How to Choose an Ecommerce Marketing Agency — Contribution Revenue, Not Just ROAS
When you’re vetting an ecommerce marketing agency, the real question isn’t “can they run great ads” — it’s “have they built our store’s data foundation properly, and can they prove results with contribution revenue instead of a single ROAS number?” About 80% of ecommerce ad performance is decided before the media buying even starts, by the quality of the measurement infrastructure: pixels, Conversion API, and product catalog setup. When data breaks, algorithms learn from the wrong customers, and you end up with a setup where ROAS looks great while actual profit shrinks. A good ecommerce agency stands out on five fronts: ① measurement setup capability, ② reporting centered on contribution revenue rather than ROAS alone, ③ a creative operations system, ④ a repeat-purchase and CRM mindset, and ⑤ adherence to platform best practices. This guide walks through those five criteria along with a checklist of vetting questions you can ask in any agency meeting.
Why Ecommerce Is a Different Game — Data Structure Decides 80% of It
Even though it’s called “marketing agency” just like everywhere else, ecommerce plays by different rules. Before we get into the evaluation criteria, let’s cover why you shouldn’t judge a store’s marketing partner by generic ad-agency standards. The short version: ecommerce results are decided far more by the quality of your measurement infrastructure than by creative or bidding tactics.
First, algorithms learn by eating data. Meta and Google’s automated bidding and machine learning need to know exactly “who bought” in order to find similar buyers. When that signal breaks, the system optimizes toward the wrong people. Third-party cookie restrictions and ad blockers mean a client-side pixel alone now misses a meaningful share of conversions, which is why sending conversions directly from the server — the Conversion API (server-side tracking) — has become standard. In other words, “setup” in ecommerce isn’t a side task; it’s the foundation results are built on.
Second, there’s a long list of events to measure, and they need to be precise. A shopper’s purchase journey runs from product view → add to cart → checkout start → purchase. Google’s GA4 ecommerce guide recommends tracking that journey with events like view_item, add_to_cart, begin_checkout, and purchase, each carrying accurate product-level parameters such as item_id, price, and currency (source: Google, GA4 ecommerce measurement guide). If this structure is missing pieces, or if currency and amounts are off, you can’t tell where customers are dropping off — or which ads actually drove revenue.
Third, the product catalog (feed) is half the ad. Meta’s Advantage+ catalog (dynamic) ads and Google Shopping both auto-generate creative from your product feed. If the feed’s images, prices, stock status, and IDs are inaccurate, the ad itself breaks — budget leaks toward out-of-stock items, or the wrong price gets shown. In ecommerce advertising, “making good creative” matters just as much as “keeping product data accurate and flowing in real time.”
In short, judge an ecommerce agency not by “how flashy can they make the ads” but by “can they design a setup where data never breaks, and use that data to prove real revenue.” That lens lines up exactly with the data-science and growth-hacking methodology Growth emphasizes — because once measurement breaks down, every optimization becomes a guess. Now let’s look at the five criteria that verify that capability in concrete terms.
5 Ecommerce-Specific Evaluation Criteria
The five criteria below go beyond the generic “is the portfolio good, is the price fair” checklist — these are what actually separate ecommerce wins from losses. For each one, we’ve laid out vetting questions you can ask directly in a meeting, in table form. If an agency dodges the answer or simply says “we’ll get you great ROAS,” that’s a sign their capability in that area is weak.
Criterion 1. Measurement Setup Capability — Do They Build the Pixel, Conversion API, and Catalog Themselves?
This is where evaluating an ecommerce agency should start, because as shown above, data structure is the foundation of ad performance. A good agency audits tracking setup before launching any campaign. They’ll check that the pixel and server-side Conversion API work correctly without duplication (including event deduplication), that GA4 ecommerce events are placed across every stage of the purchase journey, and that catalog IDs, prices, and stock sync in real time. On the other hand, an agency that says “setup’s probably already fine, let’s just start running ads” risks pouring your budget onto a broken foundation. For more on why tracking setup is a precondition for results, see our complete guide to performance marketing.
