The problem: revenue that grows while profit does not
Online stores have a specific version of the growth trap. Shopping and Performance Max campaigns are told to maximize conversion value at a target ROAS, so they push the products that sell easily: the discounted lines, the low-margin bestsellers, the items existing customers were about to buy anyway. Reported revenue climbs, the ROAS target is met, and the finance team finds that blended margin after ad spend went backwards.
Add faceted navigation that has spawned tens of thousands of crawlable URLs, a product feed with missing GTINs and generic titles, and an email program nobody has connected to the acquisition numbers, and you have the typical situation an e-commerce marketing agency is asked to fix. The fix is rarely one channel. It is getting the catalogue data, the search visibility, the paid campaigns and the retention math to work from the same numbers.
What e-commerce marketing covers at Axoria
We define e-commerce marketing as the coordinated management of everything that puts products in front of buyers and turns them into profitable orders: product feeds and Shopping ads, Performance Max, category and product page SEO, on-site merchandising and conversion, and the email and retention layer that determines what a first order is really worth. Marketplace presence is considered where it changes the acquisition math.
The unifying principle is profit-aware decision making. Every product has a margin, a return rate and a repeat-purchase profile, and those numbers should decide how much you bid for it, whether it belongs in a paid campaign at all, and which categories deserve SEO investment. Most platforms will happily optimize toward revenue; teaching them to optimize toward contribution is where the work lies.
Who this is for
Direct-to-consumer brands and online retailers on Shopify, WooCommerce, Magento, BigCommerce or a custom platform, with a catalogue large enough that feed quality and site structure matter and enough order volume for bidding automation to work. It suits stores already spending on Google and Meta who suspect the reported numbers flatter reality, and retailers whose organic visibility has stalled because of technical and structural issues in the catalogue.
Very small catalogues or single-product stores often need paid social and CRO more than feed and category work, and we will say so. Our e-commerce industry page covers how the strategy differs by store type.
What Axoria does
Product feed management
Feed structure, titles, attributes, GTINs, custom labels for margin and seasonality, supplemental feeds and rules in Merchant Center or a feed tool, so Shopping and PMax have accurate data to match on.
Shopping and Performance Max
Campaign structure split by margin tier, product role or brand, with target ROAS set per group from unit economics, brand traffic exclusions, and asset groups that do not cannibalize each other.
Category and product SEO
Category page targeting, product content, internal linking and faceted navigation control so the store ranks for the commercial terms that matter without wasting crawl budget on filter permutations.
Merchandising and on-site conversion
Collection ordering, product page structure, search and filter behaviour, cart and checkout friction, tested rather than assumed.
Retention integration
Email and SMS flows (welcome, post-purchase, replenishment, win-back) treated as part of LTV, so allowable first-order CAC is grounded in real repeat behaviour.
Profit reporting
Reporting that shows contribution margin after ad spend by campaign and category, new-customer share, and returns-adjusted revenue, reconciled to your store data.
Feeds and Performance Max: control where the platform gives you little
The product feed is the targeting layer for Shopping and Performance Max, so feed quality is campaign quality. A title such as “Blue Jacket” competes with every blue jacket on the internet; “Men’s Waterproof Hiking Jacket, Navy, Recycled Shell” matches the queries buyers use. We rewrite titles and attributes from search query data, fill missing GTINs and product types, and add custom labels carrying margin band, stock depth, price bracket and seasonal relevance so campaigns can be segmented on them.
Performance Max is powerful and opaque. Left alone it spends heavily on brand search and remarketing, reports that as new revenue, and starves genuinely new demand. Our standard setup excludes brand terms, separates campaigns by margin tier or product role so a high-margin line is not outbid by a clearance line, checks the search terms and placement reports that are available, and runs a standard Shopping campaign alongside for terms where manual control is worth it. When a PMax campaign’s reported ROAS looks too good, we test it against a holdout before trusting it with more budget.
Profit-based bidding in practice
If two products both sell at $100 but one has a 60% gross margin and the other 20%, a single target ROAS treats them identically and over-invests in the second. We either pass margin-adjusted conversion values to the platform, or split campaigns by margin tier with different targets. Either way, the bid reflects what the sale is worth to you, not what the customer paid.
Category and product page SEO for large catalogues
For most stores, category pages are the organic revenue engine: they target the commercial head terms (“women’s running shoes”, “ergonomic office chairs”) that product pages cannot. We map categories to keyword clusters, write category content that helps rather than pads, build internal links from guides and related categories, and make sure the page structure (H1, filters, product grid, pagination) can be crawled and understood.
