Nobody had run this expo before, and all we had was a deadline
So we didn't run ads. We built the room first. Before a single ticket sold, we worked out exactly who needed to be there — the buyer profiles, what actually moved them, creative built to land with each one. Then the spend: weeks of Meta ads turned that into more than 20,000 ticket buyers, and the hall filled. But a full hall isn't a successful expo — a full hall of the wrong people is just noise. So the last piece was a matchmaking layer that paired attendees with the people worth meeting, and booked over a thousand qualified conversations matched to the exhibitor's and visitor's preference.
A full hall, and a matchmaking floor booked solid.
When we took it on, the event had never run — no first-party data to retarget, no proof it would draw a crowd, and an immovable deadline: doors opened whether the hall was full or not. The brief had two halves — fill the room with real F&B buyers, then make sure those buyers actually did business once inside.
The honest gap: the Meta pixel captured only about half the tickets — the payment-page redirect broke tracking — so much of the campaign was optimised on incomplete data, and despite that, we still managed to sell over 20,000 tickets.
A first-of-its-kind expo, with no history to sell.
This was the first global B2B F&B expo bringing the full Chinese food supply chain — ingredients, packaging, machinery, processed goods — into Indonesia, the world's largest Halal-food consumption market. The demand was real and the supply was real. What was missing was the infrastructure to connect them, for a show with zero track record.
The client arrived with genuine assets: a heavy PR launch (40+ media placements), co-location with two established expos, institutional endorsements, and a strong exhibitor list anchored by Chinese suppliers. But two things were entirely absent — an audience, and a system to convert that audience into business.
- No first-party data. An inaugural event means zero past attendee list, no retargeting seed, no lookalike base. The acquisition engine had to be built on cold audiences alone.
- An immovable deadline. Unlike an evergreen store, every dollar of spend expired on event day. There was no "next month" to recover a slow week.
- A credibility gap. Selling tickets — and meetings — for a show that had never happened, with no photos, no testimonials, no proof.
- No matchmaking system. No qualification flow, no matching logic, no scheduling architecture. A registered visitor list the client didn't know how to convert into committed buyers.
- Uneven exhibitor data. Several exhibitor descriptions didn't reflect what they were actually presenting — mismatches that would break meetings if not caught and corrected pre-event.
Where the expo was at takeover
Acquire the audience, then activate the floor.
Because this was the first time a Chinese-origin B2B F&B expo had been brought to Indonesia, we started with research — how Chinese organisers run these events, how the format works across other Asian markets — then layered in Indonesian buyer behaviour to find the right way to communicate the show locally. From there we identified the ICP, validated the market categorisation, and structured an intentional set of audiences and creative sequences. Pricing design, offer timing and the registration-form journey were all part of the remit. The two workstreams ran in deliberate sequence: acquisition built the list, then matchmaking ran against it.
Act 1 — Acquisition: from zero data to a full hall.
Weeks of Meta · a lean managed budget · Tens of Thousands tickets.
1 Build the engine from zero
With no data to inherit, the first job was a working funnel — pixel events on the ticket flow, audiences built from the F&B buyer thesis, and a creative system designed to test fast rather than perfect slowly. Identity-callout statics ("For F&B Brand Owners / Professionals / Manufacturers") launched alongside broad and lookalike prospecting on Meta.
- Pixel events on the ticket flow to start collecting first-party signal from day one.
- Audiences built from the F&B buyer thesis rather than any inherited list.
- Identity-callout statics ("For F&B Brand Owners / Professionals / Manufacturers") launched alongside broad and lookalike prospecting on Meta.
2 Find the signal
A creative tournament across hook types let the audience tell us what converts. The findings were decisive — and we concentrated spend behind the winners.
| Audience segment | What they were sourcing | Hook angle that landed |
|---|---|---|
| HoReCaHotel · Cafe · Restaurant · Bar | Reliable suppliers and new menu ideas under margin pressure | Identity callout "For F&B Professionals" + everything-under-one-roof sourcing |
| Bakery & DessertSpecialty shops | Niche ingredients and equipment rarely served by general expos | Sub-category callout "Bakery shop owner?" — narrower self-ID, stronger conversion |
| F&B ManufacturingProduction | Machinery, packaging, ingredient suppliers — without flying abroad | "Ga perlu terbang ke China" barrier-removal + "For F&B Manufacturers" |
What the data forced us to change
| Dimension | Before | After |
|---|---|---|
| Video hook | Identity-list pop-ups — ~5% hook rate | Pattern interrupt ("STOP") — 14–30% hook rate, the strongest signal in the account |
| Targeting | One broad ICP video for everyone — 2.0% CTR | Geo-matched variant ("Are you currently in [City]?") — 4.5% CTR on the same creative |
| Price message | Generic "pre-sale price", no urgency | "Priced like a cup of coffee" — a concrete anchor that removed the objection |
2 Find the signal
A creative tournament across hook types let the audience tell us what converts — then we concentrated spend behind the winners.
