Culture · 2026 Refresh
BMA
operates here.
A small senior team running its own brands, defending its clients' cash, and using AI as leverage, not theatre.
Best Marketing Agency Pte Ltd · UEN 201936773M · Singapore
Why we are rewriting this now
The bar moved. So did we.
Two things shifted at the same time. AI rewrote what one operator can do in a day, and the SG SME market got smarter, stopping payment for activity and starting payment for outcomes. This deck is what we believe now, said plainly, before the next season starts.
01 · The cut
We pruned
Underperformers off. Quiet effort kept polite, we are not a polite company about cash. New hires are AI-fluent operators who can do the work of three.
02 · The lever
AI changed the unit
Claude Code, our skills repo, and async workflows mean one senior operator can carry 25 client accounts without dropping quality. The headcount math changed.
03 · The market
Marcus got smarter
SG SME founders have been burned by 2–3 prior agencies. They no longer buy reports. They buy revenue. So we sell revenue.
Who we built BMA for
Marcus, the founder we serve.
Every value, every hire, every line of copy in this deck is filtered through one person. If Marcus would not respect it, it does not ship.
The profile
- Singapore SME founder
- $600K – $5M annual revenue
- Burned by 2–3 prior agencies
- Wants customers, not reports
- Skeptical, time-poor, results-only
What he is sick of
- "Engagement was up 23%" decks
- Account managers who quote influencer playbooks
- Agencies that have never run their own ads
- Smiling Zoom calls, no answer on cash
- Singapore-polite hedging on bad news
The Marcus test. Before any deck, email, ad, or report ships, ask: would Marcus, six weeks deep into a contract he is questioning, read this and feel his money is being defended? If no, it does not leave the building.
The promise, on one screen
Performance you can deposit.
Not awards. Not impressions. Not "brand lift." We sell the kind of result that shows up in your bank account in a quarter we can both name out loud.
We measure in
SGD
Not reach. Not engagement. Cash.
We dogfood on
Own brands
SGF · Justswim SG · Justswim ID
We answer in
Hours
Not "we'll get back to you next week."
The line we will not cross. We do not promise rankings, leads, or revenue we cannot personally model from real money we have spent on real campaigns. Hope is not a forecast.
The moat no SG agency can copy
We run the brands we sell to.
Most SG agencies are pure service businesses. They sell what they have read, not what they have run. BMA is structurally different, the founder is a shareholder of operating brands that buy media every day, on the same channels we sell to clients. That is not marketing. That is the moat.
Singapore Florist (SGF)
D2C floral brand. Mother's Day, Valentine's, peak-season ad ops. Every PSG Search/Authority/Performance package has been pressure-tested here first, on real spend, before any client paid for it.
Justswim Singapore
Swim school. Lead-gen funnels, local SEO, parent-buyer journey. We have personally A/B-tested 30+ landing page variants on real Singaporean parents, and lost real money on the bad ones.
Justswim Indonesia
Cross-border launch and ops. Different market, different conversion economics, different ad platform behaviour. Same playbook tested in a second jurisdiction before we recommend it.
What this buys the client. When Marcus asks "have you actually run this on a brand like mine?",the honest answer is yes, with our own money, this quarter. No competitor can say that without lying.
Old BMA → New BMA
What we retired this year.
A lot of what passed for "agency culture" was performance theatre dressed in mission/vision/values. We binned the artefacts that did not earn their keep.
Retired
- Mission / vision / values triptych,three slides nobody reads
- BEST acronym,forced backronym from 2019
- ROAD framework,slide-deck filler
- Daily sync recital,45-min status meetings
- Junior coordinator hires,overhead without leverage
- "Engagement" reporting,vanity metric theatre
Kept and sharpened
- 5 plain-English values,each rejects an antipattern
- One promise,performance you can deposit
- 15-min standup,M / W / F only, async otherwise
- Senior AI operator hires,leverage layer
- Asana board as the visual,work tells the story
- Cash-tied weekly close,Friday demo of real outcomes
Value 01 of 05
Value · 01Skin in the game.
