The Complete Guide to eCommerce Marketing
Most eCommerce marketing advice is a list of channels. This guide is the other thing: how to work out which constraint is actually holding your store back, what each channel is genuinely for, and how to tell whether any of it worked once attribution stopped being reliable.
CHAPTERS
8 chapters
READING TIME
37 min read
LAST UPDATED
18 September 2026
LEVEL
Intermediate
Almost every marketing decision in this guide is made before a single campaign is built.
EXECUTIVE SUMMARY
What this guide argues, in five points
eCommerce marketing is not a channel problem. It is a constraint problem wearing a channel costume. Stores that grow reliably are not the ones running the most channels; they are the ones that identified the single thing limiting them, fixed it, and only then spent money on top of a machine that converts and retains.
- There is one binding constraint at any moment — traffic, conversion, margin or repeat rate. Spending on the other three is expensive theatre.
- Contribution margin after variable costs, not revenue and not ROAS, is the number every channel decision should be measured against.
- Channels are not interchangeable. Search harvests existing demand, paid social creates it, retention compounds it. Confusing the three is the most common budget mistake we see.
- Conversion rate is a multiplier on every channel at once, which is why it is almost always cheaper than the next increment of traffic.
- Attribution broke and is not coming back. Incrementality testing and contribution reporting are now the honest answer, not a nice-to-have.
The eight chapters below are the detail behind those five points, in the order we work through them when we take over an account.
- Contents
Eight chapters
Read it in order the first time. After that, the chapter index is the useful part.
01
Every store has one thing holding it back at any given moment. Marketing spent anywhere else is spent at a discount.
4 min
02
Four numbers decide what you can afford to spend to acquire a customer. Get them wrong and no amount of channel skill rescues the account.
5 min
03
Demand you can harvest and demand you have to create are different businesses. Most plans fail because they treat them as one.
4 min
04
Search harvests demand, paid social creates it, retention compounds it. Most budget waste is a channel being asked to do the wrong one of those three.
6 min
05
Every point of conversion rate improves search, paid, email and marketplaces at the same time. Nothing else in marketing does that.
4 min
06
The second order costs almost nothing to win and is worth the same as the first. That gap is the whole argument for lifecycle work.
5 min
07
Targeting is mostly automated now. Creative is the variable you still own, and the gap between concepts is wider than the gap between audiences.
4 min
08
Platform numbers no longer add up to the bank balance. The fix is not a better tracking script; it is a reporting model that expects the gap.
5 min
- Before you start
Who this guide is for
Written for the people who carry the number, whether or not they build the campaigns themselves.
Founders and heads of eCommerce
You are deciding where the next quarter of budget goes and want a way to choose that is not a benchmark deck or a vendor pitch.
In-house marketing managers
You are running several channels at once and need a defensible order of work rather than a list of forty tactics.
Teams reviewing an agency
You want to know which questions a marketing partner should be able to answer before you sign anything.
NOT WRITTEN FOR
People looking for a list of growth hacks. There is not one here. Everything in this guide assumes you would rather have a smaller number of decisions you can defend than a longer list of tactics you cannot.
CHAPTERS
- 01 Start from the constraint, not the channel
- 02 The unit economics that decide everything
- 03 Mapping demand before choosing channels
- 04 What each channel is actually for
- 05 Conversion: the multiplier on every channel
- 06 Retention, where the margin actually lives
- 07 Creative and content as the real lever
- 08 Measuring marketing when attribution is broken
CHAPTER 01
4 min
Start from the constraint, not the channel
Every store has one thing holding it back at any given moment. Marketing spent anywhere else is spent at a discount.
The question that opens almost every engagement is some version of “should we be doing more paid social?” It is the wrong question, because it assumes the constraint is traffic. Four things can limit an eCommerce business, and only one of them is binding at a time in any useful sense.
