Ecommerce growth strategy:
choose your next profitable move.

Revenue is the headline.
Contribution pays the bills.

Decide what growth must leave behind—not only how much it must bring in.

Once you have a likely growth direction, put a financial target behind it. A plan to increase sales is incomplete until you know what those sales must contribute—and how much cash the business can risk.

Make the outcome specific: “By this date, improve this measure from this baseline to this target, without crossing this cash or service limit.” Increasing contribution, accelerating payback and reducing dependence on one market are different goals.

Follow one $100 order

A customer pays you $100. That does not mean you have $100 to spend on growth. Pay for the product, delivery, payment fees and other variable costs first.

What remains is contribution: money available towards fixed costs and, beyond those, profit. It is not the same as operating profit or cash in the bank.[1]

For this guide, acquisition is shown separately: net revenue − non-acquisition variable costs − acquisition spend = contribution after acquisition.

Net revenue includes shipping charged, excludes sales taxes collected, and is after discounts and refunded revenue. Product cost is landed cost; fulfilment includes outbound shipping.

The $7 covers unrecovered return-handling, replacement and variable support costs—not refunded revenue deducted twice. Use consistent cost boundaries and do not count acquisition in both this calculation and overhead.

Set a ceiling before raising the budget

At a $49 acquisition cost, that same order leaves only $1 before overhead. Higher customer volume would not automatically make the business stronger.

With $50 before acquisition and a chosen requirement to retain $18, your acquisition allowance is $32. That is a planning constraint, not proof that customers are available at that price.

Repeat purchasing may change the picture. Use observed repeat contribution after retention costs, and compare customer groups over the same elapsed time. Shopify’s cohort reporting groups customers by first purchase and follows subsequent behaviour.[2]

Do not compare a six-month-old group with a two-week-old group as though they had equal time to repurchase. And do not let projected lifetime value hide a payback period the business cannot finance.

That is why the free D2C masterclass treats offer, acquisition and economics as one connected decision rather than separate dashboard targets.

A Growth Cell is a customer opportunity.
Not another campaign.

Separate the market you want to win from the routes you use to reach it.

A Growth Cell is the smallest customer opportunity worth making a separate growth decision about. Within a brand or product family, define the geography, use case, buyer group and buying problem.

“People who buy bathroom products” is too broad. “UK households replacing a mat that stays damp in a busy shared bathroom” gives you a clearer research question and buying situation.

A flow is a path through discovery, evaluation and purchase. One cell can have several flows: search to product page, demonstration to educational page, or email resolving a buying objection.

The distinction stops a common planning mistake: calling every channel, ad angle or landing page a new growth opportunity. Those may be valuable execution changes inside the same opportunity.

Split only when the commercial decision changes

Now consider apartment renters choosing a mat for a compact bathroom. Dimensions and door clearance might matter more than performance between uses. That could justify a separate cell—but treat it as a hypothesis until research supports the distinction.

Ask whether the promise, proof, offer, buying journey, economics or operating requirements would materially change. A demographic difference alone is not enough.

Two age groups needing the same thing, responding to the same offer and producing similar economics may belong together. A complicated map is not a more accurate map simply because it has more rows.

Record only the cells relevant to your next decision. Label each as observed, partly supported or untested. A research idea should not look like an established revenue stream.

For a new brand, begin with one customer-job hypothesis. Validate need, product and offer before building an international growth architecture.

Is the pond smaller—
or is your fishing weaker?

Do not treat a demand problem and an execution problem as the same diagnosis.

When sales slow, the instinct is to improve the ads. Sometimes that is exactly right. Sometimes the buying problem is less urgent, an alternative is more attractive, or a market no longer supports the same economics.

The pond-and-fishing metaphor is useful because it keeps two questions separate: how attractive is the opportunity, and how well are we capturing it?

Start with measurement and buying conditions

First rule out broken tracking, out-of-stock products, payment failures and changed delivery promises. Compare equivalent periods and inspect country, product, device and traffic source rather than trusting one blended average.

A store-wide conversion decline could reflect weaker traffic, a changed visitor mix, missing stock or real friction. Until you separate them, a redesign may solve the wrong problem.

Then gather several kinds of evidence. Read customer questions and return reasons. Inspect relevant search demand and alternatives. Compare the same customer opportunity across different flows. Look for patterns that agree—or contradict your first explanation.

Follow the buyer’s unanswered question

For the hypothetical bathroom brand, “Will it fit?” suggests dimension and clearance evidence. “What will delivery actually cost?” suggests clearer delivered pricing. “Why choose this over what I already use?” suggests a value or product problem.

Baymard’s checkout research identifies reported barriers including extra costs, delivery speed and trust concerns. Those findings are useful prompts for investigation—not evidence that your store has the same problem or will achieve a particular uplift.[3]

Match the intervention to the evidence. You do not need a quiz because quizzes are fashionable, or a new channel because the existing one feels difficult. You need a credible explanation of what prevents the next profitable purchase. The same systems-first diagnostic approach runs through D2C Game Insights.

Same $120,000 revenue.
Very different money left.

