Lifecycle & retention
Email, push, SMS and in-app journeys timed around real behaviour. New users get activated, quiet users get a reason to return, and nobody gets over-messaged.
CleverTap · Journeys · SegmentationAvailable for CRM & lifecycle roles
I am Jericho, a CRM &Lifecycle Specialist.I win back the customers you already paid for.
Your churn report only shows the people who cancelled. I find the ones who went quiet, build the journeys that win them back, and prove the lift against a hold-out. At Maya that was worth ₱219M.
Fintech and B2B · SQL and CleverTap in my own hands · Remote or Budapest
Trusted with growth at
- 33 Bondi
- Maya Business
- Obodo
- Wollongong Granny Flats
- D4 Bioventures
- Game Haven
- Digital District
- Aurore Maison
- 33 Bondi
- Maya Business
- Obodo
- Wollongong Granny Flats
- D4 Bioventures
- Game Haven
- Digital District
- Aurore Maison
Selected results
₱219Mabout €3.5M in sales value recovered at Maya after I led a supplier migration
245dormant borrowers reactivated, worth about €310K in new lending
−48%lower cost per lead for an Australian furniture brand, in four months
33Kmerchant community brought back to life, +27% in sales value
The quiet leak
Nobody sends a goodbye email
Churn rarely announces itself. It gets quieter, one skipped month at a time, until the number in your dashboard finally moves and it is already too late to be cheap.
01
They stop opening
Your emails still report as delivered. They simply stopped landing with anyone, and nothing in your reporting flags that as a problem.
02
They stop buying
One quiet month looks like noise. Three looks like a trend. By then the habit is broken, and habits are expensive to rebuild.
03
They never actually leave
They never cancel, so they never appear in a churn report. They sit in your database looking like customers while spending nothing.
Winning back a customer you already have costs a fraction of finding a new one. That gap is where I work.
See the methodCore strengths
Lifecycle and data are where I do my best work.
I find the behavioural signal, build the journey, and measure whether it created incremental value. The wider funnel skills help me move faster when the problem crosses channels.
Analytics & cohorts
I write the SQL and build the cohorts myself in Databricks. No waiting in a data queue, and segments based on what people actually did rather than what we assume.
SQL · Databricks · Tableau · GA4Performance marketing Supporting capability
Google, Meta and TikTok campaigns restructured until each lead costs measurably less. Same budget, better return, reviewed every month.
Google Ads · Meta · TikTokConversion & web Supporting capability
Traffic is easy to buy and easy to waste. I rebuild landing pages and funnels so the clicks you pay for turn into real enquiries.
Framer · Wix · WordPressAutomation & AI Supporting capability
Chat flows, auto replies and reporting that runs itself, so the team spends its hours on decisions instead of admin.
ManyChat · Mailchimp · AI toolsProof & measurement
Every campaign runs against a hold-out group, so we know the real incremental lift. You get a business case, not a vanity dashboard.
Hold-outs · A/B tests · ROI- SQL
- Databricks
- Tableau
- CleverTap
- GA4
- Looker
- Google Ads
- Meta Ads
- TikTok Ads
- ManyChat
- Mailchimp
- Framer
- Wix
- WordPress
- Photoshop
- JIRA
The lifecycle
Six stages where revenue leaks. I work all six.
Most CRM hires own one or two of these and hand the rest over. Owning the whole line is what lets me fix a problem at the stage where it actually started.
Acquire
Paid media that brings in people who are likely to stay, not just cheap sign-ups.
Google, Meta, TikTokOnboard
The first session decides retention. Welcome journeys that get people to their first real action.
In-app, push, emailActivate
Turning a trial or a first purchase into a habit before the novelty wears off.
Triggered journeysRetain
Watching the cohort curve, catching drop-off early, giving people a reason to come back this week.
Cohorts, SQLWin back
The quiet ones. Segmented, capped incentives, measured against a hold-out.
CleverTap, offersReport
Incremental lift, revenue per cohort, and the business case for doing it again.
Tableau, Looker, GA4- The first week decides the yearMost churn is already baked in by day seven. Onboarding journeys are the cheapest retention you will ever buy.
- Flat beats steepYou cannot stop the decay, only flatten it. A few points at day 30 compounds into a different revenue line by day 90.
- Quiet is not goneThe users sitting between day 30 and day 90 are the cheapest revenue in the business, and almost nobody messages them.
