Ecommerce & DTC marketing
Grow online revenue by moving the term that is actually stuck
Online revenue is traffic multiplied by conversion rate, average order value and repeat rate. Most ecommerce marketing cannot say which of those four it is working on, which is why the spend goes up and the profit does not.
Measurement
What we report on, and what we ignore
Sessions are not on this list. These are the numbers that tell you whether the marketing is producing customers.
- Contribution margin after cost of goods, shipping, fees and returns — the number every other metric answers to
- Revenue decomposed into traffic, conversion rate, average order value and repeat rate, reported as four lines
- New versus returning customer revenue, never blended into one total
- Repeat purchase rate and time between first and second order
- Customer acquisition cost by channel, measured against contribution rather than revenue
- Email and owned-channel share of revenue
- Branded and direct demand over time, as the measure of whether the brand is growing
- Marketplace share of total revenue, tracked as a concentration risk
- Return rate by product and by category, because it silently decides which lines are worth advertising
What usually goes wrong
Where ecommerce marketing tends to fail
Online revenue is four numbers multiplied together: how many people arrive, how many buy, what they spend, and how often they come back. Everything an ecommerce marketing programme does moves one of those terms.
Most programmes cannot say which. Spend goes up, revenue goes up, the ratio between them slowly worsens, and nobody notices until a margin review. The work below starts by identifying which term is actually limiting the business, because the answer changes the entire plan.
- Nobody can say which term of the equation moved.
- Traffic, conversion rate, average order value and repeat rate respond to completely different work, and blending them into a single revenue chart hides which one is stuck. A store with strong traffic and a two per cent conversion rate does not need more media; a store converting well with no returning customers does not need a redesign. Diagnosing this before recommending anything is the difference between a plan and a package.
- The catalogue architecture happened by accident.
- Categories were created as the range grew, products sit in several of them, some collections exist only because a campaign needed a landing page, and nothing reflects how customers actually search. The result is that the pages matching real demand either do not exist or are thin grids with a headline. Category structure is the highest-leverage organic work in ecommerce and it is almost always treated as a merchandising afterthought.
- Faceted navigation is publishing pages you never wrote.
- Filters for size, colour, price and availability commonly generate crawlable, indexable URLs. A modest catalogue becomes hundreds of thousands of near-duplicate pages, crawl activity moves away from the pages that matter, and duplicate content appears at a scale nobody intended. Deciding which facet combinations deserve to exist, which should be crawlable and which should be closed off entirely is unglamorous work with disproportionate effect.
- Acquisition costs rise and the margin cannot absorb it.
- Paid acquisition on the major platforms has become steadily more expensive while cost of goods, shipping and returns have not become cheaper. A campaign hitting its target ratio can still be losing money on a low-margin line. Until bidding is informed by product-level margin rather than revenue, the platform will optimise towards whatever sells easily, which is frequently what you make least on.
- Retention is somebody's side project.
- Email and lifecycle work usually sit with whoever has spare capacity, run as a weekly promotional send, and are judged on open rates. Meanwhile a second order from an existing customer costs a fraction of the first. In a market where acquisition keeps getting dearer, repeat rate is the term with the most profit attached and the least attention paid to it.
- The marketplace is both your biggest channel and your competitor.
- Selling through a marketplace brings volume you could not buy, and it also outranks your own product pages for your own products, keeps the customer relationship and sets the terms. It is a reasonable channel and a poor foundation. The question worth answering early is what share of revenue you could lose overnight without a plan, and what owned demand would have to exist to make that survivable.
Search behaviour
What your customers are typing
Category and product searches
The commercial core. Category pages match these far better than product pages do.
- womens running shoes wide fit
- best [product type] for [use case]
- buy [product] online free shipping
- [material] [product] under [price]
- [brand] [product] price
Comparison and validation
The step where the purchase is actually decided, and usually the step where the shopper leaves your site.
- [brand] vs [brand]
- [brand] reviews reddit
- is [brand] legit
- [product] size guide
- alternatives to [product]
Problem-led discovery
Earlier, cheaper and much less contested. Where content earns customers a marketplace never sees.
