E-commerce sector hub
E-commerce marketing where the only number that counts is what is left.
Revenue is the easiest number to grow in e-commerce and the least informative. Spend enough on acquisition and revenue always rises. Whether anything is left after the cost of goods, the shipping, the returns and the media is a different question, and it is the only one worth asking.
How do e-commerce businesses grow profitably?
Profitable e-commerce growth comes from four things working together: acquisition that is measured against contribution rather than revenue, product data clean enough for search and shopping surfaces to use, a site fast enough not to lose people on a phone, and retention that makes a second order cheaper than a first.
The most common failure is optimising toward return on ad spend while ignoring the cost of goods, shipping, payment fees and returns. A campaign with a healthy return on ad spend can be losing money on every order.
The second most common is neglecting the mechanics: product structured data, accurate availability and pricing, page speed and a checkout that works one handed on a mid range phone. None of it is exciting and all of it is where the revenue leaks.
Return on ad spend is not profit, and treating it as profit ruins businesses.
A store can run a four to one return on ad spend and lose money on every order. Whether it does depends on the cost of goods, shipping, payment processing, returns and discounting, none of which appear in an advertising platform's reporting.
Contribution margin after advertising is the number that tells you whether growth is worth having. Revenue minus cost of goods, minus fulfilment and shipping, minus payment fees, minus returns, minus media. Positive and growing is a business. Negative and growing is a slow failure with an impressive chart.
Calculating it properly takes an afternoon and changes almost every decision that follows. Products that looked like winners turn out to be margin sinks once returns are counted. Campaigns that looked marginal turn out to be the profitable ones because they sell higher margin items.
It also changes the conversation about discounting. A twenty percent discount on a product with a forty percent gross margin removes half the contribution, which is a very different statement from taking twenty percent off the price.
We set this up first, because every other recommendation depends on it. Results vary by market, budget, competition and other factors. Nothing here is a guarantee of a ranking, a lead volume or a revenue outcome.
Speed is a revenue setting, not a technical preference.
E-commerce sites get slower with every app, tag and carousel added, and nobody notices because the team browses on fast connections.
Google publishes explicit thresholds for Core Web Vitals and the discipline that matters is measuring at the 75th percentile of real page loads, segmented by mobile and desktop, rather than on a developer's laptop.
The usual culprits are predictable: too many third party scripts, unoptimised hero imagery, apps that inject render blocking code, and a theme that was fast when it was installed and has had four years of additions since.
It compounds with mobile share. Most e-commerce traffic now arrives on a phone, frequently on a connection considerably worse than the one the site was built on, and a checkout that is merely slow on desktop can be unusable there.
What good actually means.
These are published specifications rather than research findings, which makes them unusually safe to plan against.
SourceGoogle, web.dev, Core Web Vitals thresholds
A store meeting all three at the 75th percentile is in a small minority, and getting there costs engineering time rather than media budget.
A large sample study, from 2020.
Commissioned by Google and run with Deloitte and 55 across 37 brand sites and more than 30 million user sessions, monitoring mobile load times over 30 days. The year matters and is stated here.
SourceGoogle with Deloitte and 55, Milliseconds Make Millions, 2020, 37 sites
These are correlations from a commissioned study rather than a promise about your store. They are worth knowing because the effect size justifies the engineering.
Your feed is a ranking factor, a shopping surface and increasingly an AI input.
Product structured data is the least glamorous work in e-commerce and it determines whether a product can appear in shopping surfaces at all. Price, availability, condition, identifiers, shipping and return information are all things Google can display, and a store that omits them competes with one hand behind its back.
The requirements are published and specific. Getting them right is a one off engineering job plus ongoing accuracy, and the accuracy half is where most stores fail: prices that lag, availability that is wrong, and variants that do not resolve correctly.
The same data increasingly feeds answer engines. When an assistant is asked to compare products, it draws on whatever structured, accurate information it can find, and a store whose product data is incomplete is simply absent from the comparison.
