Ecommerce Google Ads management
Ecommerce advertising is not simply a matter of generating more orders. Your product feed, Merchant Center data, conversion tracking, campaign structure, order value and margins all decide whether spend produces profit or just expensive revenue. I manage that whole system with the business economics in view, not just the account.
I'm Vijay Bhabhor, an independent Google Ads consultant with 17+ years in digital marketing. I manage ecommerce accounts personally rather than passing them to a junior account manager, which is the main practical difference between working with me and working with an agency.
Where the money actually goes
Ecommerce advertising is not harder than lead generation — it is a different optimisation problem. In lead generation the account and landing page are most of the system. In ecommerce, decisions interact continuously with products, inventory, pricing, margin and checkout — all of which sit outside Google Ads and none of which the account can see.
| Dimension | Lead generation | Ecommerce |
|---|---|---|
| The conversion | Usually a lead — an enquiry, a call, a form | Usually a completed transaction with a value attached |
| Quality control | A sales team assesses lead quality later, off-platform | Product, price, checkout and delivery experience decide the outcome immediately |
| The core metric | Cost per acquisition usually carries the decision | ROAS, average order value, contribution margin and product-level profitability all matter at once |
| The offer | Often a single fixed proposition | Hundreds or thousands of products with different margins, demand and stock positions |
That last row is the one that changes the work. A lead-gen account has one offer to get right. A store has a catalogue in which the easiest product to sell is frequently the one you make least money on — and an automated campaign will find it, favour it, and report the result as success.
Six stages, of which the advertising account is one. Google's own documentation makes the dependency explicit: Shopping and Performance Max campaigns use product attributes and Merchant Center data to match products with relevant searches, and Google recommends accurate product data and reliable conversion measurement as prerequisites rather than refinements.
Most underperforming ecommerce accounts I look at are not badly configured at stage four. They are being run correctly on top of an unresolved problem at stage one, two, three or five — and no amount of campaign work reaches those.
Five checks, in this order, before I change a bid or a budget. Each one can invalidate the analysis that would follow it, which is why the order is not negotiable.
01
Selling price, gross margin, habitual discounting, return rate, shipping, and payment or marketplace fees. A high-revenue product is not automatically a high-value one: a ₹4,000 item at 18% margin with one return in eight contributes less than a ₹1,200 item at 55% that nobody sends back. Until I know which is which, any budget recommendation is a guess.
02
Titles, product type, Google product category, GTIN, price, availability, images and any outstanding product issues. Google recommends rich, accurate product data and active issue monitoring — partly because disapproved or silently omitted items cannot be advertised at all, which reads identically to poor performance in a campaign report.
03
The purchase event, transaction value, transaction ID for deduplication, the Google Ads conversion action, GA4, Tag Manager, and what the store itself recorded. Smart Bidding spends against the signals it receives, so a missing or wrong value teaches it the wrong lesson. Correct tracking does not guarantee performance; incorrect tracking prevents it.
04
Google provides product-level reporting across Merchant Center-linked campaign types, so this needs no custom work. What I want is the split: which products genuinely sell, which consume spend without returning, which are seasonal, and which are out of stock but still running.
05
Spend, conversion value, ROAS, CPA, average order value and budget distribution. Last, deliberately: these are output metrics, and reading them before the four checks above tells you what happened without telling you why. That is how accounts get adjusted monthly for a year with nothing changing.
Anyone can read a Google Ads dashboard. The work that changes results is reconciling it against your product costs and order records — that is where the disagreements are, and the disagreements are the findings.
Four campaign types with four different jobs. There is no universal mix — the right combination depends on conversion volume, catalogue size, brand awareness and how much of your demand already exists in search. Each has a dedicated page if you want the implementation detail.
Matched using your product data rather than a keyword list you write, which makes the feed the primary lever. Shopping puts the product, price and image in front of someone already searching for something close to it — usually the highest-intent inventory a store has, and the first place I look when a catalogue is under-monetised.
Uses Merchant Center product data plus your assets and audience signals to serve across Google's eligible inventory automatically. It works well and removes most manual control, which is why the inputs matter more than the settings — product data, conversion values and asset quality are what you still decide.
Captures the intent product feeds express poorly — category and comparison queries, brand searches, and the problem-led phrasing people use before they know what product they want. It is also where negative keyword discipline protects the rest of the account.
Re-engaging people who viewed products or reached checkout without buying. Efficient per conversion and easy to over-credit — it largely harvests demand the rest of the account created, so measure it separately rather than letting it flatter blended performance.
The feed is usually treated as an operations task someone set up once. For a store running product campaigns it is closer to ad copy — the material Google uses to decide what your product is and which searches it belongs in.
What actually needs to be right:
Shopping and Merchant Center implementation in depth: Google Shopping Ads and Merchant Center.
Anatomy of a product record
ROAS measures revenue against ad spend. It knows nothing about what your goods cost, what you discounted, what shipping consumed or what came back as a return. A campaign can hit a strong ROAS target and still leave very little contribution behind.
Illustrative example only — not a client result
Figures chosen to illustrate the arithmetic. Your own margin structure is the only one that determines your targets.
The target should come from break-even ROAS: one divided by gross margin. At the 35% margin above, break-even sits near 2.9×. So 5.0× is genuinely profitable — and a 2.5× campaign in the same store loses money on every order while still looking like a positive return.
This is also why benchmark ROAS figures are close to useless. A store at 70% margin breaks even below 1.5×; one at 25% needs 4×. Two businesses reporting identical ROAS can be in opposite commercial positions, which is why I would rather know your margins than your current ROAS.
