Download as ppt, pdf, or txt
Download as ppt, pdf, or txt
You are on page 1of 13

Increasing Sales

through
Recommendation Systems
Strategy and Customization
(Creating a Personalized Online Shopping Experience)

Presented by

Dart Marketing, LLC

1
Impact of Recommendations Systems

Industry research indicates that successful online retailers are generating


as much as 35% of their business from recommendations. Are you leaving
money on the table?

Current YR1 YR2 YR3 YR4 YR5


Assumptions
Revenue $ 100,000 (thousands)
Lift From Reco's 15.0% 22.5% 28.1% 31.6% 34.2%
Profit Margin 10.0%

Results
Gain in Revenue $ 15,000 $ 22,500 $ 28,125 $ 31,641 $ 34,172
Cum $ 15,000 $ 37,500 $ 65,625 $ 97,266 $ 131,438

Gain in Profit $ 1,500 $ 2,250 $ 2,813 $ 3,164 $ 3,417


Cum $ 1,500 $ 3,750 $ 6,563 $ 9,727 $ 13,144

Meanwhile, the race is on because once someone invests the time to provide
feedback to a website, he/she will prefer to shop there versus somewhere else.

Personalization increases the amount of time people spend at the site and increases
repeat visits, growing loyalty and sales.

2
Strategy

Styled to conform to
your site:
New Shoppers
 Scrollable, community recommendations
 Checkout – Accessories, cross-sells

Displays
Repeat Buyers (Personal Shopping Assistant) Community
Ratings
 Scrollable, personalized recommendations
 Profile Building: Collect product ratings
 Smart Search tied to user preferences so it shows Collect
items customers will like Ratings If
Logged-In
 Can be linked to existing search, or
 Used on it’s own. Ours displays everything the
customer needs (not just list of products). Sort by
popularity, price, etc. Remove
Unwanted
Items

Scrollable Recommendations
3
It’s All About the Data

The better the data, the better the recommendations. It’s that simple.

 Product Ratings Data: Customer


feedback improves quality of
recommendations. Collecting this data
builds a relationship.

 Sales Data: Most useful after ratings


data. Removing gift purchases ensures
matches are based on customers’
personal tastes.

 Click Data: Used as a last resort. A


poor predictor of customer purchases. Philosophy: Collect the best data, derive a
Does nothing to build a relationship. custom solution, then validate initial results by
hand. Once foundation is set, profiles update in
real time.

4
Consultative/Analytic Approach

The “Also Bought” method shows what’s popular, but not what’s most
relevant. Our affinity-based approach is much more effective.
Overview of Coffee Varieties
 A different approach: map product I2
I1C6
affinities in multi-dimensional space.  S5 C1
L5 Exotic
S2S1 S7
S6
 Successfully applied to Gevalia

Exoticness / Price
R4 C7
coffees and teas, and re-validated S3 R6 R3
R2
C2 Flavored
with Netflix movie data. B2 C4 a1
L4C3 S4 TE FR
F3 F2
B1 F9 F8 F1
F0 F6
 Extraordinary recommendations R5 F5
R8
accuracy for new shoppers - even Popular Roasts F4

better for returning customers. and Blends


Com plexity of Flavor
 Results are applicable to product and
Affinity Maps & Recommendations:
pricing strategies, offline
merchandising, cross- & up-selling The bubbles above represent products sized by
sales volume. Products close to each other are
recommended to each other.

5
Customization Dashboard

Add product details such as…


Sku, Type, Sub-type, Category, Brand, etc.
SKU SUBTYPE TYPE CATEGORY DATE
KN121
NJ441
DVD PLAYER
CD
PORTABLE
COUNTRY
ELECTRONICS
MUSIC
11/12/2007
1/18/2008

TY552 DVD MOVIE VIDEO 11/12/2007

Pricing strategy
TX551 DVD CONCERT VIDEO 11/12/2007
KN121
NJ041
DVD PLAYER
CD
PORTABLE
ROCK
ELECTRONICS
MUSIC
11/12/2007
1/18/2008 
TY552 DVD MOVIE VIDEO 1/18/2008
TR466 DVD SHOW VIDEO 11/12/2007

Dates
TX551 DVD CONCERT VIDEO 11/12/2007
KN121
NJ997
DVD PLAYER
CD
STANDARD
POP
ELECTRONICS
MUSIC
11/12/2007
1/18/2008

Customize recommendations by…


 Business rules
 Profit margin
 Inventory availability
 On-sale items and other criteria

Reports include…
 Sales stats from recommendations
 Web stats relating to recommendations
 Date periods
 Custom reports available
6
Secondary Strategies

Other uses for customer profile data …


Product/Pricing Strategies

My-Gift Store (Recommend Gifts)


 Recommendations for when bill-to and ship-to don’t match
 Email gift reminders with recommendations
 Encourage customers to add new gift recipients

New Reasons to Email


 Post-sale requests for product ratings
 Personalized promotions

Telemarketing Up-Selling

Direct Mail: Highly recommended products

Catalog Merchandising

7
Next Steps

SKU
KN121
NJ441
SUBTYPE
DVD PLAYER
CD
TYPE
PORTABLE
COUNTRY
CATEGORY
ELECTRONICS
MUSIC
DATE
11/12/2007
1/18/2008
Collect Data
TY552 DVD MOVIE VIDEO 11/12/2007
TX551
KN121
NJ041
DVD
DVD PLAYER
CD
CONCERT
PORTABLE
ROCK
VIDEO
ELECTRONICS
MUSIC
11/12/2007
11/12/2007
1/18/2008
 Gather transactions data, then refine to exclude gifts,
ship-to’s, etc.
TY552 DVD MOVIE VIDEO 1/18/2008
TR466 DVD SHOW VIDEO 11/12/2007
TX551 DVD CONCERT VIDEO 11/12/2007
KN121 DVD PLAYER STANDARD ELECTRONICS 11/12/2007
NJ997 CD POP MUSIC 1/18/2008

Analyze Data
Overview of Coffee Varieties
I2
I1C6 L5
C1 Exotic
S5
S2S1 S7

Begin by mining and mapping affinities.