Constraint
What it looks like
What actually fixes it
Traffic
Good conversion rate, healthy margin, repeat customers — simply not enough sessions
Search, paid media, marketplaces, partnerships
Conversion
Sessions arrive, add-to-cart is reasonable, checkout leaks
Product pages, merchandising, checkout, trust, speed
Margin
Revenue grows, the bank balance does not
Pricing, product mix, shipping policy, discount discipline
Repeat rate
Acquisition works, everybody buys once
Lifecycle flows, product cadence, service, subscription
If you cannot name which row you are in, that is the first piece of work — not the campaign build.
KEY TAKEAWAY
Marketing that ignores the constraint still produces activity, reports and invoices. It does not produce growth, because the bottleneck never moved.
How to name the constraint in an afternoon
1
Pull twelve months of sessions, conversion rate, average order value and gross margin. Monthly, one sheet, no attribution modelling yet.
2
Compare each against a realistic benchmark for your category — not a global average, which is meaningless across fashion and industrial parts.
3
Identify the metric furthest below where it should be. That is your candidate constraint.
4
Test it with a thought experiment. If you doubled traffic tomorrow, would the business be meaningfully better, or would you pay twice as much to lose people at the same checkout step?
The answer to step four decides where the next quarter of budget goes. If doubling traffic does not obviously double the outcome, traffic is not your constraint, and buying more of it is the most expensive way to find that out.
Our twelve-point growth assessment is the long version of this chapter, with a threshold attached to each check.
CHAPTER 02
5 min
The unit economics that decide everything
Four numbers decide what you can afford to spend to acquire a customer. Get them wrong and no amount of channel skill rescues the account.
Marketing decisions are affordability decisions. Until you know what a customer is worth and what one costs, every channel conversation is an argument about opinions. The arithmetic is not complicated, but almost nobody does it at the level that matters, which is contribution rather than revenue.
Contribution, not revenue
Revenue pays for nothing. Contribution margin — what is left after cost of goods, payment fees, pick, pack and shipping, and returns — is the pool that has to cover marketing, overheads and profit. A channel returning 4x on revenue can still lose money on a 22% contribution product.
Order value
£80
What the customer paid
Contribution
£26
After goods, fees, fulfilment and returns
Break-even CAC
£26
On the first order alone
Break-even CAC at 1.9 orders
£49
Once realistic repeat behaviour is included
A worked example on a typical 33% contribution product. The second number is the one that decides whether a channel is affordable.
KEY TAKEAWAY
The gap between £26 and £49 is the entire argument for investing in retention before scaling acquisition. Repeat rate does not just add revenue — it raises the price you can afford to pay for every new customer.
The four numbers to hold
- Contribution margin percentage, by product group rather than blended. Blended hides the products that are quietly funded by the good ones.
- Break-even CAC on first order and on twelve-month value. Two numbers, two different decisions.
- Payback period. How many months until a cohort has repaid its acquisition cost. This is a cash question, not a marketing one.
- Repeat rate at 90 and 365 days. The single input that most changes what you can afford to spend.
Hold those four and channel arguments get short. A campaign either clears break-even CAC on the cohort horizon you can finance, or it does not. Everything else — the platform, the creative format, the agency’s dashboard — is downstream of that test.
EXPERT VIEW
“The most common thing we find in a new account is not a bad campaign. It is a profitable-looking account selling a product mix that cannot support the CAC it is paying, with nobody having written the contribution number down.”
Dazzle Commerce · Growth strategy team
CHAPTER 03
4 min
Mapping demand before choosing channels
Demand you can harvest and demand you have to create are different businesses. Most plans fail because they treat them as one.
Before choosing channels, find out how much demand already exists for what you sell and how people phrase it. Existing demand is cheaper to capture than new demand is to create, and the split between the two decides your whole channel mix.
Category demand rarely arrives on one surface. Map it before deciding where to spend.
Three buckets, three different jobs
Demand type
How it shows up
What captures it
Branded
People searching your name, returning directly
Cheap to defend, dangerous to count as marketing performance
Category
People searching the product type without a brand in mind
Organic category pages, Shopping, non-branded search, marketplaces
Latent
People who would buy but are not looking yet
Paid social, creators, video, PR, partnerships
Most reporting merges the first two, which makes every channel look better than it is and hides whether you are actually growing.
