not complying with Shopping policies, the account is suspended immediately. The suspension is severe and survivable because it is impossible to miss. The quiet tier is the one that costs more over a year, the same way untracked calls do on landing pages.

Does our website price have to match the feed exactly?+

Yes — what you advertise has to be what a shopper finds on arrival. Google publishes the requirements for the price attribute, and the Shopping ads policies disallow representing your products in a way that is not accurate, realistic and truthful. The usual cause of a mismatch is timing rather than intent: a promotion goes live in the store platform and the feed regenerates on a different schedule. Align the refresh with how often prices actually change. The same claim-accuracy principle runs through print advertising.

Our products lost their images after a site migration. What happened?+

The feed kept pointing at image paths that no longer exist. The image link attribute has to resolve and comply, and a Shopping ad is mostly its image, so a broken link removes the item rather than degrading it. Migrations break this at scale over a weekend nobody is watching. Re-run the image check immediately after any change touching media paths, and treat it as part of the launch checklist rather than something to notice later. Deploy-time checking is covered on website maintenance.

Do we need GTINs for everything?+

Where the product has one, yes, and where it genuinely does not you declare that properly rather than inventing a value. The GTIN attribute documentation explains how the identifier establishes which product a listing refers to, which is what lets it match the right searches. The most common inherited defect we find is a placeholder somebody entered years ago because the field was optional in the store platform. Validate against the product data specification before submission rather than after rejection — the same "check the record first" habit we apply to directory listings.

Can we appear in Shopping without paying?+

In some cases, from the same data you already maintain. Google documents free listings for products, which let eligible items appear in unpaid surfaces. It is regularly left unconfigured because it produces no line on an invoice, which is exactly why it is worth checking. It also raises the stakes on feed quality, since the same record now serves two purposes. Keeping the on-page product structured data consistent with the feed matters here too, as it does across our technical SEO work.

Our feed gets fixed and the same errors come back. Why?+

Because the correction was made in the feed rather than in the system that generates it. The next export overwrites the fix and the disapproval returns, which turns feed cleanup into a monthly subscription to the same problem. The durable repair usually sits in the inventory or ecommerce platform, owned by somebody outside marketing, which is why it takes longer and gets skipped. Validate against the product data specification at the source. This is the same "fix the source" order we apply in CRM work.

Google says our product page is unavailable, but it loads fine for us. Why?+

Because you are not testing what Google tests. Its guidance on product pages that are unavailable exists because the destination is verified directly rather than assumed. A page can load perfectly in your own browser and still be blocked, geo-restricted, intermittently erroring or too slow when fetched from outside a logged-in session. Check as an anonymous visitor from outside your network, and treat product page health as a paid media dependency rather than only a web concern — see website performance.

What should a Shopping report actually contain?+

Performance figures, and beside them the number that gives those figures meaning: products submitted against products eligible to show. Return on ad spend is a ratio and it can look healthy while most of the catalog sits out of the auction. Add the outstanding issue count grouped by cause, and any account-level exposure, and you have a report that tells you what to do next. Google's Shopping ads overview covers the mechanics; reconciling reported revenue against orders that reached the business follows the discipline on paid search.

We are a service business. Should we run Shopping ads?+

Almost certainly not, and we will say so before you ask. Shopping advertises products with prices, images and stock levels. If you book jobs rather than ship items there is nothing for the channel to list, and an agency proposing it is selling a package rather than reading your business. For most home services operators the honest answer is local services ads and paid search. Where you do sell products alongside a service, Shopping can work for that part of the business only, under the same Shopping policies as any retailer.

Find out how much of your catalog is actually advertising.

The A.R.C. Report covers your whole marketing position, and where Shopping is in the plan we count the products you submitted against the products actually eligible to show, group the outstanding issues by cause, and flag anything that could take the account down rather than one item. Findings are yours whether or not we work together.

What the review covers on a Shopping account
  • Products submitted counted against products eligible to show
  • Outstanding issues grouped by cause rather than by volume
  • Feed price and availability reconciled against the live site
  • Identifiers and image links validated against the specification
  • Product page availability checked as a crawler would see it
  • A straight answer if Shopping is the wrong channel for your business

Explore the wider program: all services, paid search, display advertising, remarketing and the A.R.C. Report.

Request an A.R.C. Report

Tell us the store and the catalog you run, and we will tell you how much of it is currently in the auction.

No cost, no commitment. We will follow up by email or phone to walk you through the findings.