Seatfrog board · Wednesday 23 September 2026

Marketing update.

Sheran Gale, Head of Growth Marketing

Where we started, and what changed

March 2026, when I joined: one message to everyone across 68 automated sends, and the two biggest campaigns held nobody back to measure against
Efficient today: three specialists hired, every Marketing send on one platform, ten journeys designed; one is live and the rest ship in Q4, a quarter later than the June plan
Effective by the end of Q4, on the churn-risk score and real-time triggers; agentic in Q1 2027 with the AI one-to-one decisioning engine

March 2026when I joined

We are here todayTodaySeptember 2026

End of Q4 2026December

Q1 2027

Customer experience

The message

Channels

People and process

Data and models

FragmentedLaying the foundation

“You don’t know me”

One size fits all: one message to everyone

Email and push, each on its own

Teams operate in silos; no specialist skill set

No model behind a send; 7,074 customers reached, nobody held back

EfficientBuilding capability+£115k net a year, modelled; ships Q4

“You know my stage, and what I just did”

Right message, right group

Marketing’s email, push and in-app as one journey, on one platform (Braze)

Specialist skills hired; a cross-functional team, processes starting to form

Life-stage segments, personas and recency-frequency-value scoring

EffectiveScaling efficiently+£164k net a year, modelled; churn prevention live

“You knew before I did”

Right message, right customer, at the right moment

Real-time orchestration across email, push and in-app

Established process and turnaround times; end-to-end delivery run with AI agents

Churn-risk score live, propensity-to-return score Q4; every journey read against a 20% control group, the holdout

AgenticAutomated, continuously improvingNot yet sized

“My personal travel agent”

Right experience: one conversation across every touchpoint

AI orchestration across every channel

End-to-end agentic process; automated ways of working

Real-time data and machine learning; the AI decisioning engine (needed)

How the journeys were built

61 moments and ten journeys, mapped and designed in-house in one quarter

Agencies sell this map, and the Booking.com version below is one. Ours was built here, so the capability stays in the team.

1Audit every send
2Find where customers stop
3Hear why
4Map every moment
5Design the journey for each

The moment map, all 13 stages of the customer's life — **Ours** · 61 moments across the customer’s life, built in-house in a quarter from the live record.

Booking.com experience map, VML — **The agency version** · the same map, built for Booking.com by VML.

One journey in depth · churn prevention, live from 22 September

Before this journey, a customer could go quiet for 49 days before anything happened, and then it was a timer, not a signal

Nothing at allDays 1 to 49

Nothing watches an active customer for the first 49 days of a slip. No score, no trigger, no message.

The slip is invisible to us while it is still cheap to fix

The same push to everyoneDay 50, on a calendar

09:41

Seatfrog nowGot a trip coming up soon?

A calendar timer, not a signal: it fires whether they slipped or not

The same email to everyoneDay 50, the same day

Today’s day-50 email, the same to everyone

Generic, and the seasonal weekly lands on top of it: two voices in one week

1 in 4 active buyers goes quiet every month, 40,276 in August, and until this journey the slip went unnoticed for 49 days.

Churn prevention, live from 22 September · the churn-risk score re-scores every active customer nightly; a rise opens one personalised save the next day, one reminder only if it went unseen, then nothing until day 60

What ships next · turning losing bidders into returning bidders, November

Today every losing bidder meets the same four things, whoever they are and however they lost

The pushesIn the auction

17:41

Seatfrog nowYou’ve been outbid.

Seatfrog nowJust a few minutes left to bid.

Seatfrog nowBidding has ended.

Three fixed pushes, built into the app before Braze, the same words whatever they bid

The lost screenAt the close

One closed screen, the same for everyone, with no way forward

The recovery email30 minutes after

One message, whoever lost and however they lost

The campaignOn a schedule

Sent to everyone on a calendar, unrelated to the loss

Losing reads as a dead end. 81% of customers do not bid again after one loss*, and the one recovery email reached about 20 of the 14,800 who lose an auction each month, because its entry rule let almost nobody in.

*81%: Marketing’s read from the Journey 4 pack, for Analytics to confirm the population and window. 14,800 a month: Analytics’ figure, Q4 2026 Marketing Model.

From November · every losing bidder gets a straight account of the close and one reason to bid again from their own history; eight ways to lose, four segments, five channel states

What ships when

The first journey is live and carries about £39k of the £64k; five more ship in November and December and three in Q1 2027; sized at benchmark lift rates, the Q4 launches drive £64k net, £43k of it landing inside Q4

Live Ships Waits on a model, a decision or a date

Upgrades only; incremental over a 20% holdout; Q4 2026 Marketing Model, 21 September; lift rates confirmed by Analytics; nothing measured yet.