KEY TAKEAWAY
If you cannot separate branded from non-branded in your reporting, you cannot tell growth from harvesting. That separation is worth more than most new campaigns.
Sizing category demand honestly
- Start from the category terms, not the head term. The head term is expensive, generic and often not how buyers actually search once they know what they want.
- Read the results page, not the volume column. If a query returns nine articles and one retailer, it is not a commercial query however large the number looks.
- Check marketplace search alongside search engines. In several categories more product discovery starts on Amazon than on Google, and that changes the plan.
- Look for the phrasing gap. Buyers rarely use your internal merchandising language. The distance between the two is usually a list of pages you have not built.
The output of this chapter is not a keyword list. It is a one-page statement of how much of your addressable demand is already being searched for, how much has to be created, and which surfaces it lives on. Everything in the next chapter follows from it.
CHAPTER 04
6 min
What each channel is actually for
Search harvests demand, paid social creates it, retention compounds it. Most budget waste is a channel being asked to do the wrong one of those three.
Channels are not interchangeable buckets you pour budget into. Each one does a specific job, and the most expensive mistakes in eCommerce marketing come from asking a channel to do a job it is structurally bad at.
Channel
The job it is genuinely good at
What it is bad at
Organic search
Capturing category demand repeatedly at near-zero marginal cost
Speed. Nothing meaningful happens inside a quarter.
Shopping and paid search
Harvesting in-market demand the day you switch it on
Creating demand that does not already exist
Paid social
Creating demand and introducing products people were not looking for
Efficiency at the bottom of the funnel on a commodity product
Marketplaces
Reaching buyers who start their search inside a marketplace
Owning the customer relationship or the data
Email and SMS
Turning existing customers into repeat contribution
Growth. It multiplies a base; it does not create one.
Creators and affiliates
Borrowed trust in categories where proof is the barrier
Predictability and clean measurement
Run this table against your current budget split. The mismatches are usually obvious within a minute.
A defensible starting split
There is no correct budget split, but there is a defensible starting point, and it comes from the constraint you named in chapter one rather than from a benchmark deck.
- Constraint is traffic, demand exists: weight to Shopping, non-branded search and the organic category layer. You are harvesting, and harvesting is cheap.
- Constraint is traffic, demand does not exist yet: weight to paid social and creators. You are paying to be introduced, and that costs more per first order.
- Constraint is conversion: hold acquisition flat and move budget to conversion work. It is the only lever that improves every channel at once.
- Constraint is repeat rate: hold acquisition flat and move budget to lifecycle. Raising repeat rate raises the CAC you can afford everywhere else.
KEY TAKEAWAY
Do not run a channel because a competitor does. Run it because you have named the job you need done and it is the channel that does that job best at your margin.
The order channels should be added
1
Defend and measure what you already have. Branded search, direct, existing customers. Cheap, and it establishes the baseline everything else is measured against.
2
Harvest existing category demand. Shopping and non-branded search first, because they pay back inside weeks and tell you what actually sells.
3
Build the organic category layer in parallel. It is slow, so it has to start early, not once paid gets expensive.
4
Add demand creation once conversion and retention hold. Paid social into a store that does not convert is a subsidy for your competitors’ remarketing.
5
Open a second channel — marketplace or market — only when the first is stable. Operational load arrives immediately; the revenue does not.
Teams that skip steps one and two and open with paid social usually produce a quarter of impressive-looking activity, a CAC nobody can defend, and no durable asset at the end of it. Teams that run the order above are slower for a quarter and considerably harder to compete with afterwards.
CHAPTER 05
4 min
Conversion: the multiplier on every channel
Every point of conversion rate improves search, paid, email and marketplaces at the same time. Nothing else in marketing does that.
Conversion rate is the only number that improves every channel simultaneously. A move from 1.8% to 2.2% is a 22% increase in orders across search, paid, email and marketplaces at once, with no increase in media spend. That is why it usually beats the next increment of traffic on cost per additional order.
Where the leaks actually are
- Category and collection pages. Most stores optimise product pages and leave the page that decides which product gets seen almost empty.
- The product page itself. Imagery, specification completeness, delivery clarity and returns — in roughly that order of impact.
