Sew Heart Felt
Full Audit

A felt-animal gifting brand, not a slipper brand

Share of the last 12 months’ £1.47M line-item revenue. Every demand peak is a gifting occasion, and each occasion has its own hero animal.

Monthly DTC revenue, Aug 2023 – Aug 2026

Two occasion peaks and a deep summer trough. Mothering Sunday’s date decides whether Feb or Mar carries the spring peak, so always read Feb+Mar combined: £286k → £452k → £567k (+25% YoY). The Mar-2026 £380k “spike” is verified real: £337k DTC, a paid Mother’s Day blitz, not wholesale.

Channel share of revenue (de-duplicated, linear attribution)

Meta leads in steady state; Google overtakes Meta in Q4 as Christmas shoppers search and Shopping/PMax harvests the demand. Email holds 10–13%. Triple Whale pixel, every order’s revenue split equally across its touchpoints.

Platform-claimed revenue by month (trend only)

Each platform’s own attribution: the three models overlap and sum to ~111% of sales, so read the shape, never add the lines. Note Klaviyo’s line sagging from July 2026: that is the flow outage, not seasonality.

Meta: spend and ROAS, 24 months

37-month totals: £886,743 spend → £2,389,053 purchase value (2.69x blended). The two structural findings: Feb–Mar absorbs ~5x normal spend profitably (Mar 2026: £100,019 at 2.73x), and Q4 was UNDER-spent both years, with ROAS rising through the ramp (Nov–Dec 2025 at 3.45–3.48x), so budget, not the auction, was the constraint.

Google: spend and ROAS, 24 months

Ran unmanaged all summer without decay: £5.8–7.1k/mo at 4.3–4.6x. Last 12 months £121,742 → £525,958 (4.32x blended). The blend hides the split below. Dec 2025 out-spent Nov into the shipping cutoff (£29.4k @ 3.36x vs £14.3k @ 4.81x); load November this year.

Email: 23.5% of revenue, and flows died on 10 July

Last 12 months: £442,315 attributed (flows £288,722 / campaigns £153,592). An agency mass edit on 10 July collapsed flow sends 95–99% while signups (~2,700/mo) and campaigns continued: the welcome flow went from 6,754 emails in June to 159 in August. Jul–Aug 2026 are the two worst attribution months in 24. Flows do £29–38k/mo in Q4: fixing this is the highest-leverage action in the audit.

Architecture is otherwise unusually complete (103 flows incl. seasonal variants; cart/browse/post-purchase/winback/back-in-stock all exist). Genuinely missing: a purchaser cross-sell flow (37k purchaser list unused) and an occasion-anniversary winback, which matters because the cohort curve shows a month-12 anniversary echo: last Q4’s buyers are this Q4’s best reactivation list. Cadence gaps: only 9–10 sends in March (the biggest month) and Dec 2025 halved sends vs 2024 and halved revenue.

Every offer type, scored

286 email campaigns of 3,000+ recipients over 24 months, £278,924 attributed. Each send is tagged by the offer mechanic in its subject line, then measured two ways: revenue per recipient (does it create orders) and AOV (does it create bigger ones). Baseline: RPR £0.0302, AOV £34.65. A send can carry more than one tag, so buckets overlap; read each against the baseline, not against each other.

The read: discounting is not what works here, deadlines and new characters are. Launches carrying no discount at all are the single biggest earner in the account (£64,296 across 53 sends), and Black Friday wins on structure rather than on 20% being generous. Everything in the bottom-left quadrant is consuming send slots and margin for a below-baseline return.

The same offer changes value with the season, because the buyer changes

Revenue per recipient against the 24-month baseline, split by trading season. Blue beats baseline, red misses it. Blank cells had fewer than three sends. The top row is each season's own gifting occasion sent undiscounted (Christmas, Mother's Day, Father's Day, Valentine's). This grid is the argument for building an offer calendar rather than a standing discount policy.