Process
The loop I run, in the order I run it
Same six steps whether the goal is onboarding, retention or win-back. Below: the actual journey, the actual query, and the numbers they moved.
- 01
Find the money that is already yours
Not a blast to everyone. At Maya it was borrowers with a clean repayment record who had gone quiet for 90 days: proven buyers, zero acquisition cost, sitting idle.
- 02
Make it survive a Finance review
I model the upside, cap the incentive so it pays for itself on fees alone, and bring the numbers to Product and Finance. Campaigns that are not funded do not happen.
- 03
Design for the inbox, not the dashboard
Four messages, not fourteen. In-app, push, SMS then email, with buyers exiting the moment they convert and a hold-out group deliberately left untouched.
- 04
Write the query myself
Repayment history, recency, consent, all filtered in SQL against real behaviour. No ticket, no queue, and no segment I cannot defend in a meeting.
- 05
Ship it properly
Into CleverTap with the conversion goal wired, frequency caps and quiet hours set, and two versions of the offer running against each other from day one.
- 06
Prove the lift was real
Hold-out versus treated, so the number survives scrutiny. Then I change one thing, timing, channel or offer, and run it again.
A quiet borrower, back inside two weeks
- EntryQuiet 90+ daysPaid on time, two or more past loans, consented
- Day 0In-app + pushYour 30% cashback is waiting
- Day 3SMSShort nudge with a real deadline
- Day 7EmailProof, plus exactly how to claim it
- ConvertsTakes a loanExits into the retention track
No action by day 14 means one final offer, then suppression. Respecting the inbox is part of the job.
The cohort, straight from Databricks
-- Quiet but valuable Maya Advance borrowers
WITH last_loan AS (
SELECT customer_id,
MAX(loan_date) AS last_loan_date,
COUNT(*) AS lifetime_loans,
SUM(principal_amount) AS lifetime_gtv,
AVG(days_past_due) AS avg_dpd
FROM fact_maya_advance_loans
WHERE status = 'PAID'
GROUP BY customer_id
)
SELECT c.customer_id, c.msisdn, c.email, l.lifetime_gtv,
DATEDIFF(CURRENT_DATE, l.last_loan_date) AS days_quiet
FROM last_loan l
JOIN dim_customer c USING (customer_id)
WHERE DATEDIFF(CURRENT_DATE, l.last_loan_date) BETWEEN 90 AND 365
AND l.avg_dpd <= 7
AND l.lifetime_loans >= 2
AND c.marketing_consent = TRUE
ORDER BY l.lifetime_gtv DESC;
Same budget, working harder
Leads per month 104 to 144
Cost per lead $168 to $111
Cost per lead, furniture client $436 to $227
AfterBefore
Results · Maya Business · 2024 to 2025
Four problems in the data, four numbers that moved
Maya is a fintech app used by millions of small businesses across the Philippines. I ran nationwide campaigns there as Campaign Associate.
The customers everyone had forgotten
245 borrowers came back, worth ₱19.18M in loans and ₱1.26M in fees.
- Problem
- Borrowers with perfect repayment records had gone silent for months. Not churned, not contacted, just quietly parked.
- Action
- Pulled the cohort in SQL, then built a short CleverTap journey around a capped 30% cashback, with a hold-out group to keep the result honest.
- Result
- 245 came back and borrowed again. The full walkthrough sits in the process section above.
245Reactivated
₱19.18MNew loans
₱1.26MFee revenue
When the leak is not marketing, say so
An ₱82M loss became ₱219M once the supply behind the shop actually worked.
- Problem
- Products kept failing to load and payments dropped out. Maya Shop was quietly losing ₱82M in sales value and merchant trust with it.
- Action
- Traced it with Product and Engineering to the supplier rather than the app, built the case for a switch, then led the migration with no downtime.
- Result
- ₱219M in sales value, about €3.5M, 98% of products staying live, and merchants transacting again. I include this one because knowing when the answer is not a campaign matters as much as running one.
₱219MSales value
98%Uptime
₱82MLoss reversed
A dead community of 33,500 people
Entering only counted if you sold something. Active merchants rose 35% and sales value followed.
- Problem
- Between promos the merchant group went silent and sales value dipped with it. Posting more would not have fixed it.
- Action
- Ran weekly games where the only way to enter was to make a real transaction, so engagement paid for itself.
- Result
- 35% more merchants selling, 27% more sales value, and 250 to 300 entries every week with no extra spend.
