Skip to content
Faisal Mahmud TusherDigital marketing, SEO & analytics
with Faisal Tusher

Performance Marketing › Intermediate

Capstone: A Performance Marketing Programme for a Stationery Store

The capstone brings the intermediate level together in one programme for an invented online stationery store heading into its gift season. You set targets from contribution margin, split the budget across channels on marginal returns, plan how the strongest line scales and design the incrementality test that decides the next budget.
Capstone projectIntermediate levelModule 3: Testing, scaling and incrementality7 minUpdated

Brief

Inkmeadow Stationery is an invented online shop selling fountain pens and ink, notebooks and paper, and boxed gift sets. The owner runs the business with one part-time marketer. Paid media has grown year on year by adding a little to whichever channel reported the best return, and the owner now suspects some of that money is buying sales the shop would have made anyway.

The next quarter is the gift season, when gift sets sell far more than usual. The owner wants a programme for that quarter that answers three questions: what each channel should be asked to return, where the budget should go, and what test would show whether the remarketing and prospecting money is earning its keep.

Inputs and constraints

Everything below comes from the shop's order system, its analytics and the ad platforms, for the last full quarter and an earlier quarter with similar seasonality.

Illustrative example — not a real result

Inkmeadow Stationery: margins and revenue mix

Product groupContribution marginShare of revenue, normal quarterShare of revenue, gift season
Fountain pens and ink45%30%20%
Notebooks and paper40%50%40%
Gift sets30%20%40%

Contribution margin is after product, packaging, delivery, payment fees and expected returns. The owner wants to keep at least 10 percentage points of revenue as profit after ad cost.

Illustrative example — not a real result

Inkmeadow Stationery: channels at two spend levels

ChannelBiddingEarlier spendEarlier revenueLast quarter spendLast quarter revenueNotes
Shopping, all productsOne target ROAS of 4.0£7,000£30,400£9,000£36,000No split by product group
Search, non-brandTarget ROAS 3.0£4,000£14,000£4,000£14,000Loses 30% of impression share to budget
Search, brandManual CPC£800£16,000£800£16,000People searching for Inkmeadow by name
Social, prospectingMaximise conversion value£3,000£8,500£5,000£12,50070% of its buyers are new customers in the order system
Social, remarketingMaximise conversion value£2,200£13,200£2,200£13,200High frequency; never tested

Revenue is as each platform reports it. Last quarter the platforms together claimed £91,700, while the shop's own analytics, counting each order once, credited £74,000 to paid media.

Illustrative example — not a real result

Inkmeadow Stationery: budget for the gift season

The gift-season budget is £24,000 in total, including any test reserve, against £21,000 last quarter. The test reserve may not exceed 10% of the total, so £2,400 at most. Any money moved between channels must stay within the £24,000, and the owner will review the split monthly rather than weekly. The marketer's time is fixed at the same part-time hours as last quarter.

Further constraints:

  • The marketer can run one controlled test in the quarter, alongside routine management.
  • No website changes during the busiest weeks, so landing page work must be finished early or wait.
  • The owner accepts counting a new customer's expected repeat purchases only if the assumption is written down.

Task

Produce one document with these sections, in this order. Use the named site resources, show your arithmetic in tables, and state expected effects as a direction with a way to check them rather than an invented uplift.

  1. Targets from margin. Calculate the Break-even ROAS and target return for each product group and for the gift-season blend, using the break-even calculator and the ROAS calculator. Decide whether Shopping should be split by product group and whether any campaign should bid to a cost target instead, applying the target CPA versus target ROAS comparison.
  2. Reconcile channel targets. Use the gap between platform-claimed and analytics-credited revenue to set channel targets that, weighted by spend, still deliver the blended target, as the cross-channel performance planning guide describes.
  3. Assign roles and judge each channel by its role. Give every channel a role from the planning guide and say what it is judged on. Treat brand search and remarketing separately from the rest.
  4. Split the budget on marginal returns. Calculate the marginal return for every channel with two spend levels, compare it with the relevant target and propose the gift-season split. Lay it out on the media plan template, with flighting across the quarter following the media planning framework.
  5. Plan the scaling. Name the line that should scale first, the step size, the review point and the marginal return at which scaling stops, using the scaling paths and guardrails in the testing and scaling guide.
  6. Design one incrementality test. Choose between the remarketing holdout test and the budget increase incrementality test, and justify the choice. Write the hypothesis, metric, guardrails, run time and decision rule into the experiment hypothesis template, and say how the budget changes under each possible result.
  7. Summarise for the owner. One page: the targets, the split, the scaling step, the test and the decision it informs, with the assumptions listed.