Read it with the persona work. Summer buyers are 85% women and 58% aged 55+, largely buying for themselves and adding to a collection, so deep discount is the one season where it earns (+30%). Q4 is 31% men and doubles the 25 to 34 band, one-shot gift buyers who respond to a deadline and to newness, so launches (+78%) and restocks (+72%) beat everything. Mother's Day is 82% women and 81% under 55 buying for a mother, and it works essentially undiscounted (+43%): they are buying anyway, and discount is dead weight. In the shoulder months no offer clears the baseline by much, which is a demand problem, not an offer problem.

What has been run since 2023, for reference

The biggest acquisition season produces the least valuable customers

Every identified customer's first order over 12 months, classified by the occasion of the product they bought, then followed for repeat purchase. This is the persona work landing in the P&L.

Halloween and Christmas cohorts repeat at 17.3% and 15.2%. The Mother's Day cohort repeats at 6.6%, and Christmas buyers arrive with a first-order AOV of £45.00 against £29.23 for Mother's Day. The Daughter Buying for Mum persona buys one £19.99 animal, once, in a twelve-day window, for somebody else. The Accidental Collector and the Christmas List-Maker buy more items, more often, at higher value.

March acquires the most customers and the worst ones

Cohort size against the share of that cohort buying again within 90 days. Two panels rather than two axes, so the scales stay honest.

March 2026 brought in 11,265 new customers, the largest cohort of the year, repeating at 4.60%, the lowest rate measured. September and October cohorts are a quarter of the size and repeat at over 12%. This does not argue for spending less in March: the Mother's Day window is profitable on the first order and it is the single biggest revenue event of the year. It argues that March's job is acquisition and Q4's job is value, and that the March cohort needs an occasion-anniversary reactivation eleven months later rather than a generic lapse winback.

The free-shipping threshold sits above the ceiling of a natural basket

Free UK shipping starts at £45 (verified live). 38 of the 41 bestsellers are £19.99, so two animals reach £39.98. The customer cannot get to £45 by doing the obvious thing, so they pay the £3.99 either way and there is no reason to add the second item.

The distribution shows a wall, not a pull: 7.1% of orders sit in the £35 to £40 band, then it collapses to 1.9% in £40 to £45. Basket economics confirm the shape: one item £25, two items £39, three or more £68. Only 19.1% of orders contain two or more non-add-on products.

Zipify OCU is installed, firing, and running the same offer all year

Measured directly from order tags across 52,978 orders, not from the app. Two placements are live: a post-purchase offer (748 orders) and a thank-you page offer (239 orders). Both have run the same two heart-themed products every month for twelve months, through Halloween, Christmas, Mother's Day and summer.

Attach collapses exactly when volume peaks. February 1.28% and March 1.36% against January 2.99% and July 2.42%. March alone is 12,321 orders at close to the year's worst attach rate. A Mother's Day buyer who has just bought Heart and Flowers Sheep is being offered a Happy Heart Balloon Mouse, which is a substitute for what she just bought rather than a complement to it.

Low retention is not the binding problem here, and the reason is margin

Twelve months to Aug 2026, KC net DTC basis. Blended margin 85.4% on Olly's measured product costs, shipping at £3.96 per order, payment fees ~1.9%.

What kills a low-AOV, low-retention business is CAC exceeding first-order contribution, so you are forced to retain in order to recover acquisition cost. SHF does not have that problem. An 85% gross margin on a handmade £19.99 product at an £11 CAC is an unusual position, and it is why 13.9% lifetime repeat has not hurt them. Retention here is upside, not survival.

When demand actually arrives: weekly, Oct–Dec 2025

October is NOT part of the ramp: all five weeks sit at the September baseline (~£19–20k). The ramp is an abrupt 2x step in the week of 10 November, 4.5x in BFCM week, a ~3.8x plateau to mid-December, then dead after the ~20 Dec shipping cutoff.

Spend ramp: 2025 actual vs 2026 recommended

2026 dashboard targets: Sept £84,928 · Oct £131,500 · Nov £252,000 · Dec £255,800. The reshape: move money from December (past-cutoff waste in 2025) into November (where Meta’s ROAS rose with spend). Scale in steps and judge the MARGINAL pound: Feb–Mar 2026’s incremental spend returned only ~1.3x.

Priority actions