3 Compress to the deadline
Spend escalated into the D-Day window, where 50%+ of all tickets landed in just five days — even at a 70%+ price premium. The proof was clean: the core buyer was price-inelastic and urgency-driven.
Discovery split confirmed the spine: Instagram ~70%, referrals ~10%, Facebook ~5%, TikTok ~5.0%, others 10%.
Act 2 — Matchmaking: from a registered list to a booked floor.
Weeks of outreach · a 2,000-strong pool worked · Thousands of meetings.
The matchmaking team was sequenced to start once acquisition had built a viable registration pool. The job wasn't to outreach blindly — it was to design a system that could qualify buyers, extract their real sourcing needs, and route that into accurate matches. The outreach itself was the data-collection layer, not just a conversion tool.
1 Build the system before the pipeline
This program started not by sending messages but by designing what information those messages needed to collect. The matching logic required two things from each buyer — what they were sourcing, and enough context to rule out mismatches — extracted through outreach without it feeling like an interrogation. In parallel, every exhibitor description was reviewed against the actual profile and inaccuracies were flagged: a match built on a bad description fails at the meeting itself.
| Dimension | Before | After |
|---|---|---|
| Opening framing | Announcement tone — the program as an event, not a benefit. Low response. | Opens with their registration as a shared fact, then asks about their sourcing need — the buyer is the subject, not the program. Much higher response. |
| Qualification | Closed ("interested in packaging suppliers?") — yes/no, no matching value | Open-ended ("what packaging / raw material / machinery are you comparing?") — product-level specifics that feed matching directly |
| Scheduling ask | Vague ("can you come to the area?") — high drop-off | Specific and confirmable ("we'll schedule you with [category] on [date] at [time]") — a soft yes becomes an accepted slot |
| Follow-up cadence | Single follow-up, no post-booking reminders — attrition before event day | D-3, D-1 and D-day reminders with each buyer's schedule — keeps the commitment live, cuts cold-feet cancellations |
2 Live outreach — where the system got tested
Three channels in priority order — phone first, WhatsApp second, email third — mirroring how Indonesian B2B buyers actually communicate. Every touchpoint also collected the sourcing detail the matching logic needed: open-ended questions, not closed yes/no asks.
- Entrance reception + signage. A receptionist at the main gate directing buyers straight to the matchmaking area.
- Walk-in registration table. Set up inside the area so walk-ins weren't sent back to the entrance.
- Daily exhibitor pre-brief. Every confirmed exhibitor notified of their schedule each morning — fewer absences, less idle time.
- Walk-in booking form. Unscheduled buyers captured on the spot.
- Meeting-overload redistribution. Overloaded slots rebalanced to prevent table shortages at peak hours.
| Event day | Same-day meeting completion | Note |
|---|---|---|
| Day 1 | 10% | Pre-corrections baseline |
| Day 2post-corrections | 30% | Overnight fixes shipped between Day 1 and Day 2 |
| Day 3 | 19% | Sustained above baseline |
| Day 4 | 17% | Sustained above baseline |
3 On-site — managing what outreach couldn't control
The matchmaking area sat in a low-traffic section of the hall, behind a congested, organiser-managed entrance — so many confirmed buyers never found it. The on-site team caught it on Day 1 and shipped five corrections before Day 2.
The jump from 10% to 30% between Day 1 and Day 2 is directly attributable to the overnight corrections — and the buyers who showed ran nearly all their meetings. The structural lesson: the matchmaking floor must be part of the venue-layout conversation before the event, not a Day-1 fire to put out.
Build the system first, then scale what works.
Pattern interrupt beats production value.
"STOP" drove 2–3× the account-average hook rate and the highest CTR (8.9%). For a cold audience, a cognitive break outperforms polish — every new video should open with a challenge, not an identity roll-call.
For events, urgency is the conversion engine.
56.2% of tickets sold in the final five days, even at a 71% premium. The deadline is the most powerful lever you have — architect urgency earlier, don't just ride it at the end.
The outreach is the data-collection layer.
Every matchmaking touchpoint did two jobs: convert the lead, and extract the sourcing detail that made the match useful. Match quality is set before the event, in how the qualification flow is designed.
Exhibitor data quality is your problem, not the organiser's.
Inaccurate descriptions, used as-is, send buyers to the wrong meetings. Pre-event data review is the prerequisite for the matching logic to function — not an optional step.
The matchmaking floor is a physical product.
If a buyer can't find the area, weeks of outreach are irrelevant. Our next-edition standard includes a mandatory pre-event venue walkthrough, signage spec, and table allocation before Day 1.
A booking number is a baseline; show-up rate is the real metric.
We set booking targets that build in booking-to-attendance attrition. A program's value is measured at the table, not in the CRM.