- how to stop [problem] at home
- what to get someone who [interest]
- how to choose a [product type]
- do i need [accessory] for [product]
Post-purchase and service
Low glamour, high volume, and directly attached to whether a second order happens.
- [brand] returns policy
- where is my order [brand]
- how to track [brand] order
- [product] care instructions
These are examples of how customers in this market search, drawn from keyword research and from the questions that come up on sales calls. They are illustrative, not a volume claim — the actual demand in your area is something we size before recommending anything.
Buying behaviour
How your customers actually decide
Strategy follows this, not the other way round. Everything on this page is downstream of how the decision genuinely gets made.
- Most purchases involve a validation step somewhere else. Shoppers leave to check reviews or compare on a marketplace, then return through a different channel entirely, which is why last-click attribution consistently misprices the top of the funnel.
- Delivery cost and speed decide a large share of abandonments, and they decide it late. A shopper who has chosen the product will still leave at the shipping step, so surfacing the real total early tends to reduce checkout loss more than any persuasion on the product page.
- Returns policy is read before purchase far more often than retailers expect, particularly in apparel and anything with fit or size uncertainty. A clear, generous policy reduces hesitation at the point of decision even though it costs more to service.
- Product page detail beats persuasion. Dimensions, materials, sizing, compatibility and what is actually in the box remove the specific doubts that stop a purchase; adjectives do not.
- Reviews carry more weight than any brand claim, and the useful signal is the middle of the distribution. Shoppers read the three-star reviews looking for the recurring complaint, which is why an unbroken run of five stars reads as less trustworthy rather than more.
- Repeat purchases are decided by the post-purchase experience, not the marketing. Delivery accuracy, packaging, the first use of the product and how a problem was handled determine whether the second order happens at all.
Where the money goes
The channels that earn their place here
In priority order for this business, not a menu. Anything not on this list is something we would need a specific reason to recommend.
Category architecture first, products second
Category and collection pages match how people search and carry link equity down to the products, yet on most stores they are unedited grids. Restructuring them around real demand, writing them so they deserve to rank, and fixing the internal linking underneath is the organic work with the most revenue attached and the least competition from marketplaces.
Retention as the profit line, not a newsletter
When acquisition costs rise against fixed margins, the second order is where profit comes from. That means lifecycle flows rather than weekly promotions: post-purchase sequences, replenishment timing based on the actual product, win-back before the customer has gone cold, and segmentation that stops you discounting to people who would have paid full price.
Acquisition held to contribution margin
Paid is the fastest way to buy volume and the fastest way to buy unprofitable volume. We feed product-level margin into the bidding rather than revenue, separate prospecting from retargeting so the reporting stops flattering itself, and accept that some best-selling lines are simply not worth advertising.
Fix the term that costs nothing to improve
Conversion rate is the only term in the equation that improves revenue without increasing spend. The work is rarely a redesign: it is shipping cost visibility, product detail that answers the actual objection, review presentation, search and filtering that works, and a checkout that does not lose people at the payment step.
Crawl control on a large catalogue
Faceted navigation, out-of-stock handling, pagination, variant duplication and slow templates are the recurring technical failures in ecommerce. They are invisible on a small site and severe on a large one, and fixing them frequently releases organic performance that no amount of content work would have produced.
A storefront that can carry the plan
Some of this cannot be fixed in a theme. Category templates that support real content, filter behaviour you control, fast product pages under load and a checkout that works on a mid-range phone are platform decisions, and a replatform is worth doing when the current build is the thing capping conversion.
