1 Intro
"As a virtual assistant, Kit interacts with business owners through messages over various interfaces including Shopify Ping and SMS."
Kit serves as a nice UI to make ads and helps them
"create more effective and performant ads through marketing recommendation"
2 Motivation
Initial rule-based recommendations had the budget ranges hard coded into the application where the user can can choose from.
But these may not fit their needs and it's a difficult decision to make in order to maximize returns.
3.1 ML | 2 Problems
Broken down into 2 problems: 1. predict the user's budget range they're likely to spend 2. "predict the likelihood of making sales"
These are possible because data was collected from the rule-based Kit
3.2 ML | Questions @vincentchio when you say "predict the likelihood of making sales", how would this take into account the expected profitability of a given ad?
Is this "predict whether the profits from ad > ad spend" or is it "predict whether sales > 0 from this ad spend"?
3.3 ML | Training Flow
feature exs:
- marketing: 30 day avg spend, 30 day marketing sales
- shop: 30 days unique visits, 30 day total orders
training: Google Cloud's ML Engine
monitor: 1) alert when metric exceeds threshold 2) alert on outliers, detected based on z-score
4.1 New Biz | Lookalike
Facebook advertising allows you to specify a Lookalike audience.
This is often a subset of customers that are your best customers and facebook will then provide ads to other users that "look like" them. facebook.com/business/help/…
4.2 New Biz | Lookalike Recommendation @Shopify can also give tailored ad recommendations for newer businesses, that are geared to helping them develop "Lookalike" audiences which can then be used for future facebook advertisements.
4.2 New Biz | Lookalike ML
Steps: 1. How many new visitors are required to establish a lookalike audience? 2. How much are they willing to spend? 3. Will the budget be enough to acquire those visitors?
4.3 New Biz | ML Qs @vincentchio curious as to what your target variable would be in order to know whether there's enough visitors to establish a lookalike audience?
Is it based on some % increase in conversion rates being above some threshold in comparison to a random pop?
• • •
Missing some Tweet in this thread? You can try to
force a refresh
In this thread I'll highlight some important pieces from a variety @ShopifyData & @ShopifyEng blogs where they discuss applications they've built that I think would benefit Data Scientists.
Contents: 1. How Shopify Capital Uses Quantile Regression To Help Merchants Succeed 2. How to Build an Experiment Pipeline from Scratch 3. How to Use Quasi-experiments and Counterfactuals to Build Great Products 4. Categorizing Products at Scale
Other Threads On Shopify DS Applications:
- How Shopify Uses Recommender Systems to Empower Entrepreneurs: