REAL WORK. VISIBLE CHANGE.
From product impressions and conversion to content production efficiency, see what changed and how we connect analysis, planning and execution.
Food · Product traffic and conversion
Connect script planning, storyboards, video generation and publishing review around the reasons to buy. Track impressions, clicks and conversions separately to connect creative execution with business results.
The same product is compared over the weeks before and after publication, using existing case materials. These changes cannot be attributed solely to AI.
Food marketing must answer two questions: can the content attract the right audience, and can the product presentation encourage a purchase? The project developed scripts and supporting visuals around the product, then assembled, reviewed and published them. Click performance and sales performance were reviewed separately.
Input · Product information, reference content and use cases
Identify audiences, selling points and supporting shots. Place product features in specific customer situations and define the core reason the script needs to communicate.
Deliver script directions with supporting visual references.
Input · Confirmed reasons to buy and script structure
Assemble product scripts, generate storyboard images and video shots, then add voiceovers, subtitles and editing for team review.
Deliver video assets ready for publishing tests.
Input · Product traffic records before and after publishing
Compare impressions, click-through rates and conversion rates separately to check whether assets attract attention and support purchases.
Inform future content selection and variant testing.
BEFORE31,500
AFTER109,400 · Relative growth of 247.3%
BEFORE5.24%
AFTER6.61% · Up 1.37 percentage points
BEFORE2.32%
AFTER7.80% · Up 5.48 percentage points
Images from the food case. The three metrics above compare the same product before and after publishing.
Client feedback · Operations and AI creative production
Connect reference analysis, script assembly, storyboards and production tasks. Reuse proven structures for different audiences and scenarios, reducing repeated planning and handoffs for each asset.
Calculated from daily team output reported by the client. This is team output; the source does not provide a complete measurement period.
Video production involves analysis, writing, storyboards, generation and editing. Operations staff work with an AI creative producer to carry buying reasons and visual tasks through the workflow, expand a master script across audiences and scenarios, and review generated results.
Input · Reference videos and product information
Identify audiences, scenarios, selling points, openings and supporting visuals. Decide which structures to retain and which details to replace with product-specific facts.
Give future scripts reusable structures and reliable product facts.
Input · Approved scripts and different customer situations
Keep the core reasons to buy, adapt copy for audiences and scenarios, then generate storyboards and shot-by-shot production tasks.
Expand one planning effort into a batch of reviewable production plans.
Input · Scripts, storyboard images and shot tasks
Generate the required shots and connect voiceover, subtitles, editing and review. Operations staff check finished videos before publishing.
Create a collaborative workflow from creative input to finished video.
BEFOREPlan each video and collect references individually
AFTERReuse a master structure for each product, audience and scenario
BEFORESeparate handoffs for scripts, prompts and editing
AFTERConnected script and storyboard tasks, with shared team review
BEFORE30 videos
AFTER400 videos · Client feedback
Production records from the existing case. Click to view the full images.
Niu Qian's team · Food / FMCG short-form video commerce
Build a collaborative workflow from reference content to product scripts through asset analysis, product analysis and script assembly. Make proven structures searchable, product benefits matchable and production rules reusable.
Input · Reference videos and approved examples
AI breaks content down into frameworks, characters, scenarios and stories. The team reviews it before adding it to the library.
Future creators can search for and reuse proven content.
Input · Product facts, primary benefits and conditions of use
Identify purchase needs and supporting visuals, linking products with audience and scenario assets.
Ground scripts in product facts and give asset selection a clear rationale.
Input · Product briefs, asset libraries and framework rules
Retain reasons to buy, combine segments for the product and scenario, and explain their order, purpose and connections.
Creative producers review the scripts and feed revisions back into the rules.
Explore six sales scenarios covering invitations, price objections, samples and quotes.
124 reference images with search, favorites and original-image viewing.