| Vetting question | Signs of a good answer |
|---|---|
| How do you design the pixel and server-side Conversion API together? | Can walk through a concrete process for checking event deduplication and match quality |
| How far into the journey do you place GA4 ecommerce events? | Covers view, cart, checkout start, purchase — down to product-level parameters |
| How do you manage the product catalog (feed)? | Has a system for real-time price/stock/ID sync and error checking |
| How do you verify tracking accuracy before launch? | Has a process — test purchases, event debugging — for confirming data integrity |
Criterion 2. Contribution-Revenue-Centered Reporting — ROAS Isn’t a Single Metric You Can Trust Alone
This is the most important criterion in evaluating an ecommerce agency. Many agencies lead with “we hit 500% ROAS” as proof of results. But ROAS alone can’t tell you whether the business actually made money. First, ROAS is just revenue divided by ad spend — it ignores cost of goods and margin, so a high ROAS can still mean a loss. Second, most ad platforms attribute conversions in ways that favor their own channel, so the sum of ROAS each platform reports tends to overstate actual revenue. Third, ROAS tends to take credit for purchases that would have happened anyway (revenue, not incrementality). We break down the mechanics of this trap in The ROAS Trap — How a High ROAS Can Still Mean a Loss.
That’s why good agencies treat ROAS as a supporting metric and center reporting on contribution revenue and profit. Google itself points out that giving all the credit to the last click “ignores the other ad interactions the customer went through along the way,” and explains that data-driven attribution instead “calculates the real contribution of each interaction along the conversion path and distributes credit accordingly” (source: Google Ads, About attribution models). In other words, “which ad drove the sale” isn’t a single-metric question — it’s a question about the contribution of the entire path. This is also why Growth works around the idea of “one visitor who becomes revenue” rather than raw traffic volume. Here’s how the key metrics break down:
| Metric | What it shows | Limitations / caveats |
|---|---|---|
| ROAS (revenue over ad spend) | Revenue efficiency of an ad channel | Doesn’t reflect margin, incrementality, or duplicate attribution — never judge on this alone |
| Contribution revenue (attribution) | How much each channel actually contributed to revenue | Needs to be viewed across the whole path; platform-reported numbers can be inflated |
| Profit-based metrics (margin-adjusted) | Whether the ads generated real profit | Requires connecting cost, returns, and shipping data |
| Incrementality | Revenue that wouldn’t have happened without the ads | Requires experimental (holdout) design |
| Vetting question | Signs of a good answer |
|---|---|
| Do you report results with ROAS alone, or do you also show contribution revenue and profit? | Presents ROAS as a supporting metric alongside contribution revenue and margin |
| How do you handle the gap between the platforms’ reported ROAS totals and actual revenue? | Recognizes duplicate attribution and corrects using an integrated view (MMM, experiments, total revenue reconciliation) |
| How do you separate out purchases that would have happened without the ads (incrementality)? | Proposes experimental design such as holdout or incrementality testing |
| Can cost and margin data be reflected in reporting? | Takes a profit-based approach (e.g., margin-adjusted metrics) to judging performance |
Criterion 3. Creative Operations System — Do They Run Creative as a System?
In ecommerce advertising, creative isn’t something you make once and you’re done — it’s a consumable that needs constant testing and rotation. Run the same creative too long and fatigue sets in, dragging performance down. So what you should verify isn’t “can they make pretty creative” but “do they run creative through a hypothesis–production–test–learning system.” A good agency validates which messages, hooks, and formats work with data, and reallocates budget by picking winners through A/B testing — with the discipline to look for statistically meaningful differences, not one-off results. You can find methods for continuously improving ad creative in How to Keep Improving Ad Performance with A/B Testing.
| Vetting question | Signs of a good answer |
|---|---|
| How many creatives do you test per month, and against what hypotheses? | Has a pipeline that supplies and rotates creative regularly, driven by hypotheses |
| What determines a winning creative? | Judges by statistical significance and conversion metrics — not gut feel or impressions |
| How do you detect creative fatigue (declining performance)? | Monitors frequency, CTR, and conversion trends to time rotations |
| How do learnings from a winning creative feed into the next one? | Structures winning elements into a loop that accumulates into the next hypothesis |
Criterion 4. A Repeat-Purchase and CRM Mindset — Do They Look Past the First Sale?