Shopping ads sit on the same foundation. Prices and availability in the feed have to match the landing page, and mismatches between the two are the single most common reason a shopping ad is disapproved.
Find the business you actually run.
Each page covers how customers in that model buy, the channels that earn their place, the rules that apply and what honest measurement looks like.
Retail Marketing
Physical shops with an online presence, where local search and stock accuracy decide footfall.
Retail MarketingOnline Stores Marketing
The fundamentals: speed, product data, and a checkout that does not lose people.
Online Stores MarketingShopify Marketing
Platform specific structure, apps that slow a store down, and theme decisions that cost conversions.
Shopify MarketingAmazon Marketing
A closed ecosystem with its own rules, where your brand site still has a job to do.
Amazon MarketingDTC Brands Marketing
Paid acquisition economics, retention and the attribution problem that broke the model.
DTC Brands MarketingSubscription Marketing
Churn is the whole business, and cancellation rules are moving.
Subscription MarketingMarketplace Marketing
Two sided demand, where growing one side without the other breaks the model.
Marketplace MarketingDigital Products Marketing
No inventory, no shipping, and a completely different trust problem.
Digital Products MarketingWhat each channel is genuinely for.
The honest version, including the ones we would talk a store out of.
| Channel | Best for | Where it wastes money |
|---|---|---|
| Paid search and shopping | Existing demand for specific products | Broad category terms with no purchase intent |
| Paid social | Creating demand for visual, discoverable products | Commodity products with no story |
| Organic search | Category and comparison content, compounding | Chasing informational terms that never convert |
| Email and SMS | Retention, the highest margin revenue in the business | Only when neglected, which is common |
| Marketplaces | Reach and discovery | Margin, and owning the customer relationship |
| Affiliate and influencer | Reach with a disclosed material connection | Paying for sales you would have made anyway |
The row most stores underuse is email and SMS. It is the only channel where the audience is already yours and the marginal cost of reaching them is close to nothing.
What applies across e-commerce.
- The FTC's rule on fake reviews and testimonials took effect on 21 October 2024. It covers buying reviews, writing your own, and suppressing negative ones. Every review we help collect is first party, requested from a real customer, and never gated on the rating they intend to leave.
- Any material connection between you and someone endorsing you has to be disclosed clearly and close to the endorsement. Paying, gifting, discounting or employing someone all count as material connections.
- The FTC's Mail, Internet, or Telephone Order Merchandise Rule requires a seller to have a reasonable basis for any shipping time it advertises, and to ship within 30 days where no time is stated. A dispatch promise in ad copy is a legal representation.
- Automatic renewal and cancellation requirements are moving. The FTC's negative option rulemaking record is public and has been through both litigation and fresh rulemaking, and several states impose their own automatic renewal rules on top. Treat the renewal disclosure as a legal question with a marketing surface, not the other way round.
- Email and phone follow up carry their own rules. CAN-SPAM requires accurate headers, a physical postal address and a working opt out honoured promptly, and calls and texts to consumers sit under the Telephone Consumer Protection Act. Follow up sequences get built to those rules rather than retrofitted to them.
The review rule is the one that catches stores most often, because incentivised reviews, review gating and importing reviews from elsewhere are all common practices in this sector and all sit inside what the rule addresses.
The second order is where the profit is.
First orders in a paid acquisition model are frequently break even or worse. Everything after that is where a store makes money.
Know your repeat rate
What proportion of customers order again, and how long it takes. Most stores can answer neither, which makes every acquisition decision a guess.
Build the post purchase sequence
Delivery expectations, how to use the product, and a reason to come back timed to the actual consumption cycle rather than to a calendar.
Segment properly
A customer who bought once and a customer who has bought five times should not receive the same message. This is the cheapest improvement available in most stores.
See the remaining steps: The second order is where the profit is.Hide the remaining steps: The second order is where the profit is.
Fix the reasons people do not return
Usually delivery, returns friction or a product that did not match its description. None of those are marketing problems and all of them are marketing costs.
Measure cohorts, not totals
Revenue per customer acquired in a given month, tracked over the following year, tells you whether the business is improving. Total revenue does not.
