Steps to Improve eCommerce SEO With an Agency: The Joint Operating Model US Merchants Get Wrong
Most eCommerce SEO retainers stall not because the steps are wrong, but because the handoffs between merchant and agency are unowned. Here is the step sequence with a named owner at every stage.

Image via 123RF
In This Article
Handoffs Kill Retainers, Not Tactics
Baseline Before Crawl
Margin Data Sets the Priority
Why 'Hire the Agency and the Steps Happen' Is the Most Expensive Myth in eCommerce SEO
Most US merchants signing an eCommerce SEO retainer believe the sequence of work is what they are buying. It is not. What they are buying is a partner who will execute a sequence that does not run without merchant inputs. The steps themselves - baseline, priority mapping, crawl triage, structured data, category depth, consolidation, links, iteration - are widely documented. What is not documented is the handoff model that makes them actually occur.
The Handoff Model: Every Step Has a Merchant Input and an Agency Output
Each of the eight steps in a serious eCommerce SEO engagement has two halves. The agency side is the deliverable. The merchant side is an input the agency cannot manufacture: margin bands, return rates, inventory depth, dev deploy windows, product launch calendars, GA4 admin access, and the commercial judgement about which categories the business wants to own in twelve months. When the merchant half is missing, the agency still ships something - it just ships the wrong thing, on time, with a clean report.
The most common retainer failure is not a bad tactic. It is a delivered recommendation with no owner on the merchant side to accept it.
The Four Myths That Break the Step Sequence Before It Starts
Myth: Step one is a technical audit. Fact: the crawl is step one of the technical workstream, not step one of the engagement. If you crawl before validating measurement, every later decision is ranked against a revenue number you have not confirmed. Myth: more pages equal more SEO. Fact: on most US Shopify and WooCommerce catalogs the binding constraint is index bloat and internal link equity distribution, not page count. Myth: SEO runs on a separate track from the agency's paid media. Fact: the same query and product-level data should feed both, or you are paying twice for the same research. Myth: AI search changed the steps. Fact: it changed the output format - answer surfaces and entity clarity - but crawlability, entity clarity, and commercial-intent content still gate everything else.
Step 1 - Baseline and Measurement Validation (Not a Technical Audit)
Before a single crawl budget decision is made, the engagement needs an agreed revenue baseline covering the trailing 12 months, split by organic landing entry and by category. This is the number every later step is judged against.
What the Merchant Must Supply: GA4 Admin, GSC Domain Verification, Order Data
- GA4 Admin (not just Viewer) access so event-level configuration can be inspected.
- Google Search Console domain-property verification, plus access to any legacy URL-prefix properties.
- Order data export from Shopify, WooCommerce, or your ERP, covering at least the same 12 months.
- Consent mode configuration details for your CMP. If consent mode is misconfigured, the baseline is understated and every later step is measured against a wrong number.
What the Agency Delivers: Purchase Event Audit, Items Array Check, Revenue Baseline
The audit checks that the GA4 purchase event fires once per transaction, includes a populated items array, and carries the correct currency. A very common WooCommerce gap is a purchase event that fires with a transaction value but no item-level detail, which makes product-level reporting impossible.
gtag('event', 'purchase', {
transaction_id: 'T-10482',
value: 1249.00,
currency: 'USD',
items: [{
item_id: 'SKU-3391',
item_name: 'Trail Runner GTX',
item_category: 'Footwear',
price: 1249.00,
quantity: 1
}]
});
Google's own GA4 ecommerce reference is explicit about the required parameters. An agency that skips this check will spend the next six months optimising the wrong category.
The Handoff Failure: Reporting on Sessions Because Ecommerce Events Were Never Fixed
If step one is skipped or half-completed, the retainer defaults to session reporting. Sessions are easy to produce and easy to present. They are also disconnected from margin, from product-level demand, and from the paid media account - which means the SEO programme becomes unfalsifiable. Nobody can prove it worked, and nobody can prove it didn't.
A validated revenue baseline is the single most valuable artefact an eCommerce SEO engagement produces in month one. Everything after step one is a comparison against it.
Step 2 - Commercial Priority Mapping Using Margin, Returns, and Inventory Depth
Ranking categories by organic sessions is the default because it is the data the agency already has. It is also the reason many retainers produce flat profit: the highest-traffic categories are frequently the lowest-margin and the highest-return.
Why Traffic-Led Prioritisation Produces Flat Profit
A category with 40,000 monthly organic sessions, a 9% gross margin, and a 22% return rate contributes less than a category with 6,000 sessions, a 41% margin, and a 4% return rate. If the agency never sees the margin bands, it cannot know this. The merchant supplies three data columns - gross margin band, return rate band, and inventory depth in weeks - and the agency converts them into a priority tier.