- Checkout. Unexpected shipping cost remains the single most cited reason for abandonment, and it is a policy decision, not a design one.
- Speed on the templates that carry revenue. Measured on a mid-range phone on a real network, not on a laptop in an office.
KEY TAKEAWAY
Before buying more traffic, work out what a 10% relative lift in conversion rate would be worth at your current volume. On most stores it is larger than the media budget being argued over.
Testing without fooling yourself
Most eCommerce stores do not have the traffic to run a clean A/B test on anything except the highest-volume templates. That is not an argument against testing; it is an argument for testing fewer, larger things and being honest about what a result means.
- Size the test before you run it. If the sample you can gather in a month cannot detect the effect you care about, the result will be noise whatever the tool says.
- Test changes large enough to matter. Button colour does not move a business. Merchandising logic, delivery messaging and page structure do.
- Stop reading a test early. Peeking at a running test and stopping when it looks good is the most common way teams convince themselves of a lift that is not there.
- Keep a log of every test, including losers. Losing tests are the cheapest information in the account and the first thing lost at handover.
Our CRO testing template carries the sizing maths and the significance calculation, so a verdict is arithmetic rather than opinion.
CHAPTER 06
5 min
Retention, where the margin actually lives
The second order costs almost nothing to win and is worth the same as the first. That gap is the whole argument for lifecycle work.
Acquisition buys a first order. Retention decides whether that order was worth buying. On most stores the difference between a marketing plan that works and one that quietly loses money is the second, third and fourth purchase — and those are won with product cadence, service and a small number of well-built flows, not with more sends.
The flows that carry the revenue
Flow
What it is for
Common mistake
Welcome
Establish who you are before the first purchase decision
One email with a discount, then silence
Abandoned cart and browse
Recover intent that already existed
Three identical reminders; no attempt to answer the objection
Post-purchase
Reduce anxiety, set up the second order
Transactional only, then straight back to campaigns
Replenishment
Arrive when the product is running out
Sent on a calendar rather than on consumption
Winback
Recover lapsing customers before they are gone
Triggered far too late, usually at twelve months
Five flows, built properly, out-earn a year of campaign sends on almost every store we have taken over.
KEY TAKEAWAY
Raising repeat rate does not just add revenue. It raises the CAC you can afford in every acquisition channel, which is why retention work often unlocks paid media that previously looked unaffordable.
Measuring retention without lying to yourself
- Report by cohort, not in aggregate. Aggregate repeat rate rises simply because the customer base ages. Cohorts tell you whether anything changed.
- Do not use open rate as a success metric. Mail privacy protection inflates it to the point of meaninglessness. Use clicks, orders and contribution.
- Attribute flow revenue conservatively. A customer who would have returned anyway should not be counted as recovered by the email that arrived first.
- Watch list health, not list size. A large list with falling engagement is a deliverability problem being stored up.
Retention is also where category structure matters. A store selling a two-year replacement cycle cannot be run on the same cadence as one selling a consumable, and applying consumable tactics to durables produces unsubscribes rather than orders.
CHAPTER 07
4 min
Creative and content as the real lever
Targeting is mostly automated now. Creative is the variable you still own, and the gap between concepts is wider than the gap between audiences.
Once targeting became largely automated, creative became the main variable a marketer still controls. On most paid social accounts the difference between the best and worst performing concept is larger than the difference between any two targeting strategies, and it is the only lever that has not been taken away by the platform.
Volume matters, but concept variety matters more. Ten versions of one idea is one test, not ten.
Concepts, not variations
Producing forty assets that all say the same thing in different fonts is not creative testing. A concept is a distinct argument for why somebody should buy — a problem, a comparison, a demonstration, a proof, an objection answered. Each concept deserves a handful of executions; each execution without a concept behind it is decoration.
Concept type
What it argues
Works best when
Problem and agitation
Here is the thing that annoys you and why it happens
The product solves a specific, nameable irritation
Demonstration
Watch it work, unedited
The benefit is visible in under five seconds
Comparison
Against the obvious alternative, honestly
You genuinely win on a dimension buyers care about
Proof
Other people like you bought it and stayed
Trust is the barrier rather than awareness
Objection
The reason you are hesitating, answered directly
Price, fit, delivery or returns is the blocker
Five concepts tested properly beats fifty variations of one. Rotate concepts, not colours.
