Back to Blog

Business Growth

Subara: 2.47× ROAS and 214 Registrations on Meta, but Only 23 Paid

2.47× return on ad spend: LKR 46,508 on Meta ads brought 1,143 customer WhatsApp enquiries, 214 registrations and LKR 115,000 in paid registrations for Subara. The more useful story is underneath: only 23 of the 214 paid, and that one step holds the cheapest growth left in the account.

By Joshuwa Salamon Published Nov 3, 2026 Updated Nov 3, 2026 5 min read

A Sky Designers performance case study — Subara, LKR 5,000 matchmaking registrations, Meta ads, 1 to 31 October 2026

Subara spent LKR 46,508 on Meta ads in October, booked LKR 115,000 in paid registrations, and earned a 2.47× return on ad spend.

But the headline number isn't the useful part. 214 people registered and 23 paid. Everything before registration worked at a low cost, and nearly all the unrealised value sits in one step after it. Any Sri Lankan business that takes a sign-up first and the money later can use what that gap shows.

The Numbers

MetricOctober 2026
Ad spendLKR 46,508
Customer WhatsApp enquiries1,143
Cost per customer enquiryLKR 38
Registrations214
Paid registrations23
Cost per paid registrationLKR 2,022
Booked revenueLKR 115,000
Return on ad spend2.47×

Revenue is 23 paid registrations at LKR 5,000. Spend is from Meta Ads Manager, converted at LKR 335 to the US dollar, and includes a hiring campaign. Cost per customer enquiry and the enquiry count leave that campaign out.

Why a registration isn't a sale

StepPeopleRate from the step beforeAd spend per person (LKR)
Customer WhatsApp enquiries1,143n/a38
Registered21418.7%217
Paid2310.7%2,022

Stop counting at registration and October looks excellent: LKR 217 for a sign-up on a LKR 5,000 service. Count to payment and each customer cost about 9 times that. Both numbers are true, but only one of them pays the bills.

If your form, your WhatsApp flow or your CRM has a "registered" stage, report the paid number beside it every time. A sign-up costs the customer nothing, so it can't tell you how many people have decided to buy.

Is a LKR 100 enquiry worse than a LKR 35 one?

CampaignSpend (LKR)Share of spendEnquiriesCost per enquiry (LKR)
Main sales campaign36,68978.9%1,06335
Subscription campaign6,47613.9%65100
Earlier small sales campaign7701.7%1551
Hiring campaign (job enquiries)2,5735.5%3877
Total46,508100%1,53030

The main campaign brought 93.0% of customer enquiries on 78.9% of the spend. An enquiry from the Subscription campaign cost about 2.9 times as much. On that evidence you would move its budget across.

We wouldn't, not yet. Registrations and payments weren't split by campaign, so this data can't say which campaign brought the 23 people who paid. With only 1 in 50 enquiries paying, a costlier enquiry that pays more often could easily be the better buy. Label each chat with the campaign it came from, as we set out in our guide to tracking enquiry-to-booking rate, then decide the budget on cost per paid registration.

Keep hiring ads out of your customer numbers

Subara also ran a hiring campaign. It brought 387 job enquiries at LKR 7 each, the cheapest messages in the account. That's a good result for hiring, and useful to know if you need to fill roles quickly.

It also shows how easily a report can flatter itself. Blend the job enquiries in, and the account's cost per message reads LKR 30 instead of LKR 38 for customer enquiries. Those 387 people were asking about jobs, not the service. Going the other way, the 2.47× return carries the hiring spend. The customer campaigns alone cost LKR 43,935, which works out at 2.62×. Give anything that isn't selling its own line, then read each number for what it is.

Where the cheapest growth sits: the 191 registrations without payment

191 of the 214 registrations had no payment. At LKR 5,000 each that's LKR 955,000, though that is a ceiling and nobody should forecast it. Across the whole funnel, 2.0% of enquiries paid, below the 4.0% to 22.0% we showed in our enquiry-to-booking guide.

Each extra percentage point of registrations paying adds about 2 customers and LKR 10,700. As an illustration, if 20% had paid rather than 10.7%, that's about 20 more paid registrations and LKR 100,000 more revenue on the same ad spend, and the return moves from 2.47× to about 4.6×. No extra ad could do that as cheaply.

This data doesn't say why people stopped, and some who registered late in October may still pay in November. So the first job is to find out. We'd look at three things: when the fee is first mentioned, how easy it is to pay on the spot, and who follows up with a registered person and how soon. Matchmaking is a personal decision and people may want to talk to someone first, so a phone call could do more than a payment reminder. Log the reason each person gives, even in a simple sheet.

The Takeaway

When a sale happens in two steps, cheap traffic only fills the first one. Count to the point where money changes hands, split that count by campaign, and fix the biggest drop before you add budget. If you want a target to work towards, our method for setting a budget from a sales target uses these same rates. More on how we run these campaigns on our performance advertising page.


Sky Designers is a Colombo-based digital marketing agency specialising in performance advertising across Meta, TikTok, and Google. Getting plenty of registrations but not enough payments? Get in touch.