The Three-Tier Priority Grid: Defend, Grow, Deprioritise
| Tier | Typical Profile | Agency Action |
|---|---|---|
| Defend | High margin, high inventory depth, existing top-10 positions | Protect rankings, refresh category copy, fix cannibalisation |
| Grow | High margin, deep inventory, positions 11-30 | Category depth, internal links, structured data, content expansion |
| Deprioritise | Low margin or high return rate regardless of traffic | Maintain, do not invest - redirect effort to Defend and Grow |
This grid is the single most useful agenda item for a monthly working session. It also prevents the most common merchant complaint: that the agency keeps optimising pages that sell low-margin stock.
Step 3 - Crawl, Index, and Architecture Triage
This step is agency-led but merchant-gated, because nothing ships without a deploy window and a developer who accepts the ticket.
Index Bloat on Shopify and WooCommerce: Facets, Pagination, and Tag Pages
On many US Shopify and WooCommerce stores, indexed URL counts run several times higher than the true product-plus-category count. Faceted filter URLs, paginated variants, and tag pages are the usual sources. Google's crawl budget guidance for large sites frames this as a crawl efficiency problem, not a penalty problem. The fix is disciplined parameter handling plus canonical and internal link rules - not blanket noindex, which too often strips pages that do generate revenue.
The Dev Ticket Problem: Why Recommendations Sit in Shared Docs for Months
Recommendations do not fail because they are wrong. They fail because they live in a shared document with no assignee, no acceptance criteria, and no date. The structural fix is to convert every technical recommendation into a Jira, Linear, or GitHub issue with a named merchant-side owner, a definition of done, and a target sprint. If it is not a ticket, it is not a decision - it is a suggestion.
A shared backlog with named owners is worth more than any audit PDF. The audit describes the problem; the backlog closes it.
Step 4 - The Entity and Structured Data Layer
Structured data tells Google which entity a page describes, what it costs, whether it is in stock, and what it is rated. Without it, the crawler is inferring from unstructured markup.
Product, Offer, AggregateRating, BreadcrumbList, Organization
- Product + Offer - price, currency, availability, and seller, kept in sync with the PDP. Stale price markup on sale transitions is the most common markup error we see.
- AggregateRating - only where genuine first-party review data exists.
- BreadcrumbList - matches the visible breadcrumb and reinforces category hierarchy.
- Organization - verifies the manufacturer/brand entity relationship.
Google's structured data introduction makes clear that valid markup is a prerequisite for rich result eligibility, not a lever that forces rich results to appear.
Eligibility Is Not a Guarantee: Framing Rich Results Honestly
Merchants frequently ask for 'guaranteed star ratings in search.' There is no such thing. Correct markup is an eligibility outcome. Whether your competitor appears with a price panel and you do not is influenced by many factors outside the markup layer. An agency that frames structured data as an eligibility input rather than a control is being straight with you.
Core Web Vitals Reality Check: INP Replaced FID in March 2024
Interaction to Next Paint replaced First Input Delay as a Core Web Vital in March 2024, as documented on web.dev's INP reference. Any agency still reporting FID in a monthly deck is working from a stale playbook. Ask directly which metric they report and in which reporting tool. This is the fastest, cheapest vetting question in the entire engagement.
Step 5 - Category and Collection Page Depth (Where Commercial Queries Actually Rank)
Product detail pages rarely rank for head commercial terms. In categories like footwear, furniture, supplements, and electronics accessories, the top of the SERP is almost always occupied by category and collection pages - because the query 'trail running shoes' is not asking about one SKU.
Why PDPs Rarely Win Head Commercial Terms
A PDP is a single-entity document. When a shopper types a category term, Google's systems are looking for a page that represents a range. If the merchant's category pages are empty shells carrying only a filter and a product grid, the agency has nothing to work with and will over-index on PDP-level tactics that cannot move category rankings.
The Category Page Content Stack: Intro Copy, Buying Guidance, Internal Links
- Intro copy (80-150 words) above or below the grid - describes the category, not the store.
- Buying guidance block - how to choose, sizing, materials, compatibility. This is where the merchant's product expertise is the input and the agency's role is structure.
- Curated internal links - from the category to its strongest sub-collections and to the two or three PDPs that deserve internal equity.
- FAQ block on the category page - covering the questions the sales team already hears on chat and phone.
The merchant's product team writes the buying guidance. The agency structures it. Both need to be in the room for Step 5 to produce anything that ranks.
Google's people-first content guidance is the clearest published framing of why this works: pages that demonstrate first-hand expertise outperform pages that describe the category generically. On a category page, the merchant is the only source of that expertise.
Step 6 - Content and Internal Linking to Consolidate Equity
Most US catalogs do not have a content volume problem. They have a distribution problem: relevant pages exist but receive no internal links, while outdated blog posts absorb the equity.
Consolidation Beats Creation on Most Catalogs
The first content action in a mature engagement is usually not to write nine new articles. It is to merge overlapping posts, redirect weaker variants, and re-point internal links from dead content to the pages that need ranking support. This step is merchant-dependent because consolidation involves editorial judgement the merchant owns.