+35%Active merchants
+27%Sales value
33.5KMembers revived
Turning a one-off into a routine
One small reward, modelled with Finance, drove ₱467M in everyday transactions.
- Problem
- Top-ups and bill payments had flattened out. Merchants tried the feature once and forgot about it.
- Action
- Built a tiered earn-an-extra-₱12 reward and checked the margin with Finance so growth never cost more than it returned.
- Result
- ₱467M in sales value, up 28%, with a third more merchants transacting regularly.
₱467MSales value
+28%Growth
+32%Active merchants
Lifecycle teardowns · Wise and Figma
I tore down two companies nobody asked me to
One transactional fintech, one B2B SaaS. In both cases the published numbers point at a retention problem hiding inside a growth number, and in both cases I wrote down what I would test before anyone asked me to.
Each one names where the revenue leaks, gives three journeys in priority order, includes the SQL to find the segment, and states what would prove me wrong. Outside-in, no internal data, hypotheses labelled as hypotheses.
- Three testable hypotheses, in priority order
- A cadence-aware reactivation query you can read
- The measurement design, including guardrails
- What would prove me wrong within a week
Written unprompted, and the closest thing to watching me work before you hire me.
Client work · 33 Bondi · 2025 to 2026
Given a client list and a budget, here is what changed
At 33 Bondi, a technology consultancy in Sydney, I ran marketing and websites across their Australian client base.
Wollongong Granny FlatsHome building
A third cheaper, every single lead
Rebuilt Search and Performance Max around the keywords that actually convert, moved a $16K monthly budget behind them, fixed the site layout, then pushed a blog across YouTube, Facebook, Instagram and email.
+38%More leads
−34%Cost per lead
−36%PMax cost
ObodoFurniture
Twice the enquiries, half the cost
Rebuilt the Google Ads account around high value trade buyers, then layered Facebook, Instagram and email behind it. Volume doubled while the cost of each enquiry nearly halved.
+80%Conversions
−48%Cost per lead
D4 BioventuresLife sciences
A site serious enough for the science
This firm turns medical research into funded companies, but the website did not say so. Rebuilt it in Wix around their four focus areas, with content aimed at universities and founders.
WixRebuild
4Focus areas
Earlier work · 2020 to 2024
Where I learned to ship without waiting for anyone
Four years of freelance work across shops, brands and the arts. It is not the CRM work, and I am not presenting it as such. It is why a lifecycle campaign of mine never waits on a designer, a copywriter or an ads specialist.
Game Haven
Built a crowd, then sold to it
Content, creative and paid ads for a game shop, with a chat bot answering buyers around the clock.
292KReached
597KViews
8,992Clicks
Digital District
Posts people acted on
Content, ad strategy and automation for a digital products brand. Multi photo posts did most of the work.
+183%Interactions
+17%Reach
Nanowires
Paid and organic, together
Visuals and ads for a software and game store, with bundles that gave people a reason to act now.
42KReached
76KViews
976Clicks
Effluve Santé
An opening that felt like something
Copy and key visuals for an aromatherapy launch, written to make people feel first and act second.
LaunchOpening promo
CopyBrand voice
Imusicapella
Filling concert halls
Content and copy for an award winning choir, promoting tours at home and across Europe.
GlobalTour promo
ArtsAudience growth
Aurore Maison
A jam brand worth its price
Full identity for gourmet preserves: flavour names, two label systems and shop ready mock ups.
3Flavours
PremiumPositioning
Design
I do not just brief the creative. I make it.
Social, editorial and campaign design in Canva, Photoshop and Illustrator, which means a campaign never waits on a designer.
Credentials
Engineering degree, marketing job
The analytics side is not self taught by accident. Every certificate links to the live credential.
BS Computer Engineering
San Sebastian College, Recoletos de Cavite. Where the SQL, the data pipelines and the systems thinking come from.
Receipts
Check everything on this page
Anyone can put a big number on a portfolio. These ones leave a paper trail, and here is exactly what I would do in your first month.
The client sites are liveWollongong Granny Flats, Obodo and D4 Bioventures are all online. The layouts and blogs are mine.
Certificates are verifiableGoogle Skillshop, Databricks and HackerRank credentials link to the issuer, not a screenshot.
The query is on this pageReal segmentation logic, not a diagram of one. Read it in the process section and judge for yourself.
Every number has a methodHold-out groups and date ranges, so the lift is incremental rather than a flattering before and after.
