Shopify Illustrates Indirect AI Monetization Model

E-commerce use cases for artificial intelligence often will be indirect, as now is the case for 

Shopify Magic, the AI platform now offered to Shopify merchants.

The company says Shopify Magic "is a suite of free AI-enabled features that are integrated across Shopify's products and workflows to make it easier for you to start, run, and grow your business." In other words, the AI monetization comes from the fees Shopify earns from its customer retail operations. 

In 2023, about 68 percent of Shopify revenue was generated by Merchant Solutions, while about 32 percent comes from subscriptions.

In 2022, Shopify's revenue came from the following sources, subscriptions and merchant services, of which about 71 percent of total revenue was generated by Merchant Solutions. That includes such features as: 

• Shopify Payments, a payment processing service that allows merchants to accept credit and debit cards online and offline.

• Shopify Shipping: A shipping solution that helps merchants get discounted shipping rates from major carriers.

• Shopify Capital: A financing solution that provides merchants with loans to help them grow their businesses.

• Shopify App Store: A marketplace where merchants can find and install apps to add new features and functionality to their stores.

• Shopify Partners: A program that connects merchants with experts who can help them with everything from design and development to marketing and customer service.

Shopify also makes money from Subscription Solutions, which generate about 29 percent of total revenue. Think of that fee as similar to a mobile subscription, providing merchants access to the Shopify platform, as well as features and tools to help them create and manage their online stores.

Use of Shopify Magic is provided at no additional charge. Shopify expects to monetize the features by attracting more merchants who will use more Shopify services, including payment processing, shipping and finance, for example.

So Shopify monetizes indirectly, by gaining and retaining more customers, while generating fees for providing a variety of services supporting their retail customer sales.

Using Customer Twin Models Beyond Recommendations

“Customer twin models” (a digital twin model focused on consumer behavior) seem to be finding early application for retailers selling products and content. The obvious early use case is recommendations based on prior behavior. 

Amazon uses customer twin models to personalize product recommendations, for example, based on past buying and searching behavior. 

Netflix uses customer twin models to recommend movies and TV shows. Spotify uses customer twin models to personalize music recommendations.

Since a customer twin is a virtual representation of a real customer, using data including demographic information, purchase history, website browsing behavior, and social media activity, other business use cases seem viable as well. 

Customer twin models can be used to create personalized experiences across various touchpoints, including websites, mobile apps, and in-store interactions. Businesses can tailor content, navigation, and product offerings to each customer's unique needs. 

Such models can enable businesses to proactively identify customers who are at risk of churning or dissatisfaction.

Customer twin models also can be used to detect fraudulent activities, such as unauthorized transactions or account usage anomalies, or to optimize marketing campaigns and improve customer acquisition strategies.

They can additionally be used to develop predictive analytics models that forecast future customer behavior, such as purchase intentions, product preferences, and potential churn. 

The models also can enable businesses to offer personalized pricing and promotions tailored to each customer's unique needs and purchase history.

In the telecommunications industry, customer twin models can be used to optimize network performance and capacity planning. By analyzing customer usage patterns and traffic data, telecommunications companies can anticipate demand fluctuations and allocate resources accordingly. 

Why Good Advice So Often Fails to Elevate Connectivity Service Provider Outcomes

Like any other business, internet service providers get some common recommended approaches for sustaining their profit margins under competitive market conditions. And one might argue there are good reasons why most--if not all--of these suggestions seem to have relatively little impact on industry fortunes.

If one assumes connectivity demand is both ubiquitous but also relatively inelastic, there is only so much customers are likely to spend on connectivity. As a percentage of household income or firm income, for example, connectivity spending is relatively low and does not change much.

Country	Household Connectivity Spending	Business Connectivity Spending
United States	2.0%	5.0%
Canada	3.0%	4.0%
Mexico	2.0%	3.0%
Brazil	3.0%	4.0%
Argentina	4.0%	5.0%
China	2.0%	3.0%
Japan	1.0%	2.0%
South Korea	2.0%	3.0%
India	3.0%	4.0%
Indonesia	2.0%	3.0%
Germany	2.0%	3.0%
France	2.0%	3.0%
United Kingdom	2.0%	3.0%
Italy	2.0%	3.0%
Spain	3.0%	4.0%
Egypt	4.0%	5.0%
South Africa	3.0%	4.0%
Nigeria	2.0%	3.0%

Among the range of recommendations are revenue-enhancing and cost-reducing measures. On the revenue side of the business model often include measures such as investing in network upgrades to increase capacity and improve performance. 

