TikTok Ads for helicopter tours in Bali: $22,281 ad spend, 11,721 leads at $1.90, ~$154,800 revenue, 595% ROMI

The task

The client runs helicopter tours in Bali: scenic flights over the island, transfers, charters for hotels and agents. Average order is about $470. The main paid channel is Meta Ads, running since 2024 with a steady flow of leads.

What hurt:

• One channel, one risk. All paid traffic came from Meta. Any ad account issue or auction spike hit sales directly.
• The Bali audience is tourists, and they are on TikTok. People decide what to do on the island once they are there, scrolling the feed. We needed a channel that catches a person in the moment and takes them straight into a conversation.
• The sales team lives in messengers. A lead is only worth something if it turns into a chat with a manager within the hour.

Task: build TikTok into a second sales channel with the same economics as Meta: leads into a conversation under $2 and ROMI above 300%.

What we did

• “TikTok Direct conversation” as the objective, not website leads. Since April 2025 most of the budget runs in campaigns where the result is a started chat on TikTok. The person stays on the platform and the manager gets a warm conversation instead of a phone number.
• Lead forms as a second track from December 2025. A separate campaign with instant forms and ad sets per language: Russian, English, Arabic, Korean, plus a “travellers in Bali” segment. A form lead is more expensive but comes with contact details and intent.
• WhatsApp click test. From April to September 2025 we sent traffic straight to WhatsApp. A click cost $0.13, but conversions in this setup were not tracked, and we folded it into Direct.
• Two ad accounts: a standard Indonesian one and an Australian one. The Australian account was opened in January 2026 for a hypothesis: an Australian in Bali stays on their home carrier’s roaming, so they can be reached through Bali geo plus Australian carrier targeting. We tried Bali geo plus a tiny uninhabited patch of Australia, Australia plus Bali geo at the same time, lead forms in 15 languages and a separate Direct campaign for Australians on the island.
• The Australian account was switched off in August 2026. $2,010 spent, 192 leads, 3 payments worth $1,146. None of the hypotheses produced a steady flow of Australians. The only cheap segment was the Indonesian form, 48 leads for $33, but those were locals, not tourists.
• Search and website only as tests. Search and website campaigns got $447 and did not beat Direct economics.

Results

Campaign type	Ad spend	Result	Cost
TikTok Direct conversations	$13,648	10,503 conversations	$1.30
Lead forms	$7,232	1,179 leads	$6.13
WhatsApp clicks	$954	7,130 clicks	$0.13
Search and website	$447	39 conversions	—
Total	$22,281	11,721 leads	$1.90

• 22.6M impressions, 90,200 clicks. The main account runs at $650–1,650 a month and has delivered 600–970 conversations a month since June 2025.
• Best ad set is the very first Direct one: 4,511 conversations at about $1.10 each.

Money (spend → payments → revenue):

	Ad spend	Payments	Revenue	ROMI
2026 (January–September), CRM actuals	$11,281	176	$82,410	630%
2025 (April–December), modelled	$11,001	~165	~$72,400	~558%
Total	$22,281	~341	~$154,800	595%

• 2026 actuals come from the client’s CRM: leads with a TikTok source and the deal value on each. 4,672 leads → 176 payments (3.8%). Average order $468, median $386.
• 2025 is modelled from the ad account’s 2025 conversions (6,520 conversations and 130 forms) and 2026 revenue per conversion of the same type. WhatsApp clicks are excluded from the model: a lower-bound estimate.

Insights

• Direct and lead forms return the same revenue per dollar, in different ways. A conversation costs $1.30 and converts to a payment at 2.6%. A form costs $6.13 and converts at 7.8%. In 2026 both tracks brought about $41,000 each. Choosing between them is about manager workload, not money.
• TikTok did the “second channel” job. 595% ROMI at $1.90 per lead is the same level as Meta for this client. The channel has run for a year and a half without dips.
• The roaming hypothesis did not work. Carrier targeting and geo tricks produced no flow of Australians: $2,010 to test, account closed. The cheap way to test an idea is a separate low-budget account, not experiments in the main one.
• The first ad set is still the best. The simplest combo, “Direct + scenic flight”, has not been beaten in a year and a half: 4,511 conversations at $1.10.