We do not recommend tactics we have not lost real money on first.
What we mean
- Jim is a shareholder of Singapore Florist (SGF)
- Jim is a shareholder of Justswim SG + Justswim Indonesia
- BMA dogfoods every service on these brands first
- We have personally lost real money on bad ads
In practice
- Every PSG package is pressure-tested on our own brands before any client pays for it
- When a tactic fails, we eat the cost, not the client
- Operators are encouraged to run their own micro-brands too
- Insights flow from our own ad spend, not other people's case studies
Rejects
Agencies that have never run their own ads, never paid out of their own bank account for a bad campaign, and learn the playbook entirely from clients' wallets. That is not strategy, it is tuition paid by someone else.
Value 02 of 05
Value · 02Real money or real silence.
If you cannot show it in cash, do not show it in the slide.
What we mean
- Every claim ties to SGD
- Reports lead with revenue, CAC, LTV, not impressions
- If the cash story is bad, we say it is bad
- If we are not sure, we run the campaign 14 more days before claiming a win
In practice
- Monthly client review opens with: spend in, revenue out, ROAS, net
- "Engagement up 23%" gets cut from every deck before it ships
- Pixel + GA4 + GHL are reconciled to actual order value, not to platform-reported conversions
- If we cannot reconcile, we say "we cannot reconcile",out loud
Rejects
Vanity metrics, ratio gymnastics, "engagement" theatre, screenshots of reach without revenue context, and the agency favourite, picking the one chart that went up and pretending the rest of the dashboard does not exist.
Value 03 of 05
Value · 03AI multiplies, humans decide.
AI is leverage. The human owns the call and the outcome.
What we mean
- Claude Code runs daily on every operator's machine
- Our private skills repo is the leverage layer, clone day 1
- AI drafts. Operator decides. Operator signs.
- If the model is wrong, the operator is wrong, not the model
In practice
- bma-pitch-deck turns a Fireflies call into a custom PSG deck in 20 min
- bma-seo-scorecard runs a S$2,500 forensic audit, automated
- pr-pitching-campaign drafts and schedules journalist outreach end-to-end
- Every output gets a human read-through before it touches a client
Rejects
"AI agency" theatre, ChatGPT cosplay, AI-as-crutch operators who paste raw model output into client emails, and anyone who blames the model for a bad recommendation. The accountability chain ends at a name, not a prompt.
Value 04 of 05
Value · 04Truth before comfort.
Say the awkward thing on day one, not in week six, when it costs the client money.
What we mean
- Deliver bad news early and direct
- Singapore-polite hedging is a tax on the client's cash
- "Maybe we can try…" is not a recommendation
- If a campaign is failing, we say "this is failing" in the call, not in the email after
In practice
- Kickoff call: we tell Marcus what we will not do for him, before what we will
- If a client's offer is broken, we say so before we run a single ad
- If a teammate's draft is not good enough, the feedback is in the next 30 minutes, not "let's circle back"
- If we are wrong, we say it on the call. We do not redraft history in the report.
Rejects
Sandbagging, false comfort, deferred bad news, "we are still optimising" three months in, and the SG-agency reflex of softening a real problem until the client churns and is shocked we did not warn them. We warned them. In writing. On day one.
Value 05 of 05
Value · 05Defend the client's cash.
When their goals conflict with our quarter, their cash wins.
What we mean
- If a client should pause spend, we say pause
- If a package is wrong for their stage, we downgrade it
- If a renewal does not earn its fee, we do not renew it
- Their P&L is the deciding vote, not our pipeline
In practice
- One PSG client this year: we recommended they not renew the Performance Stack and step back to Authority. They stayed two more years on the smaller package, and trust.
- We will tell a Marcus to spend less on us and more on his offer first, if his offer is the bottleneck
- Scope creep without renegotiation = we rewrite the SoW or we walk
- The 5% renewal commission only triggers when retention is real, never on artificial extensions
Rejects
Renewal-at-all-costs, scope-bloat upsells, sunk-cost rationalisation, "let's give it another month" when the data already said no, and the entire SG-agency template of selling more to a client who needed less.
