Algorithm based personalization - IRD
Integration requirement document
Congratulations for taking the red pill, the increase in your conversion rates is just round the corner and thank you for giving us the opportunity to work with you – we really appreciate it opportunity and would make the most out of it. ❤️
The Smartech Magic 🔮
1/ Smartech’s recommendation algorithms predict what product each customer is likely to buy next, with high accuracy
2/ Powered with this knowledge, it is then able to show the most relevant and personalized products across every touch-point
3/ Smartech does this with a super-quick integration from the client-side
Integration 🔧
There are two key parts to this integration:
There are two parts to this integration
- Smartech getting the user, interaction and product data for the AI training
- Smartech publishing the most relevant and personalized products at the required real-estate:
- Home page
- Product listing page
- Product display page
- Personal boutique
- Exit-intent popup
- Personalized search (Beta)
- Personalized marketing journeys
- Personalized emails in the journeys
- Personalized app notifications in the journeys
- Personalized browser notifications in the journeys
- Personalized ads for acquisition (Beta)
What we require from you
- Implement the JavaScript in the header of your website
<script src='//cdnt.netcoresmartech.com/smartechclient.js'></script>
<script>
smartech('create', 'VALUE HERE');
smartech('register', 'VALUE HERE');
smartech('identify', '');
</script>
Customer ID details: Please put the unique customer ID in the ‘identify’ quotes. For non-signed in user, you can keep it empty.
Example:
smartech(‘identify’, ‘[email protected]’); when George is signed in
smartech(‘identify’, ‘’); when George is not signed in
This enables us to track the user interaction and render the product widgets
- Implement the JavaScript on your 'thank you page'
var BOXX_PURCHASED_PRODUCT_IDS=["Insert product IDs here"]
Example code:
<script>
var BOXX_PURCHASED_PRODUCT_IDS=[19020336, 19030400, 191001201]
</script>
Why ?
This enables us to track the purchases, and map the corresponding purchased products to each use
- Test Products or free products or 100% discount coupon codes
Why ?
This enables us to test the complete integration
What happens if we don’t get this: We would have to buy the products for testing. Alternatively if COD is available, we would have to let you know each time we buy a product on COD so that you don’t ship it. 🤧
- We need two empty category listing pages – one for personal boutique, and one for the liked products. You can name these:
- "Personal boutique" (or anything else that works for you)
- "Liked products" (or anything else that works for you)
What happens if we don’t get this: We would not be able to develop the boutique pages. 🤧
Note:
If you already have a “wishlist” function, and you would like us to integrate your wishlist function to the boutique page, please provide us the function for the same, and we would do the needful.
- Your LIVE product feed
This needs to contain any information that is displayed in the front-end. This usually includes (but needs to be specific to your portal):
- Product link
- Image
- Hover image
- Name
- MRP
- MSP
- Currency
Please note that the values should be exactly the same as what is present in the front-end.
Any information that is used to filter the products by the user in the front-end on the listing pages. This usually includes (but needs to be specific to your portal):
- Category
- Sub-category
- Sub-sub-category
Why ?
This enables us to show the right categories and sub-categories on different listing pages.
Any information that helps in better defining the product for content-based recommendations. This usually includes (but needs to be specific to your portal):
- Color
- Shape
- Size
- Design
- Pattern
- Brand
Why ?
This enables us to show better recommendations, specifically for websites which have slightly lesser traffic, or for long-tail product views.
What format should this data be in one of these 4 formats are work best for us:
- XML format
- Live CSV format
- You provide us your API
- You use our API
Note:
This enables us to show better recommendations, specifically for websites which have slightly lesser traffic, or for long-tail product views.
What happens if we don’t get this: We can scrape your product listing pages, and use that for display. However, that sometimes leads to a few issues:
- We scrape with each user going to the corresponding product in the listing page. This leads to long-tail products not getting scraped. This leads to poorer discovery and a decrease in improvement in results
- Depending on the structure of your website, we may not be able to integrate on your listing page
- Under some specific scenarios, specifically for websites that sell in multiple countries, the integration is less stable and would need a much more thorough user testing from the client-side as well
- The product refresh rate may be low – which means that it may take some time (usually a day) before a product attribute update (like price change, availability change) is reflected in our widgets
- GA read access specifically for the following data: users, page views, ATC, and Purchases.
Note:
This enables us to test the complete integration. We do get the same data from our integration as well, and an independent source to compare the values against is great.
What happens if we don’t get this: We would like to have the same data (users, page views, ATC, and purchases) at a daily level for the last 1 month. We may also keep asking for this data once in a while. 🤧
We are sure you have most of the things mentioned above. Feel free to reach out to use if you have any questions. 🤝
The Process 📍
Preparation for integration:
- Our onboarding POC would work with you to get us the above-mentioned data
- While you get us this data, we would provide you a demo of the UI, and get your approval on locations, widget UI, and nomenclature
Integration process
- This will take 5 working days, during which we work on your portal to personalize it
- During this time, we would typically speak with you 2-3 times for:
- To demo the revised website in the demo mode, and take any feedback
- To show you the changes requested if any
- To get your approval to go live
Can we pre-fix these meetings please?
Go-live process:
- Once we get your approval, we will go live in a staged manner, in one of the two ways:
- By phasing the users for whom we go live
- By phasing the features on which we go live
- Typically, we go live on all discussed features in a day and go live for all users in 2-3 days
Post-live process:
-
After going live, we will speak with you once a week to showcase you the results. The first time maybe 2 weeks after going live.
-
These are the results you should expect:
End of 2 weeks:
- Increased CTR’s
- Walk-through of the same on your dashboard
End of 5 weeks: Increased ATC and Conversions (typically 5%) in either:
- A/B testing model or
- Last click attribution
Let’s get going 🚦
Updated about 4 years ago