KEY TAKEAWAY
Creative is a research output before it is a production output. The best-performing ad on most accounts is a customer’s own sentence, lightly filmed.
Where the ideas come from
- Reviews and support tickets. The phrasing customers use when they are not being marketed to is the highest-converting copy you will find.
- Pre-purchase questions. Every repeated question is an objection that belongs in an ad and on the product page.
- Returns reasons. They tell you which expectation the current creative is setting wrongly.
- The competitor set. Not to copy, but to find the argument nobody in the category is making.
CHAPTER 08
5 min
Measuring marketing when attribution is broken
Platform numbers no longer add up to the bank balance. The fix is not a better tracking script; it is a reporting model that expects the gap.
Last-click attribution was always a simplification. Consent banners, tracking prevention, in-app browsers and the shift of discovery onto surfaces that do not pass a referrer turned that simplification into a distortion. Pretending otherwise produces confident reports and bad decisions.
Platform-reported numbers and the numbers in the accounts stopped agreeing some time ago. Plan for the gap.
Three layers that work together
Layer
Question it answers
Cadence
Platform reporting
Is this campaign working relative to last week?
Daily, for optimisation only
Blended and contribution reporting
Did the business get better, after all costs?
Weekly and monthly
Incrementality testing
Would this revenue have happened anyway?
Quarterly, per channel
The mistake is using layer one to answer layer two questions. Platform reporting is a steering wheel, not a scoreboard.
KEY TAKEAWAY
If the sum of platform-reported revenue exceeds the revenue in your accounts, you are double-counting. That is not a reporting nuisance; it is the reason budgets get approved for channels that are not earning them.
The reporting set worth building
- Blended CAC and blended contribution at the top. Total marketing cost, total new customers, total contribution. Nothing to argue with.
- New versus returning customer revenue, separated. Growth is the first number; the second is the base being harvested.
- Non-branded organic landings by template. The honest measure of whether the organic layer is compounding.
- Cohort contribution by acquisition month. The only view that shows whether the customers you bought this quarter were worth it.
EXPERT VIEW
“We ask two questions of every marketing report. Does the sum of the channels equal the money in the bank, and does the report say what we should stop doing. Most reports fail both.”
Dazzle Commerce · Analytics team
- Worked examples
Three decisions, before and after
The three arguments that come up on almost every account we take over, and how they resolve.
BUDGET ALLOCATION
Scaling paid into a conversion problem
Before
Paid social budget raised 60% while site conversion sat at 1.1% on mobile
After
Budget held flat for one quarter; product template and delivery messaging rebuilt first
The extra spend was buying sessions into a template that lost them. Fixing the multiplier first made the same budget produce more orders, and made the eventual increase defensible.
REPORTING
Counting branded search as growth
Before
Blended ROAS reported at 4.2x, with branded search inside it
After
Branded separated; non-branded acquisition reported against contribution
Branded search harvests demand other channels created. Leaving it in makes acquisition look efficient and hides whether the business is actually reaching new people.
CHANNEL CHOICE
Opening a marketplace to fix a margin problem
Before
Marketplace launch planned to add volume on a 22% contribution product
After
Launch deferred; pricing, shipping thresholds and product mix corrected first
A marketplace adds referral fees to a product that could not already fund its own acquisition. More volume at negative contribution is a faster route to the same problem.
Anonymised from accounts we manage. The shapes recur; the numbers will not match yours exactly.
- Take these with you
Three checklists
Written to be worked through rather than read. Nothing here needs a tool you do not already have.
Before you plan
Half a day, once a year and after any major change.
- Binding constraint named in one sentence, with the number behind it
- Contribution margin calculated per product group, not blended
- Break-even CAC written down for first order and twelve months
- Repeat rate at 90 and 365 days known
- Branded and non-branded demand separated in reporting
- Category demand sized on search and on marketplaces
Every quarter
Ninety minutes with the numbers in front of you.