Evaluation criteria

CriterionWhat a strong answer showsCommon gap
TargetsBreak-even and target return per product group and for the gift-season blend, all from contribution marginOne target for the whole account, or targets from last quarter's averages
SeasonRecognises the blended floor rises as gift sets take a larger shareCarrying a normal-quarter target into the gift season
Target typeShopping split by product group; target type chosen per campaign with a reasonKeeping one Shopping target because it "works on average"
ReconciliationChannel targets tightened by the observed over-claiming so the blend holdsEvery channel set to the blended target and the blend still missed
RolesBrand, demand capture, demand creation and remarketing each judged on a fitting measureProspecting judged on last-click return and cut
Marginal returnsMarginal return calculated per channel and compared with the target, not the channel's averageMoving budget towards the highest average return
ScalingA named line, staged steps, a review point and a written stop rule"Increase budget gradually" with no stop rule
TestOne test, chosen for the decision it informs, with a pre-written decision rule and a run time that fits the quarterTwo tests at once, or a test with no rule for acting on the result
AssumptionsRepeat-purchase and forecast assumptions written down beside the figures they affectHidden assumptions that make prospecting look better than the data
CommunicationAn owner summary readable in minutes, with directions rather than promisesInvented percentage uplifts

Check your arithmetic against the figures below after working them out yourself. They are not the only defensible answer; the criteria above judge the reasoning.

Illustrative example — not a real result

Reference figures for self-checking

  • Break-even and target return: pens 1 ÷ 0.45 ≈ 2.2 and 1 ÷ 0.35 ≈ 2.9; notebooks 2.5 and 1 ÷ 0.30 ≈ 3.3; gift sets 1 ÷ 0.30 ≈ 3.3 and 1 ÷ 0.20 = 5.0.
  • Normal-quarter blended margin: 0.3 × 45% + 0.5 × 40% + 0.2 × 30% = 39.5%, so break-even ≈ 2.5 and target 1 ÷ 0.295 ≈ 3.4.
  • Gift-season blended margin: 0.2 × 45% + 0.4 × 40% + 0.4 × 30% = 37%, so break-even ≈ 2.7 and target 1 ÷ 0.27 ≈ 3.7.
  • Over-claiming: £91,700 ÷ £74,000 ≈ 1.24, so platform-reported returns run about a quarter above what analytics credits.
  • Blended paid return on analytics revenue last quarter: £74,000 ÷ £21,000 ≈ 3.5, above the normal-quarter target but below the gift-season one.
  • Shopping marginal return: (£36,000 − £30,400) ÷ (£9,000 − £7,000) = 2.8, above break-even but below target.
  • Social prospecting marginal return: (£12,500 − £8,500) ÷ (£5,000 − £3,000) = 2.0 on platform revenue, below break-even before any repeat-purchase assumption.
  • Non-brand search: 3.5 reported and losing impression share to budget, the clearest candidate for a staged increase.
  • Social remarketing: 6.0 reported, unchanged at constant spend, and untested.

A strong programme notices that the gift season makes every target harder at the moment budgets rise, so the first increase goes to the line that is both above target and constrained by budget, not to the line with the highest reported ROAS. It also treats the remarketing figure as a claim to be tested: an Incrementality result is what decides whether that money stays or funds prospecting.

Submission checklist

0/10

When you have finished, compare your targets with the eyewear target-setting exercise, your split with the language school budget exercise and your test with the furniture remarketing holdout exercise, then return to the intermediate performance marketing level to mark the level complete.

Pages this lesson applies

Terms used on this page

Also related

Related