The website
What the site has to do for this customer
- Category pages that are real pages — merchandised, written, and structured around how people actually search
- Faceted navigation with explicit rules about which combinations may be crawled and indexed
- Product pages carrying the specifics that remove doubt: dimensions, materials, sizing, compatibility, box contents
- Delivery cost and timing visible before the checkout, not revealed at the payment step
- Returns policy written plainly and linked from the product page rather than buried in the footer
- Reviews on the product page with the full distribution visible, not a curated selection
- Internal search and filtering that works, because a shopper who searches and finds nothing leaves immediately
- Product images that load fast at full quality on a mid-range phone on mobile data
- A checkout tested on real devices, with guest checkout and the payment methods your market expects
- Structured data on products that matches what is actually on the page, including price and availability
Constraints
What the rules allow, and what they do not
Consumer protection rules govern more ecommerce marketing than most operators expect: how a sale price is justified against a previous price, what a stated delivery time commits you to, how subscription renewals must be disclosed, and what a returns policy must say as a minimum. The requirements differ by country and by state or province, and enforcement in this area has been active. We write to whatever applies to your markets and ask you to confirm the detail with your legal adviser.
Tracking and personalisation carry a separate obligation. Consent requirements for analytics and advertising cookies, and the disclosures required when customer data is used for targeting, vary by jurisdiction and are increasingly enforced. This affects measurement design as well as copy, so it is decided at the start of an engagement rather than retrofitted.
Product claims are the third area. Anything asserting a health, environmental or performance benefit needs substantiation, and sustainability language in particular is now scrutinised closely in several markets. Where a claim cannot be evidenced we remove it rather than qualify it into something vague.
Questions
Questions we get from this industry
Our category is dominated by marketplaces. Is organic search worth it?
Yes, but not by fighting them on the terms they win. A marketplace has authority and breadth no single brand can match on generic product queries, and spending two years attacking that head-on is a poor use of a budget.
Where a brand wins is on its own name, on specific category and attribute searches that need genuine merchandising, and on problem-led content that a marketplace has no reason to publish. Those searches also produce customers who arrive knowing who you are, which is worth considerably more than a marketplace order.
How do we know whether to work on traffic or conversion?
By looking at where the equation is unbalanced. If your conversion rate is well below what is normal for your category and device mix, more traffic makes the problem more expensive rather than better. If conversion is healthy and volume is flat, the constraint is upstream.
The diagnostic usually takes a couple of weeks and it changes the whole plan, which is why we do it before recommending any channel. It is also common to find the real constraint is repeat rate, which neither traffic nor conversion work would have addressed.
Why is faceted navigation such a problem?
Because it multiplies. A catalogue with a few thousand products and filters for size, colour, price and brand can generate hundreds of thousands of crawlable URLs, most of them near-identical and none of them written by anyone.
The effect is that crawl activity spreads across pages with no commercial value while your new lines and important categories get visited less often. The fix is deciding which facet combinations reflect real demand and deserve to be indexed, and closing off everything else properly.
We have tried email and it just produced discount-hunters. What changes?
Usually the structure. A weekly promotional send to the whole list trains people to wait for the next offer, and the revenue it produces is largely revenue you would have had at full price.
Lifecycle flows behave differently because they are triggered by behaviour rather than by the calendar: post-purchase onboarding, replenishment timed to how long the product actually lasts, browse and cart recovery, and win-back before someone has drifted. Segmenting so full-price buyers stop receiving discounts is often the single largest margin improvement available.
Should we reduce our reliance on the marketplace we sell through?
Reduce the dependency, not necessarily the revenue. Marketplace sales are real sales and walking away from them to make a point is rarely sensible.
What is worth doing is building enough owned demand — organic visibility, a list, branded search, repeat customers — that a change in the platform's terms is a bad quarter rather than a crisis. A useful test is asking what share of revenue would vanish tomorrow if your account were suspended, and whether you could live with the answer.
How long before organic work affects revenue?
Technical and category structure fixes can move within one to three months because they release performance on pages that already exist, and they are usually where we start for that reason.
New content and authority work is slower, commonly three to six months before it carries meaningful revenue and longer in competitive categories. We sequence so the faster work is producing while the slower work compounds, and we are explicit about which is which.
Find out what is realistically winnable in your market
A strategy call is a working session on your ecommerce business specifically — your area, your competitors, the searches that matter and what it would take to compete for them. If we do not think we can move it, we will tell you.
If we don't deliver the work we agreed to deliver for reasons within our control, you don't pay for the undelivered work. Read our guarantee
Last updated · Reviewed by Zubair Afzal