Seatfrog board · Wednesday 23 September 2026
Marketing update.
Sheran Gale, Head of Growth Marketing
Where we started, and what changed
Fragmented when I joined in March, and still fragmented for the customer
In Q3 we built the capability to be efficient and effective; it ships in Q4
Agentic, the AI one-to-one decisioning engine, follows in Q1 2027
Fragmented: the comms map of 13 August 2026 and Marketing’s record. The values are the Q4 2026 Marketing Model (20 Sep), net a year, each journey counted under the level of capability it needs; together £279k, and £379k with the acquisition tests.
How the journeys were built
Ten journeys, one method, run in a quarter
- 1Audit every send
- 2Find where customers stop
- 3Hear why
- 4Map every moment
- 5Design the journey for each
One journey in depth · churn prevention, live since 17 September
Today: nothing watches for 49 days, then one calendar email to everyone
The churn-risk model decides who enters, and when, not a calendar
One personalised save the day after the slip, one knock if it went unseen, then silence
The churn-risk model
Sessions, searches, bids and trips, re-scored every night. The band shifts, and the journey opens the next day.
The churn-risk model re-scores every active customer’s behaviour nightly. A shift in their churn band, never a calendar timer, is what opens the journey.
Journey 8, live since 17 September 2026. Measured on the slip-to-return rate behind a 20% holdout; worth £156,265 net a year (£679,413 gross, upgrades only). Copy and timings from the Journey 8 pack; the example traveller is its returning commuter, London Euston to Leeds.
What ships next · Journey 4, turning losing bidders into returning bidders
Today every loser sees the same screen and gets the same email, whoever they are and however they lost
Losing reads as a dead end. 81% of customers quit after one loss, and the one recovery email reaches about 20 of the 14,800 customers who lose an auction each month, because its entry rule is broken.
From November · each loser gets a straight account of the close and one reason to bid again, drawn from their own history: four segments, eight outcomes, five channel states. Click in to play; click outside to move on
What ships when
The churn-prevention journey is live; the rest of the engine ships in November and December and the acquisition tests start in November. Together they drive £64k net in Q4, £43k of it landing inside the quarter
Q4 2026 Marketing Model (20 Sep): net revenue driven by each month’s launches, the journeys and the acquisition tests together; the rest of the £64k phases into 2027 with the launch dates. Journey dates as the strategy page’s journey table carries them.
The marketing funnel
Every sign-up is organic and the engine now holds them; the gap is the middle of the funnel, where consideration is made, and nothing of ours works it yet
- AwarenessNothing paid. Word of mouth, press, our partners’ channels.3 · paid, once a month’s sign-ups clear the line
- ConsiderationOrganic social, unboosted; customers and creators make the content, we do not amplify it.One piece of content, every platform2 · boosted organic social from November, against a holdout
- Conversion29,665 sign-ups a month from the store listing and search; 19.8% buy within 30 days.Q4 tests: search, the store page, the rebuilt website
- Loyalty loopThe lifecycle engine: ten journeys by December, each behind a 20% holdout.1 · live now
Sign-ups and the 30-day rate: the Data for Board Reporting sheet and the Q4 2026 Marketing Model KPIs tab, August 2026. The order is the plan’s: the loyalty loop first, consideration next, paid awareness once a month’s sign-ups clear the line set with Finance and Analytics.
What customers and creators already post about us, unpaid
The social strategy for the consideration layer: partner with the creators who already make this content, pay them by referral, put boosted spend only behind the posts that travel, and read what a boosted view buys against a holdout before we scale it
Our own channels: 2,121 followers on TikTok (281 posts, 900 to 3,000 plays each) and 12,500 on Instagram (456 posts)
- Creators make it, not usA creator programme with a brief, paid in upgrades and per referral; they already share codes unprompted. Many creators, many posts, a small cost each. Not yet sized.
- Only what travels gets moneyEvery post’s first days are read; the winners are boosted, ours and theirs with the creator’s permission (TikTok sells this as Spark Ads). Nothing unproven gets a pound.
- Read it against a holdout, then scaleThe November test boosts the posts that already exist, against a holdout. As sized: £2,500 a month, about 1,000 installs and 700 sign-ups a month, about 140 first-time buyers, £11.5k net a year at run rate; the cost per install is assumed until the test measures it. Then the creator programme scales and the winners go where we sell.
Plays, likes and dates read from TikTok and Instagram on 21 September 2026; the 59 TikToks tagged #seatfrog were read one by one and the biggest unpaid posts chosen. None paid for. The two loops are the deck’s own files; the Instagram post is its cover frame.
The November test as sized buys about 700 sign-ups a month (£2,500 at an assumed £2.50 an install, 70% registering) and about 140 first-time buyers; 10,000 a month is fourteen times that, and the test prices it before anything scales
Appendix