Operating Principle (the bonus value)
Value · 06One bar, one team.
There is no senior/junior gap on quality. Same standard whether you are GM or designer week one.
What we mean
- The work is judged by the work, not the title
- A junior can correct the GM in a review, and is expected to
- "This is fine for a junior" is a phrase we do not use
- If you would not show it to Marcus, do not show it to us
In practice
- Every operator can clone the same skills repo and ship at senior level
- Friday demo: same review bar for everyone
- Hierarchy decides who owns the call. It does not decide what passes.
- Promotion is recognition, not protection
Who runs this
A small senior team. No coordinator layer.
Nine people. Everyone owns a function. Nobody hides behind a hierarchy. If you join, you sit one Slack-thread away from the GM and one Asana card away from a real client outcome.
Jim Ng GM · Owner
Founder & General Manager. Strategy, sales, executive decisions. Shareholder of SGF and Justswim. The dogfooding is his.
Catheryn Wong Director
Director, Finance & Operations. Invoicing, payroll, vendor relationships, PSG compliance. The cash discipline lives here.
James Yap Strategist · Sales
Inbound deal owner. Pitch + onboarding handoff. Joined May 2026 to run the close motion.
Juzel Albert Account Manager
Owns day-to-day client AM motion across the portfolio. The single point of contact most clients hear from week to week.
Mark Baulo Head of Content
Owns content quality across the portfolio. Vets every scope before it ships.
Rachelle Chong Head of Design
Owns visual systems for key accounts incl. NTU. Sets the design bar across the portfolio.
Lau Chin Yee Designer
Production design across client portfolios. Ships layout, ad creative, and brand systems alongside Rachelle.
Frydda Mae Cruz Designer
Production design across client portfolios. Ships layout, ad creative, and brand systems alongside Rachelle.
Xavier Fok AI Senior Operator
Joined 6 May 2026. Carries the 25-account portfolio. Example A of the new hire model.
The hard line
What we do not tolerate.
Some things end the conversation. Not "we will coach this." Not "let's discuss." End of conversation. Read these once and never test the line.
Fake testimonials · ever
AI-generated reviews, ghost UGC, paid-for star ratings, fabricated case studies. Singapore CCCS treats this as a regulatory matter. We will never put a client on the line for it. Same-day termination, no severance.
Vanity dashboards
Slides that hide cash with engagement charts. If the client cannot deposit it, it is not a result. Reports that lead with reach get returned for rewrite.
Sandbagging
Holding a known-bad campaign result for the next monthly review instead of flagging it the day it broke. The client paid for honest information in real time.
Tribal knowledge
If you are the only person who knows how a client's funnel works, you have created a liability, not value. Document, repo, hand off, or do not start the work.
Singapore-polite hedging
"Maybe we can consider…" when you mean "this is broken." Polite is fine; vague is theft. Be polite and clear, never polite and unclear.
AI without judgement · scope creep · surprise churn
Pasted ChatGPT slop into client emails. Quietly absorbing scope without renegotiating. Letting a client churn on a Friday with no warning across the prior 30 days. All disqualifying.
The brightest line we draw
No fake testimonials. Ever. Full stop.
This gets its own slide because it is the single fastest way to end your service agreement and the single fastest way to put a client in regulatory trouble. Read this once. Internalise it. Never test it.
What is forbidden
- AI-generated reviews on Google, Facebook, Carousell
- Fabricated case studies or named-client quotes
- "Influencer" UGC paid to misrepresent product use
- Fake star ratings, vote-buying, review swaps
- Editing real reviews to remove client-attributable detail without consent
Why we are absolute about it
- Singapore CCCS treats fake testimonials as a regulatory matter
- BMA is a PSG-grant approved vendor,one breach risks the vendor status of the entire firm
- One fake review uncovered = the client's brand, our vendor status, and your role all gone in the same week
- There is no business case strong enough to justify it. None. Ever.
