Learning from Amazon’s

Operating Model
Engineering a customer-centric transformation

By Michael Ross, Co-Founder and Chief Scientist, DynamicAction

Over 25 years, Amazon has developed a strategic

approach that is firmly centered on its customers and
powered by data. So how can you leverage Amazon's
experience in creating and operating a customer-
focused model in your business?

For the 4 core characteristics of Amazon’s successful

model and 12 rules for putting it into practice, read on.

dynamicaction.com
Learning from Amazon’s operating model
How to engineer a customer-centric transformation
Like it or loathe it, Amazon is a colossus that can’t be ignored. In 2018, Amazon did
$117b in first pary sales and $160b in third party sales for a total of $277b in gross
merchandise volume . Many naysayers complain that Amazon doesn’t make any
profits—this is simply not true. Amazon made over $3b in profits last year. Even
more importantly it takes a very long-term view of customer profitability, with a
model that allows it to navigate the trade-off between growth vs. profitability.
Over 25 years, Amazon has developed a strategic approach that is firmly centered Data is the new oil—it
on its customers, and powered by data. All retailers need an “Amazon strategy,” and needs to be extracted,
we believe this has two fundamental components:
processed and refined to
1. A value proposition based around exclusive products, be turned into
service and experience that gives a sustainable point of something
difference. useful.
2. An operating model that takes the best bits of the Dr. Andreas Weigend,
Amazon approach. former Chief Scientist at Amazon
and Advisor at DynamicAction

So how can you apply some of this Amazon-style

customer-focused model to your business?
Setting the scene: how consumer data changed the world
Profit in retail has always been driven by customers. But historically, they have
mostly remained anonymous, with the exception of local, traditional forms of retail
where a shopkeeper knows each customer by name. Retailers have been able
to make good enough business decisions just by looking at the profitability of a
channel, store, product or category.

Today, digitally-enabled consumers have changed the economics and dynamics

of retail. With almost unlimited choice and immediate access, consumers are
empowered and redefining the shopping experience. They are seamlessly shifting
between a range of devices, and are researching, shopping and even returning
in new and complex ways. Importantly, they also now leave a “digital exhaust”—a
cross-channel breadcrumb trail made up of every impression, click, purchase,
loyalty credit, basket addition and return.

This dynamic drives intense competition for every potential customer. As well as
Amazon, other obvious winners are Google, FedEx and, of course, consumers. The
critical challenge for retailers and brands is to find a profitable path, work out how to
make sense of all their data and rethink their approach to making decisions.

2
The so what: a seismic shift in the fundamentals
of retail
These changes in consumer behavior are catalyzing massive shifts in the
fundamental economics of retailing. Retailers who have been accustomed to Data is only valuable
analyzing hundreds of stores, or thousands of products, suddenly have to if it helps you make a
contend with understanding millions of individual customers.
decision.
• New revenue drivers: Revenue is now driven by customer acquisition and Dr. Barney Pell,
retention, a more complex dynamic than traditional comp sales (like-for- artificial intelligence pioneer and
likes). It is critical to understand which customers are high value (hang on to DynamicAction Advisor
them, and find more like them); and which lose money (find out why and/or
stop doing business with them).
New costs: A new set of variable marketing costs (per impression, per click,
DynamicAction
•

empowers us to make
per transaction), and variable “per order” costs (picking, packing, postage,
returns) are replacing the more predictable costs of staff and rent.
Crucially, profit no longer correlates with revenues. decisions and take
• New data and decisions: Consumers create deep lakes of data that action quickly and
managers can use—each shopper’s digital exhaust has millions of data promptly as a merchant
points. Managers need to collate massive and messy data sets to fully team. I look at
understand customer profitability, and keep pace with the inexorable trend opportunities across all
functional areas to pick
towards surgical, semi-automated decision making.
New organizational interconnection: Actions and decisions were once
•
relatively easy to understand and could be assessed within a functional silo,
the opportunities that
by store or by region (which drove the evolution of traditional retail make the most sense at
organization structures). Now they are more complex and interconnected. that point in time.
Merchandisers, marketing and store/site operations must triangulate to Kamlah Frances
decide how the next dollar will be made or spent on products, marketing or Web Analytics Specialist
customers.