A store that raises repeat rate by a few points can afford to pay considerably more to acquire a customer, which changes what is possible in every paid channel.
Attribution is worse than it used to be, and pretending otherwise is expensive.
Platform reported conversions are estimated, deduplicated inconsistently across networks, and will collectively claim more orders than the store actually received. Adding up the numbers from three ad platforms and comparing the total to real revenue is the fastest way to see it.
The workable approach is to treat platform data as directional and to manage the business on totals that cannot be double counted. Blended acquisition cost, meaning all marketing spend divided by all new customers, is crude and honest. Contribution after media is the profitability check. Cohort revenue tells you whether it is improving.
Post purchase surveys are more useful than they look. Asking a customer where they first heard of you, in one question at checkout, produces a signal no pixel can and frequently contradicts the dashboard in instructive ways.
The vanity metric across this sector is return on ad spend reported by the platform that spent the money. It is the number most stores are managed to and the one most likely to be flattering.
What we actually do for a store.
Each is a separate scope and can be bought on its own.
E-commerce SEO
Category and product pages built to be found and to convert, on clean product data.
Shopify Design and Development
Storefronts built around the buying decision rather than around a theme demo.
Google Ads Management
Search and shopping campaigns measured on contribution, not platform reported revenue.
Meta Ads Management
Demand creation for products people discover rather than search for.
Website Speed Optimization
Core Web Vitals work at the 75th percentile, where the revenue actually is.
Conversion Rate Optimization
Finding where the checkout loses people and fixing that before buying more traffic.
What actually changes, and what does not.
Everything in this table is something you can look at. None of it is a projection, and none of it is a number we would have to promise you in advance. The work on a store's category and product pages either produces these or it has not been done.
Open the full comparison: What actually changes, and what does not.Hide the full comparison: What actually changes, and what does not.
| What a shopper meets | Before | After |
|---|---|---|
| Category pages | A grid, a title tag and nothing else | A page that explains how to choose, sitting above the grid |
| Product copy | The manufacturer's description, word for word, same as every reseller | Copy nobody else is running, built from the questions your support inbox answers most |
| Product data | Whatever the import brought in | Complete structured data, so a listing appears correctly wherever it is quoted |
| Speed | Measured on the theme demo | Measured on a mid range phone, with your real images and your real apps installed |
| Searches for your own brand | Marketplaces and resellers above you | Your own pages, carrying the answers a buyer is actually checking |
| Reviews | Held on a platform, invisible to search | On the product page, marked up, and collected by you |
| Reporting | Return on ad spend, treated as profit | Contribution after cost of goods, shipping and fees, which is the number you can bank |
This does not change your margin, your shipping cost or what a marketplace charges you, and it will not outrank a marketplace on a generic category term. It changes what happens on the pages you actually own, which is the only surface where the work compounds.
Your website works alongside the store platform you already run.
We do not replace your store platform, we do not migrate it, and we do not ask you to change it. The site links to it from the places a customer is ready to act, and embeds the vendor's own widget where the vendor publishes one. Being plain about the mechanism: that is all it is. Nothing is synced, nothing we build reads or writes your records, and the names below are systems our clients run, not partners of ours. In e-commerce the stack is the business, so this section is less about linking out and more about not fighting the platform: the store stays where it is and the work happens around it.
Store platforms
The platform owns the cart, the checkout and the customer account, and it should keep owning all three. A custom checkout is a liability and the platform's own is not, so we build around it rather than over it.
- Shopify and Shopify Plus
- WooCommerce
- BigCommerce
- Adobe Commerce and Magento Open Source
- Ecwid by Lightspeed
Subscriptions and repeat purchase
The subscription tool owns the hardest page you have, which is the one where somebody manages or cancels. We do not replace it. We do argue for making it easy to reach, because a cancellation somebody cannot find becomes a chargeback instead.