Ecommerce profitability comes from repeat purchases, not the first one. Customer acquisition cost (CAC) keeps rising, so the ability to get a one-time buyer to buy again determines whether the business is sustainable. Yet many agencies only watch new conversions and ROAS, ignoring customer lifetime value (LTV) or repeat-purchase flow. A good agency designs retention alongside acquisition — post-purchase email/message flows, retargeting by segment, and budget allocation based on LTV rather than first-purchase-only. This connects directly to the Customer Decision Journey (CDJ) lens Growth emphasizes: a purchase isn’t the end of the journey, it’s the start of the next one. You can read more about this lens in our guide to the Customer Decision Journey (CDJ).
| Vetting question | Signs of a good answer |
|---|---|
| Do you measure success by first purchase (new customers) alone, or also by LTV and repeat rate? | Tracks LTV and repeat-purchase rate as core metrics alongside acquisition |
| Do you design flows to bring customers back after their first purchase? | Runs CRM (email/message) and retargeting segmented by customer type |
| How do you split budget between new and existing customers? | Allocates by channel/segment by weighing CAC against LTV |
| Do you segment customer data for use? | Differentiates messaging using RFM or purchase-history-based segments |
Criterion 5. Platform Best Practices — Do They Handle Automation Correctly?
Meta and Google’s ad products are increasingly shifting toward automation (Advantage+, Performance Max, and the like). Automation is powerful, but it only works properly when it’s fed good data and configured correctly. So what you should verify isn’t “do they manually micromanage everything” but “do they use automation the way the platform recommends, backed by proper signals and structure.” Best practices include keeping the product catalog at sufficient scale, optimizing toward lower-funnel events (purchase), and avoiding frequent structural changes that disrupt the learning phase. You can see real-world use of Meta’s Advantage+ catalog ads in our Meta ASC (Advantage+ Shopping) campaign guide.
| Vetting question | Signs of a good answer |
|---|---|
| How do you use automation like Advantage+ or Performance Max? | Explains a principle of using automation only once signal quality and structure are in place |
| What do you set as the optimization goal? | Optimizes toward lower-funnel conversions like purchase — not views or clicks |
| What do you avoid to protect the learning phase? | Follows discipline against frequent structural changes and sudden budget swings |
| How do you keep up with platform policy and updates? | Has a process for updating operations based on official guidance |
Red Flags — Agencies You Should Avoid
If you spot any of the following signs, revisit the deal before signing. These are risk patterns that show up especially often in ecommerce.
Red flag 1 — They promise “guaranteed X% ROAS.” ROAS moves depending on margin, incrementality, and duplicate attribution, so promising a specific number can be a sign they either misunderstand what measurement actually means, or they’re using tricks to make the number look good (like pouring budget into branded keywords and retargeting to take credit for purchases that were already going to happen). The same risk shows up in SEO with guarantee-based promises. Google explicitly states that “no one can guarantee a #1 ranking,” and warns against vendors who guarantee results or keep their methods secret (source: Google, Do I need an SEO?). The same logic applies to “guaranteed ROAS” — it’s not a healthy sign. We cover this mechanism in more depth in Why Agencies Push Guarantee-Based Contracts.
Red flag 2 — All you get is a black-box report. This is when, at month-end, all you receive is a single PDF with a few ROAS numbers — no visibility into which campaign, creative, or segment produced them, and no direct access to the data yourself. Without transparent measurement, you can’t verify whether results are real, and all the learning disappears the moment you end the engagement. A good agency gives you direct access to the ad accounts and analytics tools, and explains the reasoning behind the numbers — what hypothesis they tested, what they changed, and what improved.
Red flag 3 — They want to skip setup and go straight to spending more. An agency that says “just increase the budget and revenue will follow,” without ever auditing the data structure, is pouring cost onto a broken foundation. If measurement is inaccurate, the algorithm learns from the wrong customers, and the waste scales right along with your spend. Increasing ad spend should only ever come after you’ve confirmed measurement is accurate.