The Amazon way: a customer-focused operating model

Retailers are beginning to recognize that the traditional operating model no longer works. It is incredibly easy NOT to make
money—to design propositions that customers love, shareholders hate, and that drain working capital from the business. The
Amazon approach effectively adds a customer dimension across the business. This new operating model has four core
characteristics.

3
 4 Core Characteristics
of the Amazon Model
1
Consumer Profit
OLD MODEL AMAZON MODEL
• Focused on revenues, which made • Deep understanding of profit by customer and by SKU.
sense when costs were mostly fixed • Focused on lifetime customer profitability.
(e.g., rent, staff) and discounts and • Comprehensive view of the purchase process across
promotions were tightly controlled.
channels, from first marketing touch to delivery and
returns.

2
Surgical Actions
OLD MODEL AMAZON MODEL
• Actions are blunt, based on sparse • Focused on specific actions and applied at the right level
data and reliant on physical of granularity: customer, SKU, keyword, site page, store
and transaction level.
execution.
• Developed model of “high value actions” where they
• Predominantly taken at the broad understand the impact of customer actions on long
level of stores, categories or term value.
customer segments. • Approach to type 1/2 decisions where focus is on speed
to action for reversible actions.

3
Customer Centered Metrics
OLD MODEL AMAZON MODEL
• Comp sales and outcomes are used • Measurement of things that matter to customers to
to tell a simple story. drive the right conversation (controllable input metrics).
• De-averaging of metrics and reports, with the focus on
exceptions and outliers.
• Celebration of waste as an opportunity to improve

4
Automated & Coordinated Execution
OLD MODEL AMAZON MODEL
• Physical execution within silos, • Semi-automated execution across silos enabled by
driving “good-enough” results. workflow/APIs, as well as coordinated action across
internal teams.
• Leaders are trained to manage hybrid teams of humans
+ AI.

4
 12 rules for operating
like Amazon
Focus on customer profit
1. Build your plan based on customer acquisition and retention. Review performance and
customer cohorts by month, and against the plan.
2. Understand which customers make or lose you money, and how (e.g. by channel, offer,
basket).
3. Design a value proposition, such as delivery options, free delivery, promotions or
markdowns, aligned to customer profitability.

Take surgical actions

4. Add a customer-focused lens to strategic decisions around range, stores and
marketing. What are the products, brands or keywords that acquire and retain high-
value customers?
5. Take tactical actions at product/SKU, customer and customer/SKU level. Avoid using
blunt discounts and broadcast promotions to solve a SKU overstock.
6. Profit-hunt with precision across keywords, landing pages, transactions, sessions, etc.

Discuss customer-centric metrics

7. Measure customer outcomes, but focus on controllable, actionable input metrics
rather than catching every data point.
8. Design reports that de-average performance to focus on outliers, failures and
anomalies you can learn from.
9. Develop insight that recognizes the distribution of customer lifetime value.

Semi-automate execution
10. Recognize the need for connected teams, data and action.
11. Automate the trivial and the repetitive, and ensure that the entire tech stack is
powered by clear and connected data.
12. Empower team members to act quickly and decisively.

5
About DynamicAction
DynamicAction is a retail analytics guidance system that leverages
cloud software and a proven success program to catalyze the new
customer-first operating mindset in retail. DynamicAction
empowers retailers with a clear path to navigate their
transformational journeys with AI-powered metrics. It enables
faster, better decisions to deliver profit, analytics and
visualizations for immediate insights, prioritized opportunities and
prescribed actions to take online and in-store.

Forward thinking retailers across the globe rely on

DynamicAction’s advanced analytics and retail-built practices to
holistically run more efficient organizations and formulate laser
target strategies to uncover their most profitable customers.
Forrester Research recommended DynamicAction as the key
prescriptive analytics technology to replace predictive analytics in
retail, and the National Retail Federation awarded DynamicAction
for its ability to "significantly improve or radically alter how retailing
is done."

Headquartered in Silicon Valley, DynamicAction has offices in

London, Sofia and Dallas.

Connect with us at
www.DynamicAction.com

Twitter
@DynamicAction

LinkedIn
DynamicAction