Other frequently-recommended stops also include:

• Bundling new products

• Create or raise prices for higher-usage customers

• Create or raise prices for higher-performance product tiers

• Maintain or create usage caps

On the cost side of the business model, ISPs often are urged to reduce operating costs or find methods to reduce capital investment. 

The issue is often that some of these steps help, but also require some additional costs, or can reduce take rates, increase churn or reduce market share, all of which are generally negatives. 

It might be very hard to cite any particular tactic whose revenue upside is clearly and consistently greater than the costs required to implement the policies. At least hypothetically, about the best an ISP can generally do is institute policies that increase revenue slightly more than new costs are imposed. 

Arguably, the cost of network upgrades--copper access to fiber for telcos; DOCSIS upgrades and FTTH for cable operators; 5G upgrades for mobile operators--impose high costs. In many cases, creating and then bundling new products also imposes new costs, even when new revenue also is generated. 

As a thought exercise, consider the benefits of margin protection that various tactics might represent, balanced by the cost of instituting those measures. 

Consider the cost of upgrading an access network to fiber-to-home, for example, or creating new products to bundle, reducing operating costs. All might help, but also require investments or additional costs. 

In the table, the degree of help for protecting margins also is accompanied by increases in total capex and opex costs, for example. 

At least in principle, if prices can be raised without causing customer churn, the upside clearly outweighs the cost (instituting price increases that most customers will accept). But nearly all the other tactics, while potentially yielding higher revenues, might also cause customer irritation and raise the risk of lost market share. 

Profit protection technique	Estimated profit margin protection	Cost percentage
Network upgrades	10%	20%
Bundling new products	5%	10%
Reducing operating costs	3%	5%
Raising prices	2%	1%
Charging for heavy usage	1%	1%
Using data consumption caps	1%	1%

In a way, the relatively-modest return from such tactics illustrates an aspect of the “low-growth” nature of the connectivity business in general; something the industry shares with other capital-intensive industries with utility characteristics. 

Demand is ubiquitous but also relatively inelastic. Under such conditions, there are limits to buyer spending propensity, no matter what is done to increase value, features or adjust prices.

Problem: 5G Cost More than 4G; 6G Will Cost More than 5G

By some estimates, the cost of 4G and 5G networks has gotten more expensive, and 6G is expected to be more expensive than 6G. 

There are several reasons, including the cost of new spectrum; the need for greater numbers of small cells, each supported by optical fiber connectivity; the cost of more-complicated radios; perhaps higher-cost engineering and higher site acquisition costs. 

Technology	Cost per location	Cost per square mile
4G	$10,000-$50,000	$500,000-$2,500,000
5G	$25,000-$100,000	$1,250,000-$5,000,000
6G (estimated)	$50,000-$200,000	$2,500,000-$10,000,000

All of that drives mobile operator concern about the business model for 5G and 6G. Some of that concern is about revenue, but much of the issue is the ever-higher need for capacity. 

By general agreement, mobile operator capacity gains have historically been driven by use of smaller cells (network densification) and allocation of additional spectrum. But most observers would tend to agree that denser architectures have contributed the most. 

In addition to use of smaller cells and additional spectrum, Wi-Fi offload, better radio technologies and modulation techniques also have contributed. And the mix of contributors arguably has changed over time. For example, Wi-Fi offload was not a factor for 2G networks.

In the 4G and 5G era, Wi-Fi offload might represent as much as 75 percent of mobile device data (principally internet access), but rarely less than 45 percent of total mobile internet data. 

Country	Percentage of mobile phone traffic offloaded to Wi-Fi
United States	60%
China	70%
India	50%
Japan	65%
South Korea	75%
United Kingdom	55%
Germany	60%
France	50%
Brazil	45%
Russia	55%

As mobile executives resist the ever-growing amount of capital they must spend to increase capacity, data offload might be one of the most-fruitful ways to add effective capacity while containing capital investment, at least to a point.