What Does a Good Partnership Look Like? — A Data and Account Ownership Checklist
Even an agency that passes every evaluation criterion above can still leave you empty-handed at the end of the contract if the collaboration structure is weak. Data and account ownership is where ecommerce agency relationships most often break down. If your ad accounts, pixel, catalog, and analytics data are all locked inside the agency’s own ownership, you lose everything you’ve built the moment the engagement ends. Be sure to confirm the following before signing.
| Item to check | Why it matters | Healthy state |
|---|---|---|
| Ad account ownership | If the account is in the agency’s name, recovering data and history at the end is hard | Account is in your (the advertiser’s) name, with the agency connected as a partner |
| Pixel, Conversion API, catalog | If measurement assets are locked to the agency, learning doesn’t transfer | Owned by your business, with the agency holding only operating access |
| Analytics access | Without access, you can’t verify results or bring anything in-house | You always have direct access to GA4 and dashboards |
| Reporting cadence and format | A monthly PDF can’t accumulate improvements | Real-time dashboards plus regular reviews to share hypotheses and results |
| Transition process | Without a handover, assets evaporate at the end | Data, account, and creative handover terms are written into the contract |
The core principle is simple: measurement assets and ad accounts should belong to the advertiser, with the agency holding only operating rights — that’s the healthy structure. This way, your data and learning survive even if you switch agencies or bring things in-house. Ask “whose name are the account and pixel under?” early in any meeting, and you’ll quickly learn whether that agency is ready to work as a long-term partner.
How Growth Is Different
Growth is an ecommerce marketing partner that works around one principle: one visitor who becomes revenue matters more than traffic volume. Before increasing spend, we audit the pixel, Conversion API, and catalog setup so data never breaks. We prove results with contribution revenue and profit, not a single ROAS number, and we run creative through a hypothesis–test–learning system. Rather than stopping at acquisition, we design the entire customer journey through repeat purchase and LTV — that’s what it means to work on a data-science and growth-hacking foundation. The five criteria above are, in effect, a description of how Growth operates.
If you’re looking for an ecommerce or online store marketing partner, take a look at how we approach it in Growth’s performance marketing service, and reach out for a consultation to talk through whether it fits your store.
You can see the full picture of this topic in “Ecommerce Marketing Strategy Guide — Full Funnel, Data, and Platform.”
Frequently Asked Questions (FAQ)
What should I check first when choosing an ecommerce marketing agency?
Measurement setup capability. The foundation of ecommerce ad performance is data structure. In your first meeting, ask how they design the pixel and server-side Conversion API together, and how they manage GA4 ecommerce events and the product catalog. Google’s own GA4 ecommerce guide recommends accurately tracking view, cart, checkout, and purchase along with product-level parameters (Google GA4 ecommerce guide). Avoid any agency that wants to skip setup and jump straight into running ads.
Is it okay to evaluate an agency on ROAS alone?
No. ROAS is just revenue over ad spend — it doesn’t account for cost or margin, so a high ROAS can still mean a loss. Ad platforms also tend to attribute conversions in ways that favor themselves, so the sum of reported ROAS tends to overstate actual revenue. Google itself notes that giving all credit to the last click “ignores the other interactions the customer went through,” and that data-driven attribution instead distributes credit based on each interaction’s real contribution along the path (Google Ads attribution models). Choose an agency that treats ROAS as a supporting metric and reports contribution revenue and profit alongside it.
Can I trust a contract that promises “guaranteed X% ROAS”?
Be cautious. ROAS shifts with margin, incrementality, and attribution method, so promising a specific number can be a sign the agency either misunderstands measurement or is using tricks to make the number look better (like taking credit for purchases that would have happened anyway). It’s the same logic behind Google’s warning that “no one can guarantee a #1 ranking” and that you should be wary of vendors who guarantee results or keep their methods secret (Google, Do I need an SEO?). An agency that clearly explains how it defines, verifies, and transparently reports results is the healthier choice over one promising a guaranteed number.
Who should own the ad account and data when working with an agency?
It must be the advertiser — you. If your ad account, pixel, Conversion API, catalog, and analytics data are all locked under the agency’s name, you lose everything you’ve built the moment the engagement ends. The healthy structure is one where you own the measurement assets and accounts, and the agency holds only operating rights. Before signing, confirm in writing the account ownership, data access rights, and transition process at the end of the engagement. That way, your data survives even if you switch agencies or bring it in-house.
