'It's an arms race': Carousell spends five-digit sum per month, hires more staff to fight scams
Since she joined Carousell's trust and safety team in December 2020, Ms Jessica Chen, 28, has seen her fair share of fraudulent listings.
These include those advertising Covid-19 essentials like face masks and popular items like concert tickets.
In these cases, after paying for the items on Carousell, buyers do not receive the items they pay for and discover they have been scammed.
The company's reputation suffered when police figures released in 2019 showed 70 per cent of e-commerce scams in 2018 took place on Carousell.
Recognising the need to tackle e-commerce scams, it has been investing in more talent and software to tackle scams.
Ms Chen, a policy and escalation manager at Carousell, said between 2019 and 2021 the platform's trust and safety team doubled its headcount, but declined to reveal numbers.
This enabled them to better investigate reports against users or listings faster, said Ms Chen.
The move paid off when in 2021, Carousell's share of e-commerce scams fell to 37 per cent.
Every month, Carousell also spends a five-digit sum for a subscription to Sift, a fraud prevention software. Used by Ms Chen and her team, it helps them investigate and review fraud cases.
Suspicious accounts are assigned a Sift score, based on parameters such as how new the account is, transaction history and e-mail addresses used.
Those whose scores exceed a certain threshold will be automatically suspended.
Ms Chen said: "We make use of both the automation and manual moderation for review. The main benefit of automation is that it is immediate, reducing the time that bad actors remain at large on the platform."
The most challenging but rewarding part of the job for Ms Chen is keeping up with the constantly evolving nature of scams.
She said: "What I enjoy most is the investigation process, trying to find patterns between scam cases you feel might be connected due to their modus operandi.
"It is an arms race. After the scammers behind these accounts realise you are able to detect them, they will then change their methods and the race begins again."