Faceted navigation is the recurring technical problem. A store with 3,000 SKUs and filters for size, colour, brand and price can generate hundreds of thousands of URL combinations, most of them thin duplicates that consume crawl budget and dilute ranking signals. The typical treatment is a combination of canonical tags to the parent category, noindex on low-value combinations, robots rules for parameter patterns, and selectively indexing high-demand facets (such as “brand + category”) as proper landing pages. Product pages get unique descriptions, structured data (Product, Offer, AggregateRating only where reviews genuinely exist) and image optimization. This work is delivered with our technical SEO and on-page SEO specialists.
Retention as part of the acquisition equation
How much you can afford to pay for a first order depends entirely on what happens after it. A skincare brand with a 45% 90-day repeat rate can run first-order acquisition at break-even or below; a furniture retailer with rare repeat purchases must make margin on the first sale. We build the repeat-purchase curve from your order data by cohort and product category, and derive allowable CAC from it rather than from a generic ROAS target.
The email and SMS program is then judged on its effect on that curve. Post-purchase flows, replenishment reminders timed to product usage and win-back sequences are the cheapest revenue most stores have, and improving them raises the ceiling on paid acquisition. We work in Klaviyo or a comparable platform, and the retention numbers appear in the same report as the acquisition numbers.
Marketplaces and seasonal planning
Marketplace considerations
Amazon and other marketplaces can add volume but change the math: fees, reduced customer data and price transparency. We assess marketplace presence on incremental contribution and brand control, and we watch for channel conflict, where marketplace listings undercut your own store in Shopping results. Marketplace advertising itself is scoped separately when it makes sense.
Seasonal and promotional planning
Peak periods are planned months out: inventory-aware budgets, feed and campaign changes staged before demand rises, promotional pricing reflected in the feed (sale price attributes, Merchant Center promotions), and post-peak scaling back before CPCs rise and margins fall. Bidding automation needs lead time to adjust, so promotions are not sprung on it.
Our process
Catalogue and economics review
Margin, return rate and repeat behaviour by category from your order data; feed quality diagnostics in Merchant Center; a crawl of the store to size the faceted navigation and duplicate content problem.
Measurement fixes
Purchase events deduplicated, new-versus-returning flagged, margin data or product-level values available to bidding, and returns reflected in reporting.
Feed and structure rebuild
Titles, attributes and custom labels rewritten; campaigns restructured by margin tier and product role; category targeting and facet handling implemented.
Launch and stabilize
Campaigns run at conservative targets while bidding models learn on clean data, with weekly search term, product and placement review.
Optimize on contribution
Targets adjusted by product group from margin data; SEO content and internal links rolled out by category priority; retention flows tuned against the repeat curve.
Scale through the calendar
Budget steps up in the groups that hold contribution targets, with the seasonal plan governing inventory-aware pushes and pull-backs.
Tools we typically work with
Google Merchant Center, Google Ads and Microsoft Merchant Center for Shopping and PMax; feed management tools such as DataFeedWatch or Channable for larger catalogues; Meta Ads Manager with catalogue sales campaigns; GA4 and Google Tag Manager for measurement, usually with server-side tagging on Shopify or a custom stack; Klaviyo for retention; Screaming Frog, Search Console and Semrush or Ahrefs for the SEO side; Looker Studio for profit reporting. We work inside your accounts and you keep ownership.
How results are measured
The headline metric is contribution margin after marketing cost, split by new and returning customers, with returns-adjusted revenue rather than gross order value. Beneath it: new-customer CAC and payback, MER across all channels, organic revenue by category, feed and Shopping impression share on priority products, and repeat rate by cohort. Platform ROAS is reported as a diagnostic alongside these, never as the outcome.
Reports are monthly with a weekly summary during peak season, and they reconcile platform numbers to your store’s order data so that the revenue we discuss is the revenue that shipped.
What to expect and common challenges
Restructuring Shopping and PMax campaigns resets learning, and performance typically wobbles for two to four weeks before settling; we stage changes to limit this and avoid doing it in peak season. Feed rewrites at scale need either a feed tool or developer support for bulk changes. Faceted navigation fixes on some platforms need theme or app changes, and organic gains from category work compound over months rather than weeks.
We will not run campaigns to a revenue target that ignores margin, report brand-search revenue as if it were new acquisition, or promise a ROAS figure. We will tell you when a product line is unprofitable to advertise, even when the platform says it is performing.