AI-Powered Campaign Optimization for E-Commerce Mega Sales

Project Background

A leading domestic beauty brand operates across Tmall, Douyin, and JD.com, with over 400 SKUs and an average order value of 80–200 RMB. In past mega sales, the brand relied mainly on manual monitoring and experience-based bid adjustments. The 6-person ad team had to manage 200+ campaigns simultaneously during peak periods. For this year's Double 11, the brand decided to adopt the SUPERTRANGO AI advertising system, aiming to validate whether AI could replace manual efforts to achieve better advertising efficiency in complex e-commerce mega-sale scenarios.

Core Challenges

During mega sales, advertising faces three typical challenges:

Delayed budget allocation. Traffic prices across the three platforms fluctuate sharply. Manual budget adjustments typically take 2–4 hours to respond, while traffic peaks often last only 30–60 minutes. As a result, a large portion of the budget is consumed during inefficient periods.

Accelerated creative fatigue. The Double 11 cycle lasts 20 days. Consumer click-through rates on repeated creatives decay more than 3 times faster than on normal days. Manual screening and replacement of creatives cannot keep up with the decay curve.

Fragmented cross-platform data. Tmall, Douyin, and JD.com use inconsistent data definitions. The team had to manually consolidate three reports every day. By the time a global judgment was formed, the traffic bonus window had already closed.

Solution

SUPERTRANGO built a "dedicated AI advertising engine for mega sales" for the brand. Key actions included:

Real-time budget reallocation. The system monitors traffic prices and conversion rates across all platforms at 5-minute granularity. When the CPA of a traffic placement exceeds the threshold, it automatically shifts budget to better-performing placements. On Double 11 day, the system executed over 1,200 budget adjustments, all without human intervention.

Creative lifecycle management. AI scores each creative in real time based on click-through rate, conversion rate, and engagement rate. When a creative enters its decline phase (click-through rate drops by more than 15% for 2 consecutive hours), the system automatically reduces its bid weight and directs budget to new creative test groups. Throughout the mega sale, the brand launched 180 new creatives, and AI automatically selected 23 high-potential creatives for scaling.

Unified cross-platform decision model. The system maps advertising data from all three platforms into a single conversion attribution model, with "omnichannel GMV" as the sole optimization goal rather than single-platform ROI. This means the system will deliberately sacrifice short-term ROI on one platform to maximize overall GMV.

Execution Highlights

Three days before the mega sale, the team originally planned to manually set budget caps for each platform. After simulation, the SUPERTRANGO system recommended increasing Douyin's budget share from 40% to 55%, reasoning that Douyin's seeding-to-conversion funnel was more efficient during the warm-up period. After the brand adopted the suggestion, Douyin contributed 48% of total GMV in the first 3 days, validating AI's predictive capability.

Between 0:00 and 2:00 AM on Double 11, during Tmall's traffic peak, the system detected an abnormal surge in conversion rate from search traffic driven by a top livestreamer's room. It automatically increased search ad budget by 200% while reducing feed ad spend. This adjustment was completed within 15 minutes, precisely capturing the instant traffic bonus.

Results

Key Takeaways

AI's value lies in "speed," not "judgment." The brand's original advertising strategy logic was effective, but manual execution speed could not match the pace of traffic changes during the mega sale. AI compressed strategy execution from hours to minutes—this was the core driver of GMV growth.

Cross-platform synergy is a growth lever. Single-platform ROI optimization has a ceiling. SUPERTRANGO's decision model, which targets omnichannel GMV, gave the brand the confidence to "take a loss" on one platform in exchange for overall gains. This global perspective is difficult for manual teams to maintain consistently.

Creative production capacity requires upfront preparation. AI screens creatives very efficiently, but only if the brand has prepared a sufficient creative arsenal. During this mega sale, the brand prepared 180 creatives in advance. If creative supply is insufficient, AI's optimization space will be significantly limited. We recommend brands stock at least 3 times their normal creative volume before a mega sale.