The consequence is plain. Discovery of fake testimonials produced or facilitated by any operator = same-day termination for cause. No notice period. No severance. Documented in the service agreement, signed before day one. We are not flexible on this.
How we hire
The Marcus-fit operator.
We hire fewer people, more senior, with AI fluency and dogfooding instinct. The interview filters for one thing: would this person be respected by Marcus on the first call.
We hire for
- AI-fluent,uses Claude/Cursor/n8n daily, can build workflows
- Dogfooding mindset,has run their own brand, store, or campaign
- Performance-only orientation,talks in cash, not impressions
- Comfortable with truth-telling,can deliver bad news to a founder
- Senior judgement,owns the call, does not escalate every decision
We do not hire for
- Pedigree without output
- "Brand strategists" who have never bought media
- Junior coordinators who need a manager to do work
- Anyone who frames AI as a threat instead of leverage
- Anyone who has built a career on smiling in client meetings
Example A: Xavier Fok joined as AI Marketing Senior Operator on 6 May 2026. AI-fluent, runs his own micro-businesses on the side, comfortable being measured on retention, not activity. The hire model from here forward.
How we fire (said out loud)
Performance gates and ethics gates.
If we say it on the way in, nobody is surprised on the way out. Two clocks run on every operator: a performance clock with KPI gates, and an ethics clock that has no gate at all.
Month 3 · Early signal
First gate
60-day cohort retention ≥ 75% (Standard) / ≥ 50% (PSG). Below = formal review, written remediation plan, second chance.
Month 6 · Conversion gate
Second gate
Retention ≥ 85% Standard / ≥ 50% PSG. NRR ≥ 100%. Below = service agreement ends, no conversion to permanent.
Three strikes
Performance pattern
Three documented misses on quality bar across a 90-day rolling window = exit conversation. Coaching is real, but it has an end date.
Ethics breach
Same day · no severance
Fake testimonials, client data leak, material misrepresentation, hidden conflict of interest. No clock. No notice period. No severance.
This is the deal we make on day one. The KPI numbers are in your service agreement, not buried in a Notion page. You always know where you stand. We do not run a "guess what we are thinking" company.
How we pay · the structure
Base + renewal commission + phantom equity track.
Above-market upside if you operate well. Fair fixed pay if the season is slow. We are not the cheapest agency to work for, we are the most leveraged. The math reflects that.
Base · Senior Operator
$5K+
SGD/month after 6-month service term · CPF on conversion
Renewal commission
5%
of renewed contract value · paid quarterly · uncapped
Phantom equity track
Y2+
opens after Year 1 KPI gates pass · senior operators only
Why renewal commission, not new-business commission. New-business pay rewards selling. Renewal pay rewards keeping the client alive and growing. We want operators incentivised to defend cash, not chase logos. Every $1 you retain = $0.05 in your pocket. Forever.
How we pay · phantom equity
The long-tail upside, gated honestly.
Phantom equity is not a recruiting trinket. It is a multi-year payout that vests against actual portfolio performance. Two operators in the team's 5-year history will have it before 2027 ends. That is intentional.
| Stage | What unlocks | Gate |
|---|
| Months 0–6 | Service agreement, base pay, monthly invoicing | Standard onboarding |
| Month 6 conversion | Permanent role, CPF, 5% renewal commission live | M3 + M6 KPI gates passed |
| Year 1 close | Annual review · NRR ≥ 105% on portfolio · ethics record clean | Eligibility for phantom equity grant |
| Year 2 grant | Phantom equity units allocated against pool · 4-year vest | By GM nomination, board approval |
| Liquidity event | Sale, dividend distribution, or scheduled buyback | Vested units paid in SGD |
Why this matters internally. If you stay senior at BMA for 4+ years and your portfolio retains and grows, the phantom equity payout will outweigh every monthly base cheque you collected on the way there. That is the design.
How we work
Daily cadence. Async-first.