What I would do in your first 30 days
Week 1
Audit the leak
Pull the cohort curves and find where users go quiet. You get a one-page map of the drop-off points and what each one is worth.
Week 2
Fix the obvious
Ship the highest-value journey that is missing, usually onboarding or win-back, with a hold-out group from the start.
Week 3
Instrument it
Events, goals and a Tableau dashboard everyone trusts, so retention is a number the team sees weekly rather than quarterly.
Week 4
Report and repeat
Incremental lift versus hold-out, what it earned, and the next two tests ranked by expected value.
FAQ
The objections, answered before you have to ask
You have two years of experience. Why should I take the numbers seriously?
Because they are dated, converted, and measured against hold-out groups, and because you can check the parts that are checkable. Two years is two years: I am not going to claim I have run a department. What I have done is own a lifecycle loop end to end at a fintech with millions of users, and the method transfers even where the scale does not. Treat the numbers as evidence of process, not of seniority.
What kind of role are you looking for?
CRM, lifecycle or retention marketing at a SaaS or product company. The work I am proudest of sits exactly there: finding a segment in the data, building the journey, and proving the revenue. Titles vary, the job is the same.
Can you really write your own SQL, or do you need a data team?
I write it myself. The cohort query in the process section above is mine, pulled from Databricks. It means I can test an idea the same afternoon I have it rather than joining a queue, and I never hand over a segment I cannot explain.
Do you only do CRM, or the wider funnel too?
CRM and lifecycle is where I am strongest, but I have run paid media, rebuilt landing pages and designed the creative. That usually helps: I can fix the ad, the page and the follow up instead of filing a ticket for each.
How do you know a campaign actually worked?
A hold-out group. A random slice of the audience is deliberately left alone so the comparison is real, not just a before and after screenshot. If the lift is not there, I would rather find out and change the offer.
Your experience is fintech and agency. We are B2B SaaS. Does that transfer?
The mechanics do, and they are the hard part. A dormant borrower and a trial user who stopped logging in are the same problem wearing different clothes: find the segment behaviourally, work out what reactivation is worth, build a journey that respects the inbox, prove the lift against a hold-out. What changes in SaaS is the vocabulary and the event model, activation and seats and expansion instead of transfers and GTV, plus a longer feedback loop because contracts renew annually rather than transacting weekly. That is a few weeks of learning your data model, not a career change. The teardown on this site is me doing exactly that translation unprompted.
Have you owned a full CRM stack, or only executed inside one?
Owned the loop, not the procurement. At Maya I built segments in Databricks, orchestrated journeys in CleverTap, wired conversion goals and guardrails, and reported incrementality to Product and Finance. What I have not done is select and implement a CRM platform from scratch, or manage a CRM team. If the role needs either on day one, I would rather you know that now.
How do you work with engineering and data teams?
By arriving with the query already written. Most friction between CRM and engineering comes from vague asks, so I bring the segment definition, the events I need fired, and why. At Maya that meant working through Kyuubi and JIRA with Product and Engineering on the marketplace migration, which is also how I learned to tell the difference between a marketing problem and a plumbing problem.
What would you want in the first week?
Read access to the warehouse, the event taxonomy, the current journey inventory with message volume per user, and the retention curve by cohort. From those four things I can usually tell you where the money is leaking before the end of the week, which is what the 30-day plan above is built on.
Where are you based, and can you work with our timezone?
Budapest, Hungary, so Central European Time. I have worked to Australian hours at 33 Bondi and Philippine hours at Maya, so overlap with the Americas or APAC is familiar rather than theoretical. Open to remote or hybrid, available now, and I usually reply the same day.
Next step
Skip the take-home test. Give me a real problem.
Most CRM interviews end with a hypothetical exercise. Send me a real retention problem instead, and I will show you the segment I would investigate, the journey I would test, and how I would measure the lift. If the thinking is useful, we talk about the role. If not, you have lost one email.
- 01You send the problemWhere people go quiet, and roughly what the product does. Two sentences is plenty.
- 02I reply within a dayThe segment I would pull first, the journey I would build, and how I would measure the lift.
- 03You judge the work, not the CVIf the thinking fits how your team works, we talk. If it does not, no follow-up email from me.
For recruiters
Looking for a CRM, Lifecycle or Retention Marketing role
- Available now
- Budapest · CET
- Remote or hybrid
Get in touch
- jerichocarloat@gmail.com
- +36 70 573 4973
- LinkedIn profile
- References from Maya and 33 Bondi, on request