One caveat worth stating plainly: Google Ads can only optimise towards values you actually send it. Bidding to profit rather than revenue requires passing margin-adjusted conversion values, and that is a measurement project before it is a bidding one.
Campaign-level reporting treats a catalogue as one product. Look at economics and conversion performance per product and four situations appear, each calling for a different decision rather than a different bid.
High margin · converts well
The straightforward case. Push budget here while demand and stock hold, and check that scaling has not quietly moved the acquisition cost past what the margin supports.
High margin · converts poorly
The most valuable quadrant, because the economics already work. Look at traffic quality, price against competitors, the product page, and friction in the path — the fix is usually not in the account.
Low margin · converts well
These products flatter every report they appear in. Automated bidding will favour them. The question is whether the revenue they generate leaves enough behind to be worth the budget it absorbed.
Low margin · converts poorly
Rarely worth continued paid acquisition in its current form. Sometimes the answer is exclusion; sometimes it is repricing, bundling, or leaving the product to organic and email.
These are diagnostic starting points, not automated rules. A low-margin product can be worth advertising if it reliably brings customers who return — but that has to be demonstrated from your order history, not assumed because it is convenient.
What happens after an account is taken on, in order. The first two steps repeat monthly in miniature — feeds and tracking both drift.
01
Account history and structure, Merchant Center, feed, tracking, product economics and the landing pages campaigns actually send people to. The output is a written view of what is limiting performance, ordered by likely impact.
02
Purchase events, transaction values and deduplication validated, then reconciled against your own order records for a fixed period. Nothing automated gets trusted until the numbers agree, because Smart Bidding inherits whatever error is present. Detail: ecommerce conversion measurement.
03
Which products, campaigns and constraints deserve attention first — judged on what is most likely to move contribution, not on what is quickest to change. Usually three or four things, not thirty.
04
Search term and negative keyword review, product performance and budget allocation, feed issues, asset testing where the campaign type uses assets, and observations on landing pages passed back to you or your developer.
05
Monthly reporting against business economics — spend, conversion value, ROAS, CPA, average order value, product mix and contribution — rather than a screenshot of one platform metric that happened to improve.
06
More investment where demand, conversion performance and margin all support it. Efficiency normally declines as spend rises, so the useful question is where the return crosses break-even, not whether ROAS held.
Not sure whether the account is the problem? That is what the audit answers — and if the constraint turns out to be elsewhere, I would rather tell you than sell you a retainer.
Request an auditCampaign strategy does not change by platform. What changes is how the catalogue reaches Merchant Center and how reliably purchases are measured — which decides whether the strategy can be executed at all.
The Google and YouTube channel connects the catalogue to Merchant Center and handles much of the conversion setup, which makes it quick to establish and easy to leave partially configured. The usual things to verify are whether variants and availability came across as expected, whether purchase values include or exclude shipping and tax, and whether a second tag from a theme or app is duplicating conversions.
More configuration and more control. Feed generation and conversion tracking usually run through plugins, which is fine until several of them are doing overlapping work — the recurring problems are inconsistent product data between feed and site, and duplicated or missing purchase events on custom checkouts.
Nothing is provided, so nothing is assumed. Feed generation, Tag Manager, GA4 and purchase tracking all have to be built — which means they can be built correctly, and against your margin data. I work in Laravel myself, so recommendations arrive in a form a developer can act on.
The same product data serves more than one channel, worth knowing before you pay to build it twice. It also underpins Meta Ads for ecommerce, and the product and category pages it points at are the same ones ecommerce SEO works on.
Spending more money does not fix every ecommerce problem, and paid traffic tends to expose weaknesses rather than compensate for them. In these situations I would recommend fixing something else first — sometimes instead of taking the account on.
If any of these describes your situation, the useful next step is diagnosis rather than a campaign build. My digital marketing diagnostics cover the common cases, including low ROAS and conversion tracking problems.
This is about whether paid acquisition is economically sensible for you right now, not about excluding anyone. Most stores get there; the question is whether they are there yet.
Conditions that make it work
Clear product-market demand, a functional store that converts, product economics with room for an acquisition cost, fulfilment you can rely on, an accurate catalogue, purchases that are measurable, and enough budget to produce data the bidding system can learn from.
The businesses this typically describes: D2C brands and online retailers on Shopify or WooCommerce, in fashion, beauty and personal care, jewellery, home and lifestyle and consumer products — plus manufacturers selling direct, who often have the best margins in the room and the least developed advertising.
Signals it is too early
No validated demand for the product. Margins too thin to carry any realistic acquisition cost. A broken or high-friction checkout. Inventory that cannot be relied on. Tracking nobody trusts.
And one that is a mismatch rather than a stage: an expectation of guaranteed ROAS. I cannot promise a return, because the return depends on demand, competition, your pricing, your store and your margins as much as on the account. Anyone promising a number is either guessing or planning to explain later.
Deeper implementation detail on the parts this page only summarises.
Shopping campaign structure, Merchant Center setup and product feed detail.
How PMax is set up, fed and constrained specifically for product businesses.
Asset groups, audience signals, reporting limits and ongoing PMax optimisation.
Keyword strategy, match types, negatives and search campaign implementation.
Audience lists, dynamic remarketing and re-engaging shoppers who did not buy.
Purchase tracking, transaction values, GA4 and Tag Manager for stores.
I'll review the account, the product data, the conversion tracking and your business economics before recommending what should change. If Google Ads is not the main constraint, I'll tell you that — it is a more useful answer than a campaign rebuild that cannot work.