We run a small senior team. We do not run status meetings. The Asana board is the daily sync visual, if your card moves, we see it. If it does not, we ask once, then we look at the gate.
| Day | Format | What gets covered |
|---|
| Mon | 15-min standup · WhatsApp voice or Meet | Week's top 3 client priorities. Blockers. Wins from the weekend. |
| Tue | Async only | Asana cards move, Slack threads close, no calls unless a client asks. |
| Wed | 15-min standup | Mid-week recalibration. Anything off-track gets escalated by EOD. |
| Thu | Async only | Production-heavy day. Designers ship. Operators ship. Quiet. |
| Fri | Weekly close + demo · 30 min | Each operator demos one real outcome. Cash impact, not slide design. |
WhatsApp is primary. Slack is for threads, Asana is for tasks, Drive is for files, Fireflies is for calls. WhatsApp is for "the thing actually happening right now." If your operator can't be reached on WhatsApp inside business hours, that is a problem.
The leverage layer
Claude Code + the BMA skills repo.
Every operator clones the same private skills repo on day one. The repo turns the senior judgement we have already encoded into reproducible workflows. One operator with these skills runs at the throughput of three without them.
bma-pitch-deck
PSG pitch deck generated from a Fireflies call transcript. Outputs Google Slides + parallel .md brief. 20 minutes from call to deck.
bma-seo-scorecard
45-slide forensic SEO + AI Search audit. The hand-built version is valued at S$2,500. Now automated and re-auditable monthly.
bma-hr-onboarding-docs
Letter of Offer + Service Agreement generated from official templates. Plus full access provisioning checklist. Same-day hire-to-onboarded.
pr-pitching-campaign
End-to-end PR pitching. Researches verified journalists, drafts personalised pitches, schedules Gmail send at optimal SGT.
Repo: Best-Marketing-Agency-Main/bma-skills. Cloned on day one. Synced to ~/.claude/skills/. Tested with a dry-run before any live client output. The leverage is not hypothetical, it ships in week one.
The performance bar
Good vs great, the operator delta.
"Good" keeps the client. "Great" compounds the client. The delta between the two is rarely effort, it is one extra signal sent to the founder. Multiply that across 25 accounts and a quarter and you have the difference between a job and a career.
Good
- Ships the monthly report on time
- Closes the deal at quoted price
- Replies to client within 24 hours
- Hits the campaign brief as written
- Renews the contract
Great
- Ships the report + one specific recommendation the client acts on this week
- Closes the deal at quoted price + retains 12+ months
- Replies in hours + pre-empts the question they were going to ask Friday
- Hits the brief + tells you which line of the brief was wrong before you wrote it
- Renews the contract at higher MRR with no scope inflation
Why this company exists
A note from Jim.
I have spent my own money on bad ads. I have run real brands, Singapore Florist, Justswim SG, Justswim ID, through the same channels we sell to clients. Every tactic in the BMA playbook has cost me cash before it ever cost a client cash.
I built BMA because I was tired of two things. SG agencies that talked about strategy without ever buying a media impression of their own. And the Marcus archetype, the SG SME founder, $600K to $5M, three agencies deep, who stopped trusting marketing because nobody had ever shown him a number he could deposit.
This deck is the company I wanted to work with when I was Marcus. Five values. One bar. AI as leverage, not theatre. Real money or real silence. If that sounds like the place you want to operate from, the door is open and the standard is high.
Jim Ng · Founder & GM, Best Marketing Agency Pte Ltd
Two doors out of this deck
Operate at this bar,
or operate elsewhere.
If you are on the team, this is the standard from this season forward. If you are reading as a candidate, apply. The bar is high, the upside is real, and the seat is earned.
For the team
Welcome to the new bar. Read this deck twice. Print the values page. Ask one question this week about a value you are not sure how to live yet. We rewrite this together.
For the candidate
Apply at bestmarketing.com.sg · subject line "Senior Operator · 2026". Tell us one campaign you have personally lost money on, and what it taught you. That is the first filter.
Best Marketing Agency Pte Ltd · UEN 201936773M · Built in Singapore.