|
|
Impact Study - AI Stylist and Personalization
Only AI stack with products that power Omni-channel strategy:
AI that can be applied across website, app, social channels, stores, warehouses & more
Maijindou Magento COMMERCE
BETTER DISCOVERY
HIGHER ENGAGEMENT
BETTER CONVERSIONS
HIGHER CLTV
Maijindou Magento’s AI STYLIST - ENSEMBLE GENERATOR
WHAT?
• The ensemble generator acts as a website’s AI-powered stylist
personalized for each shopper to showcase curated looks that they’ll love.
• An cross sell engine combined with the ideas of visual and fashion attributes, personalized for each user. Also provides context for the product being viewed.
WHY?
• Shoppers like to see accessories and other items of clothing
that complement and complete a look, even if it’s not what the model is wearing.
• Shoppers like to see how versatile the product they are viewing really is. When you show them a product page that includes pairing suggestions in different colors – they’ll view more, click more, and buy more.
HOW?
• Maijindou Magento’s Ensemble Generator is built by taking the following parameters into consideration along with the shopping patterns on a website --
• Visual affinity
• Inter-product Correlation
• Price affinity
• Brand affinity
• The algorithm is built and learns to tap the right mix of these levers ensure that the curated looks are seamlessly delivered without any manual
Findings:
• Engagement -- For majority of the engagement, Mondays seem to have the maximum clicks on CP. And, a large
proportion of this comes from the repeat users who have already logged in and have become organic users to the site.
• Buys – Mondays have the maximum buys through CP. This perfectly ties up to the trend of browsing late on Sundays and
people would normally take less than 24 hours to decide and buy the products.
Maijindou Magento’s ENSEMBLE GENERATOR IMPACT ANALYSIS
Maijindou Magento’S ENSEMBLE GENERATOR FOR FASTER DECISION MAKING AND BETTER USER JOURNEYS
Findings:
• Users who engage with CP tend to make decisions faster in order to get to the next step in their journey. Users who engage
with CP take almost only half the time to proceed to the next milestone in their journeys.
• Users who engage with recommendations on the site show favorable behavior, especially those with CP engagement. There
is a very good conversion rate for users who interact with CP alone - 19% of users who click on CP, end up buying the same
product.
• These clearly indicate the relevancy and effectiveness of the personalized CP recommendations.
Maijindou Magento’s ENSEMBLE GENERATOR IMPACT ANALYSIS
A/B TEST RESULTS FOR Maijindou Magento’S ENSEMBLE GENERATOR
TEST CONDITIONS
• We conducted an A/B test for a leading jeans wear
brand, to determine the value of the Maijindou Magento’s Ensemble Generator.
• Half the visitors to the product details pages were
shown our recommendations while the other half were not.
Maijindou Magento’s ENSEMBLE GENERATOR
IMPROVEMENTS AND NEXT STEPS IN PIPELINE
Next steps / Recommendations:
1. Product optimization based on entry point (Referrer):
• Split traffic based on referrer types –
o Ad/Search based retargeting
o In-house Social media platforms (like Facebook, Instagram, Pinterest)
o Email marketing / retargeting
• In the list above, ways 2 & 3 are the cheapest ways to bring in more users / repeat users back to the website.
• For social media, CP could be combined with Maijindou Magento’s SHOP THE LOOK to make it more valuable.
2. Product optimization based on day of week:
• Since most of the users browse on Sundays and buy on Mondays, the Cross Product based Email campaigns can be
targeted for these days for maximum conversion and better AOV and AOS through Cross Product.
3. Product optimization based on design / location of widget:
• Maijindou Magento’s Ensemble Generator works as a great tool to provide context and this has been established by clients who
have had the widget on their product description pages.
4. Product optimization based on visual and fashion preferences:
• Maijindou Magento’s Ensemble Generator is next going to take in style, occasion and color matching choices as part of its
algorithm.
|
|
发表于 2019-10-11 14:54:00