October

November

December

Q1 2027

ActivateDrive the First Win: turn intent into a first transaction£5.9k net driven in Q4

Turn losing bidders into returning bidders

Prompt the first purchase

Recover the abandoned upgrade

GrowBuild the Habit Loop: turn one win into a habit£4.7k net driven in Q4

Turn the first win into a second

Deliver the First Class experience, and referrals: sizing with Analytics

Recognise returning customers on arrival: needs product work

DefendReactivation: catch the first slip while a save is cheap£39.4k net driven in Q4

Churn prevention: catch the first slip

RecoverReactivation: win back the lapsed£3.5k net driven in Q4

Seasonal reactivation: the never-bought wave, sent

Seasonal reactivation: the buyers’ wave

Win back lapsed customers: offers wait on the propensity-to-return score

AcquireScale Efficient Acquisition: capture the demand that exists, then create it£10.2k net driven in Q4

Search and AI visibility on the current site, 31 October

App store page

Boosted organic social: £2,500 a month, read against a holdout

Rebuilt website, phase 1, 19 December

Partner launch go-to-market, Trip.com first

What the engine needsBuild the Lifecycle Engine: who builds each piece, and what it makes possible

Churn-risk score: live, scores nightly (Analytics)

Propensity-to-return score: sizes the win-back offers (Analytics, Q4)

Live data feed, warehouse to Braze: real-time triggers (Engineering, in scoping)

Losing-bidders app changes: pushes to Braze, a way forward on the lost screen (Engineering, November)

AI one-to-one decisioning engine: a build-or-buy decision in Q4 with Dirk and Analytics; not yet sized

First measured readEvery journey behind a 20% holdout; dates to confirm with Analytics

Churn prevention: save rate vs the holdout

Seasonal wave: purchases vs the holdout

Losing bidders: loss to re-bid within 30 days

Boosted social: cost per sign-up and the 30-day rate

Every Q4 journey read; the next board sees incremental effect

Net revenue driven: £64k in Q4, £43k landing inside itby month of launch; modelled at benchmark lift rates

£13.7k

£20.3k

£29.7k

the rest of the £64k lands in 2027 with the launch dates

Q4 2026 Marketing Model (21 Sep): net revenue driven by each month’s launches, the journeys and acquisition together. Journey dates as the strategy page’s journey table carries them; first-read dates to confirm with Analytics.

The marketing funnel

29,665 sign-ups a month, all organic, and the lifecycle journeys are starting on them; the gap is the middle of the funnel, where people decide to try us, and nothing of ours works it yet

The plan

Awareness

Paid awareness waits for a line we propose and set with Finance and Analytics after the first holdout reads, in November: the share of a month’s sign-ups buying within 90 days, and payback on a bought sign-up inside the window Finance sets.

Consideration

Committed: from November, £2,500 a month behind the posts that already travel, ours and creators’ with permission, read against a holdout Analytics designs. Proposed: a creator programme paid per referral, sized after that read.

Conversion

Search and AI visibility in October, the app store page in November, the rebuilt website on 19 December.

Loyalty loop

Seven journeys live by December and all ten by the end of Q1 2027, each behind a 20% holdout.

59 posts tag us on TikTok alone, none of them paid for by us; the biggest unpaid post about us, a creator’s reel on Instagram, reached 752,000 plays.

752,000 plays, May 2024@holidayexpert on InstagramA creator with 660,000 followers on our Secret Fares with TransPennine Express: “I’ve found another secret train hack”

204,300 plays, May 2025A customer on TikTokA customer’s first-class upgrade, filmed for her followers, with her own Seatfrog referral code in the caption, unprompted

63,500 plays, October 2023A customer on TikTok“The only way I want to travel from now on. It made my trip so much better”

The second source of growth: distribution partners, Trip.com first, sell the upgrade where the ticket is sold. Their economics sit with Iain and Tony; Marketing’s part is the partner launch go-to-market in Q1 2027.

Sign-ups and the 30-day rate: the Data for Board Reporting sheet and the Q4 2026 Marketing Model KPIs tab, August 2026. Plays and dates as shown on TikTok and Instagram on 21 September 2026; none paid for by us. The order is the plan’s: the loyalty loop first, consideration next, paid awareness once the line is set.

Appendix