S6
R4 C7 
Exoticness / Price

S3 R6 R3
R2
C2
C4 a1 Flavore d
B2 L4 C3

Create demo to compare new recommendations with


S4 TE FR
F3 F2
B1 F9 F8 F0 F6
F5
F1

R5
R8

current ones.
Popular Roas ts F4
and Blends
Com ple xity of Flavor

Further Personalize and Enhance Initial Solution


 Motivate shoppers to share ratings to further personalize
their recommendations.

8
Why Dart? Its People

You will have a business relationship with experts in the industry…


Craig Tomarkin, President (Affinity mapping, modeling, research, analysis)
Craig has spent his career converting ideas into profit. He helped GM design and launch the
world’s first free “rewards” credit card, resulting in 5 million accounts in the first year. For Gevalia
Coffee, he developed an innovative product mapping technique that optimized cross-selling,
pricing, and new-product strategies – a precursor of his current eCommerce recommendations
strategy. Craig holds a BSM from the A.B. Freeman School of Business at Tulane.

Paul Delano, Technology Expert (Java, eCommerce, SEO, hosted solutions)


Paul’s innovations in artificial intelligence and collaborative search have led to his being awarded
four patents. He created the first Internet commerce site for PC Flowers.com as well as the
infrastructure for a nationwide interactive television system. He has taught Java courses at
companies like JPMorganChase and Hewlett Packard. Paul received a MS in Computer & Systems
Engineering from Rensselaer Polytechnic Institute and a BS from Carnegie Mellon.

Phil Goodhart, Direct Response Marketing Expert (Client support, eCommerce)


Phil is a veteran of the Danbury Mint, a leading direct marketer of consumer merchandise. He
was recognized as a premiere marketing strategist, as well as an innovator in identifying new
product opportunities. He managed the development of the Danbury Mint’s first eCommerce site.
Phil earned his MBA at Harvard and BA at Princeton.

9
Dart Vs. The Competition

Distinguishing Benefits Dart Others


Interactive, scrollable recommendations (√) ( )
Collect product ratings to enable customers to train site (√) ( )
Extraordinary accuracy through Affinity Mapping techniques (√) ( )
Solution based on sales data and customer product ratings (√) ( )
Analytical, human-based approach tailored to each client (√) ( )
Mapping, pricing strategy, and other analytic consulting services (√) ( )

Other Benefits
Easy to integrate
Hosted solution
Client can customize recommendations
Search engine included
Optional performance based pricing

10
Example: Movie Recommendations I

Our recommendations stand up to the world’s best.

DARTM Netflix Blockbuster Best Buy Walmart


Caddyshack Caddyshack Caddyshack Caddyshack Caddyshack

Nat Lamp.'s Animal House Tommy Boy NL's Van Wilder Caddyshack II Natural Born Killers
Nat Lamp.'s Vacation NL's Animal House NL's Christmas Vacation DC Cab
Stripes N's Vacation Anchorman Bio Dome
Airplane! Airplane! Blues Brothers Back to School

Caddyshack: Slapstick Comedy. Chevy Chase, Bill Murray, Rodney Dangerfield.


1980.

Key Findings: Dart’s system doesn’t recommend Caddyshack II, even though it’s a
sequel. Since it had a different cast, and customers did not rated it as highly as the
original, it is not as relevant a match.

Note: There are thousands of examples on our web site. Click on dvd demo.

11
Example: Movie Recommendations II

Our recommendations stand up to the world’s best.

DARTM Netflix Blockbuster Best Buy Walmart


Sleeper Sleeper Sleeper Sleeper Sleeper

Bananas Annie Hall Bananas The Lonely Guy Star Trek IV


Love and Death The Producers Hardware Wars Modern Romance The Court Jester
Everything You Always Wanted to Know About Sex Bananas Love and Death What's New Pussycat? The Time Machine
Take the Money and Run Take the Money and Run The President's Analyst Hannah and Her Sisters

Sleeper: Slapstick Romantic Comedy. Woody Allen, Diane Keaton. 1973.

Key Findings: Dart’s system recommended Woody Allen romantic comedies


exclusively and chose titles from the matching time period. His movies match very
closely, indicating he is a genre unto himself
(i.e. a “Woody Allen” movie).

Note: There are thousands of examples on our web site. Click on dvd demo.

12
Contact Info

DART Marketing, LLC

Craig Tomarkin, President


CTomarkin@DartM.net
203-259-0676

Phil Goodhart, VP Business Development


PGoodhart@DartM.net
203-261-4731

http://Dartm.net

13

You might also like