Compare contribution after the costs of the growth plan—not just the sales forecast.

Return to the $100 first order, with $50 in non-acquisition variable costs. The baseline is 1,000 new customers, one first order each, and $32,000 in acquisition spend.

Now model two ways to reach 1,200 customers. One increases acquisition spend. The other assumes a better buying journey produces more customers from the same spend.

Plan A raises acquisition cost per customer from $32 to $36. Revenue grows by 20%, but contribution falls by $1,200.

Plan B leaves $28,000 after acquisition, or $26,000 after the assumed $2,000 implementation cost. That is $8,000 above baseline.

This does not prove conversion work is always better. The improvement may fail. Additional traffic may have stronger repeat purchasing. Your current journey may already work well.

The example proves something narrower: equal revenue outcomes can have very different economic value. Neither result is operating profit, and neither captures inventory funding or cash timing.

The next customer may not cost what the last one did

A profitable historical average does not make every expansion profitable. Ask what the additional spend produces, not only whether the account’s average still looks acceptable.

In Plan A, $11,200 of extra acquisition spend buys 200 additional customers: $56 each. Those customers add only $10,000 before acquisition. The extra spend reduces contribution by $1,200—even though average acquisition cost is $36.

Also separate attributed purchases from purchases caused by the activity. Google’s Conversion Lift reporting estimates incremental conversions rather than treating all attributed sales as additional sales.[4]

A suitable controlled test can strengthen that assessment. A before-and-after improvement alone may also reflect seasonality, other campaigns or stock changes. When evidence is limited, keep the commitment proportionate and reversible.

Put cash and evidence
before the excitement score.

A growth idea earns a commitment by reducing meaningful uncertainty—not by winning a presentation.

Compare candidate bets using the same questions: target cell, suspected constraint, expected contribution, supporting evidence, downside cash exposure, time and delivery requirements.

Use low, base and high cases. State which assumptions drive the difference. A precise forecast built on weak assumptions is still weak evidence.

Give different evidence different commitments

For the hypothetical bathroom brand, recurring size questions justify testing clearer fit information. Research suggesting a compact-bathroom need may justify a message-and-offer pilot.

An Australian launch still needs local validation. A new product supported only by a concept needs demand and supply investigation before a substantial inventory order.

These four ideas should not receive equal budgets. Nor should they be compared only on projected revenue: they are at different stages of evidence.

A waitlist records interest, not completed purchases. Early sales improve the evidence but do not establish repeatable acquisition, low returns or operational reliability. Match the next investment to what the test has actually demonstrated.

For a capacity-constrained team, my starting point is one main growth bet plus a small research track. Finish a learning cycle before adding more work. Your output is a decision brief: cell, constraint, test, economics, owner and stop condition.

Write a test brief—not a task list

A useful brief for the first opportunity would read:

In this scenario, I would start with the fit question because there is a specific observation and a bounded test. That is a judgement about this example—not a universal rule that product-page work comes first.

Your next 90 days:
diagnose, test, then scale—or stop.

The calendar should produce better decisions, not just more completed tasks.

Ninety days is a planning horizon, not a promise that every opportunity can be validated in a quarter. Supplier lead times, buying cycles, traffic and return windows may require longer.

Days 1–14: establish what is true

Reconcile the numbers. Review purchases, questions, returns and operating constraints. Write down the baseline and separate observations from explanations.

Finish with a target outcome, one cell, a likely constraint and a comparison of possible bets. Avoid rebuilding the store before deciding what the rebuild must fix.

Days 15–30: build the smallest complete route

Small does not mean disconnected. A use-case ad leading to an irrelevant page may test inconsistency rather than the opportunity you intended to test.

Connect message, destination, truthful proof, offer, checkout and measurement. Assign one decision owner and identify the product, financial and operational inputs they need.

For a controlled test, define eligibility, assignment, a commercially meaningful effect, sample requirements and the analysis method before launch. A different test design may be needed when traffic cannot support the intended comparison.

Days 31–60: inspect early; do not celebrate early

Check for broken events, allocation problems, stock differences and obvious customer harm as the test runs. Quality control should start immediately.

That is different from repeatedly checking a result until it looks significant. Microsoft’s experimentation guidance highlights the need to account for early peeking and repeated measurements in the statistical method.[5]

Allow relevant buying and return windows to mature, or clearly identify estimates for later returns. With low volume, combine research and a bounded pilot without calling a noisy result statistical proof.

Record whether the hypothesis failed, the execution failed or the result was inconclusive. Those are three different lessons—not one bucket labelled “didn’t work”.

Days 61–90: make the next commitment

Compare the result with the original economic target and guardrails. Scale in steps when the evidence supports it. Revise when you have learned that the constraint is different. Stop when the opportunity cannot meet your requirements at an acceptable cost.

Document what appears transferable: the customer job, message, proof, offer or operating setup. That becomes useful input for the next flow or geography.

Finish the quarter with a decision and a clearer model of the business—not just a longer list of things the team shipped. If you want direct help turning that decision into a focused operating plan, explore ways to work with D2C Game.