- Recharge
- Bold Subscriptions
Email, SMS and reviews
These already own the messages and the review requests. What the site controls is what they have to work with: whether a product page earns the signup, and whether reviews are shown on the page and marked up rather than left sitting in a dashboard.
- Klaviyo
- Attentive
- Mailchimp
- Yotpo
What we actually change
Not the system: the path to it. Where the button sits on the page, whether it says what happens when you press it, how many taps it takes from a phone, whether it appears again at the point a customer has finished reading and decided, and whether somebody who is not ready yet has a second way to reach you. That path is ours, it is measurable, and in most store audits it is the part doing the damage.
Where the handoff is only a link
Some vendors publish an embeddable widget and some publish nothing at all. Where there is nothing to embed and no deep link worth pointing at, the site sends the visitor to your booking page or your login and stops there. That is a perfectly good outcome and we would rather say so than describe a seam we cannot remove. What we can do is make the destination unsurprising, so nobody arrives wondering whether they are still dealing with you.
Get an e-commerce growth review.
We start with contribution after media, then look at your product data, your site speed and what your repeat rate is actually doing.
One caveat on all of that. These are descriptions of rules as they are published today, not legal advice about your situation. They differ by state and they change, sometimes quietly, so check the current wording with your own counsel or compliance officer before you rely on any of it. Where a rule touches your marketing we write to the stricter reading and send it to you for sign off before anything publishes.
Straight answers.
What is a good return on ad spend?
The question is unanswerable without your margins. A four to one return is excellent for a high margin product and loss making for a low margin one once shipping and returns are counted.
Manage to contribution after media instead. It takes an afternoon to calculate properly and it changes almost every decision that follows.
Why do our ad platforms report more sales than we received?
Because each platform claims conversions it believes it influenced, using different attribution windows and no shared deduplication. Added together they will always overstate.
Treat platform data as directional and manage the business on blended acquisition cost and contribution, which cannot be double counted.
How important is site speed really?
Enough that it is worth engineering time rather than media budget. Google publishes explicit Core Web Vitals thresholds and the measurement that matters is the 75th percentile of real loads on mobile.
Most stores get slower steadily as apps, tags and imagery accumulate, and nobody notices because the team browses on fast connections.
Do we need product structured data?
If you want to appear in shopping surfaces, yes. Price, availability, identifiers, shipping and return information are all things Google can display, and omitting them removes you from the comparison.
The same data increasingly feeds answer engines, which means an incomplete feed now costs visibility in two places rather than one.
Can we offer a discount for a review?
Not safely. The FTC's rule on fake reviews and testimonials covers buying reviews, writing your own and suppressing negative ones, and incentives conditioned on a positive rating sit squarely inside that.
Asking every customer for an honest review, with no condition attached to the rating, is both compliant and more useful, because a perfect rating profile is itself a credibility problem.
Should we sell on marketplaces as well as our own store?
It depends on whether the reach is worth the margin and the loss of the customer relationship. Many brands run both deliberately, using marketplaces for discovery and their own store for repeat purchase.
The mistake is letting the marketplace become the whole business, because the rules, the fees and the customer data are all controlled by somebody else.
How long before e-commerce marketing produces results?
Paid channels produce data within days and a reliable read on contribution within a few weeks. Organic search and retention compound over months.
We will not promise a revenue figure. We will show contribution after media every month and be honest about which channel produced it.
Where this comes from.
Primary documentation and published research behind the guidance on this page.
- Google Search Central: product structured data (opens in a new tab)
- Google Search Central: creating helpful, reliable, people-first content (opens in a new tab)
- Google Search Central: ecommerce site structure best practices (opens in a new tab)
- Google web.dev: Core Web Vitals (opens in a new tab)
- FTC: final rule banning fake reviews and testimonials (opens in a new tab)
- FTC: endorsement guides, what people are asking (opens in a new tab)
- FTC: complying with the Mail, Internet, or Telephone Order Merchandise Rule (opens in a new tab)
- FTC: negative option rule, rulemaking record (opens in a new tab)
Talk to the team
A short call, a look at how the business currently shows up, and a straight answer on what we would do first.