- Budget split re-tested against the current constraint, not last year’s
- Blended CAC and blended contribution reviewed against target
- New versus returning revenue reported separately
- One incrementality test planned or read
- Creative concept list refreshed from reviews and support tickets
- Lifecycle flows checked for silent failures and stale content
Before adding a channel
Fifteen minutes that regularly saves a quarter.
- The job this channel does is named, and no existing channel does it better
- Break-even CAC on this channel is affordable at current contribution
- The store converts well enough for the traffic not to be wasted
- Operational load of the channel is understood and staffed
- A stop rule is agreed before launch, not after
- Measurement is in place before the first pound is spent
Copy these into your own tracker. There is no download form and no email wall.
- Frameworks
Two frameworks worth keeping
The two decisions that get reargued every quarter, written down so they stop being reargued.
FRAMEWORK ONE
The channel affordability test
Before a channel is debated on strategy, it should pass arithmetic. Take contribution per order, decide the cohort horizon you can finance, and compare the channel’s realistic cost per new customer against it. Most arguments end here.
Contribution per order
Orders per customer, year one
Affordable CAC
£15
1.2
£18
£26
1.9
£49
£40
1.4
£56
£65
2.6
£169
Affordable CAC here is break-even, not target. Decide separately what margin you want left over, and finance the payback period.
If a channel cannot reach the right-hand column at realistic volume, no amount of optimisation will rescue it.
FRAMEWORK TWO
The marketing priority order
When everything needs doing, the order matters more than the list. Each step assumes the one above has cleared, because work at step five is wasted while step one is wrong.
1
Measurement integrity. Branded separated, contribution reportable, platform numbers reconciled to the accounts. Without this, nothing below can be judged.
2
Unit economics. Contribution and break-even CAC per product group. This sets what is affordable everywhere else.
3
Conversion. The multiplier on every channel at once, and almost always cheaper than the next increment of traffic.
4
Retention. Raises the CAC you can afford, which changes which acquisition channels are viable.
5
Harvest existing demand. Shopping, non-branded search, and the organic category layer started in parallel.
6
Create new demand. Paid social, creators and video, once the four steps above hold.
New sales channels and new markets sit outside this order. Settle the six steps on your primary channel before adding a second.
- In summary
Key takeaways
Eight sentences you could hand to somebody who is not going to read the guide.
1
Name the binding constraint before choosing a channel. Traffic, conversion, margin and repeat rate are four different businesses to fix.
2
Contribution margin, not revenue and not ROAS, is the number every channel decision should be measured against.
3
Separate branded from non-branded demand or you will mistake harvesting for growth for as long as the reports last.
4
Search harvests demand, paid social creates it, lifecycle compounds it. Asking a channel to do the wrong job is the most common budget error.
5
Conversion rate is the only lever that improves every channel at once, which usually makes it cheaper than the next increment of traffic.
6
Raising repeat rate raises the CAC you can afford everywhere, so retention work often unlocks acquisition that looked unaffordable.
7
Creative is the variable you still control. Test concepts, not colour variations, and take the concepts from customers rather than from competitors.
8
Expect platform numbers and bank numbers to disagree. Blended contribution and incrementality testing are the honest answer.
If you want a single first action: calculate contribution per order on your top product group and compare it to what you currently pay for a new customer.
- Go deeper
Where this work sits in the practice
Three services that pick up the chapters above.
Full-funnel growth run as one plan across search, paid, creative and retention, measured on contribution.
The multiplier from chapter five, run as a sized testing programme rather than a list of opinions.
The five flows from chapter six, built, segmented and measured by cohort contribution.
- Keep reading
Related guides
The rest of the guide library, all published in full.
Complete guide · 37 min
The business the channels sit inside: the growth equation, stages, cash and the operating cadence.
Complete guide · 38 min
The long version of the organic half of chapter four.
Complete guide · 38 min
The long version of the paid half: profit targets, account structure, feeds and incrementality.
Want this run on your store rather than read about it?
Our marketing engagements open by naming the binding constraint and modelling contribution by channel before we touch a single campaign.