Anchor Discipline and Hub-to-Spoke Mapping
Define three to five anchor variants per target page and specify them in the shared backlog. Random anchors dilute signal. Map each category hub to its supporting content and each supporting piece back to the hub with a controlled anchor. Google's SEO Starter Guide is a reasonable shared vocabulary document for this conversation if the merchant team is small and inexperienced.
Step 7 - Digital PR and Link Acquisition With a Realistic Cadence
Link acquisition is the step most likely to be quietly abandoned, because both sides are waiting for the other to supply a hook.
What a Sustainable Monthly Cadence Looks Like
A realistic US engagement targets a defined number of outreach conversations per month and a defined number of placements per quarter, with the merchant supplying proprietary data: customer survey results, product testing data, category price indices built from their own catalog, or original usage data. Without merchant-supplied data, the agency is reduced to pitching generic commentary, which does not earn links from publications that matter.
Why Link Volume Is the Wrong KPI
Counting referring domains tells you almost nothing. A single placement in a category-relevant publication is worth more than forty directory mentions. Report link acquisition against the Defend/Grow tiers from Step 2 and ignore aggregate volume.
Step 8 - The Iteration Loop Tied Back to the Step 1 Baseline
Steps 1 through 7 are a single pass. The compounding value comes from the loop.
Monthly Working Session, Shared Backlog, Quarterly Re-Prioritisation
Once a month, the merchant and agency work the same backlog together for 60 minutes. Once a quarter, re-map priorities against refreshed margin and inventory data. If a category that was in Grow has become a clearance category, it should move to Deprioritise without a debate.
Quarterly re-prioritisation is the difference between an SEO retainer that compounds and one that slowly drifts away from the merchant's actual commercial situation.
Who Owns What: The Governance Table That Keeps the Steps Compounding
| Step | Merchant Input | Agency Output | Handoff Failure If Unowned |
|---|---|---|---|
| 1. Baseline | GA4 admin, GSC, order data | Purchase event audit, revenue baseline | Reporting on sessions; no falsifiable revenue line |
| 2. Priority mapping | Margin bands, return rates, inventory depth | Defend/Grow/Deprioritise tiers | Optimising high-traffic, low-margin categories |
| 3. Crawl and index | Dev access, deploy windows, ticket owner | Technical recommendations as tickets | Recommendations sit in a shared doc for months |
| 4. Structured data | Product and review data feeds | Product, Offer, AggregateRating, BreadcrumbList, Organization markup | Stale price markup; ratings dropped in search |
| 5. Category depth | Buying guidance, sizing, FAQ content | Content structure, internal link placement | Category pages stay thin; PDP tactics over-applied |
| 6. Consolidation | Editorial judgement on which posts to merge | Redirect map, anchor specification | More content added instead of equity redirected |
| 7. Digital PR | Proprietary data, survey results, testing data | Outreach cadence and placement reporting | Generic pitching; no placements |
| 8. Iteration | Attendance at monthly session, refreshed margin data | Backlog maintenance, quarterly re-map | Drift; retainer continues but priorities are stale |
Named owners, a fixed meeting cadence, and margin-led re-prioritisation are the three mechanics that keep the steps compounding rather than drifting.
Which Merchant Profile Fits Which Engagement Shape?
Retainer pricing varies enormously in the US, and any figure below is an estimate, not a benchmark. We present ZAR ranges with a USD reference because that is how our international client base compares them, and because no two scopes are identical.
| Merchant Profile | Engagement Shape | Indicative Retainer (estimate) |
|---|---|---|
| $1M-$5M store, solo marketing lead | Baseline + priority mapping + category depth + quarterly iteration; dev tickets handled by the merchant's existing agency or freelancer | R35,000-R70,000 / ~$1,900-$3,800 per month |
| $5M-$15M store, 2-4 person team | Full eight-step sequence including structured data, consolidation, and a modest digital PR cadence | R70,000-R150,000 / ~$3,800-$8,200 per month |
| $15M-$30M store, in-house dev capacity | Full sequence plus technical delivery support, server-side data collaboration, and a quarterly strategy workshop | R150,000-R300,000 / ~$8,200-$16,400 per month |
If your team cannot supply the merchant-side inputs described in the governance table, the honest move is to start with Steps 1, 2, and 5 only. Adding Steps 3, 6, and 7 before those inputs exist produces the same outcome as buying a gym membership: a monthly charge and no adaptation. If you want to see how the framework applies to a specific catalog, explore the framework or request a growth audit.
Sources
- Google Search Central - SEO Starter Guide: The Basics
- Google Search Central - Large site owner's guide to managing your crawl budget
- Google Search Central - Intro to how structured data markup works
- web.dev - Interaction to Next Paint (INP)
- Google Analytics Help - Measure ecommerce (GA4 ecommerce event reference)
- Google Search Central - Creating helpful, reliable, people-first content


