How the world’s second largest phone maker grew online ads GMV by 6X

Xiaomi Indonesia
5,800
2.84X

ROAS

6X

Ads GMV

Cost Savings

317

Hours Saved

Share of Search

Items Sold

Xiaomi, the world’s second largest phone maker, has multiple stores on top marketplaces like Tokopedia and Shopee across Southeast Asia. As a leader in the digital space, Xiaomi adopted retail media early -  marketing to consumers at or near the point of purchase, or point of choice between competing brands or products - by activating MyAds and TopAds to reach the millions of shoppers now starting their product discovery journey right on the marketplace. 

But to scale the performance of advertising on Shopee, Tokopedia and Lazada and gain more digital share of shelf, Xiaomi Indonesia Head of Ecommerce, Akshay Singh, was looking for “a better tool to boost GMV and increase our (product) rank in search”. 

Before Epsilo

The ecommerce team at Xiaomi was using the seller tools available on the individual marketplaces themselves in Indonesia to run ads but there were five key obstacles shared with the Epsilo team when the two began their partnership.  

  1. Lack of visibility to keyword performance: no clear way to know exactly the search volume and performance per keyword and keyword category 
  1. Tedious and manual management of keyword ads: thousands of keywords were in the system but it was unclear whether the team was utilizing the right combinations with SKUs to maximise sales
  1. Campaign optimisation: manual maintenance was required to manage back to back campaigns ex. boosting extra keywords to secure traffic (PV) and GMV targets 
  1. Ad budget management: without accurate sales and inventory forecasting, ad budgets and SKUs would run out subsequently leading to a drop in share of search (rank in search results)  
  1. Lack of real time data: it was difficult to optimize performance post campaign as data was fragmented and report exporting was limited to only covering the last 90 days 

After integrating with Epsilo


The Epsilo Client Solutions team set up the Xiaomi account within 48 hours to begin solving the obstacles shared above. 

Solution #1: Keyword Analysis Report and Keyword Discovery Tool

To maximise the return on ad spend (ROAS), the team needed to know which keyword actually drove sales. Epsilo automatically categorized all the keywords and % contribution to sales with the “Keyword Analysis Report” and “Keyword Discovery” tool. These features on Epsilo provide insights and performance visibility down to the keyword level such as below:

The purpose of the keyword analysis is to answer questions below that are important to optimise ROAS and reduce wasted ad spend:

  • Should I continue to bid on competitor KWs?  
  • Was my product launch successful? 
  • Is the cost of the bid worth the GMV gained?
  • Do Brand/Generic/Competitor KWs drive better ROAS?
  • Do shoppers only buy my product during campaigns? 

Solution #2 & 3: Mass Upload and Rule Management 

To run keyword ads, a seller must map each SKU to a set of keywords. Imagine your online portfolio contains 50 SKUs and each product has on average 30 possible related search keywords to target a potential shopper. 


That means to set up keyword ads for one period, 150 repetitive actions at minimum need to be taken to map each SKU-KW ad. When preparing for a campaign, there are usually additional promotional SKU bundles (with GWP) and additional search keywords in relation to the specific campaign/sale (ex. Xiaomi sale) if applicable multiplying the time needed to set up a campaign.

Using the “Mass Upload” feature in Epsilo, multiple keywords can be mapped to multiple SKUs (ex. “Mobile” to all SKUs in the Xiaomi portfolio under category: Mobile). Using this tool allowed the team to bid on 3X more keywords (over 490 keywords versus the original 146).


The Xiaomi team also wanted to maintain their top of search ranking for certain products and instead of manually checking and intervening in case their keyword was outbid by a competitor, the team used the “Rule Management” feature on Epsilo to automate the bidding to ensure Xiaomi’s SKUs were always in the top five rank for specific keywords. 

This change resulted in +698% growth in ads GMV as product impressions and clicks grew by 4.4X and 6.1X, respectively, - more visibility on top selling SKUs (aka Hero SKUs). 

Solution #4: Wallet Management and Inventory Alerts 

When ad budget runs out or inventory runs out, there is a loss of potential sales and ad spend is wasted on driving shoppers to SKUs that are are OOS (out of stock). “Alerts” and “Rules” can be set up in the Epsilo platform to send triggers to the seller/marketer when the ad budget or Wallet Balance is running low based on historical spend data (Traction and Wallet Days of Coverage) and when SKUs are at risk of running out of stock based on historical sales data. 

The Xiaomi team saved approximately 317 work hours through rules automating over 5,800 actions*. 

Solution #5: Customisable Data Reports

To analyse post-campaign performance of a product launch, the Xiaomi team was able to collaborate and access one dashboard to assess how ads did across marketplaces. The Epsilo dashboard aggregates KPIs such as Total GMV, Ads GMV, ROAS, Ad Cost, CvR, Items Sold and CTR for different marketplaces in one control center. 

This made it easy for reporting, data extraction and decision making for campaign optimisation in the future. Overall within two months (March and April 2021), Xiaomi saw their marketplace ads GMV contribute 4.5X more than the two months prior. 


Want to understand how Epsilo can help you scale online? Let us know how we can help at our Contact page.

*(SKU x keyword) x times adjust per day x number of day per month

(635) x 1min x 30 days = 19,050 mins / 60 = 317 hours


Other Case Studies

Have questions?

Get in touch with us to understand why sellers and partners trust Epsilo to drive marketing ROI, increase productivity and make better data driven decisions for their ecommerce businesses.

Thank you for registering
Oops! Something went wrong.