How to predict trends for e-commerce?

Forecas(ng for Internet Companies

E-‐commerce: methods and trends 2015 Class: Prognose und Prognosemethoden

für Medienmärkte, SoSe 2015, FU Berlin Dozent: Prof. Dr. Klaus Goldhammer by Yevstakhiy Biletskyy, 311816

Berlin 2015

E-‐commerce. Agenda of PresentaNon

Success Factors for E-‐commerce E-‐commerce forecas(ng challenges Before Forecas(ng. Metrics to consider Funnel Op(miza(on Predic(ng demand Predic(ng pricing Models for Forecas(ng in E-‐commerce -‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐-‐ Discussion Literature

the B2C e-‐commerce sales is e x p e c t e d t o g r ow 1 5 . 6% w o r l d w i d e t h i s y e a r a n d approximately 13% in 2016

StaNsta.com, 2015 Source

Figures from a study RetailMeNot (2014) and Centre for Retail Research (2014) via Econsultancy.com January 2015

Customers spend more money on online shopping. Year by Year. More and More!

How to manage this poten(al?

Success Factors for E-‐commerce

Erfolgsfaktoren im E-‐commerce – Deutschlands Top Online Shope vol 3, ECCC & Hermes, 2014 Source

Criteria


E-‐commerce Goals? For Retailer:

Cost Effec(veness For Client: Product Sa(sfac(on

How? Purchasing Planning-‐>Strategy-‐>Forecast

E-‐commerce Forecast: (1)  Demand Planning

(predict demand and profit)

(2) Inventory Management (make sure items are available)

Bad News

Forecas(ng is almost always

wrong L

Biggest Challenge: Accuracy of Forecas(ng

E-‐commerce Forecast Challenges:

(1)  Behaviour of customers (2)  Seasonal changes (3)  Trends (4)  Demand pa]erns of compe(tors (5)  Mul(channel Commerce (6)  Classical Forecast methods fail

Before Forecas(ng: What we need to do?

(1) Understand the Lifestyle Value of the Customer

(2)  Break customers into groups

(3)  Introduce Net Promoter Score

(4) Monitor Customer Sa(sfac(on

(5)  Indicate ROI

Define Conversion

Funnel as a level of loyalty

Funnel as a decion making

process

Predic(ng Demand

Sales Forecast Techniques for Emerging Industries, MonarchControls [source]

For Old Products: Historical Product Data, Demographics, (me series analysis, cohort analysis

For New Products: Data Mining + Forecast Experience + Trends + Cohort Analysis = Regression Analysis (dependent elements)

Forecas(ng a good price

Consider: (1) Margins (2) Unique Selling Proposi(on (3) Lose Leader Strategy (upselling) (4) Effec(veness of Promo(ons

Campaign-‐>Conversion-‐>ROI Test, Adjust and again Test

Ok, but How? J

Neteconomy Models (1) Cohort Analysis – break users into related groups with a common element

Time Series Analysis – measurements over a Nme interval Regression Analysis – complex relaNonship between variables

Online Marke(ng Tools as a forecas(ng method (2) SEO – keywords -‐> Your site -‐> conversion Dispaly Ads – display ads -‐>Your site -‐> conversion Email MarkeNng – email campaign -‐> Your site -‐> conversion Test with Google AnalyNcs!

New Technology CRM communicate with Your customer (3) -‐ grade transacNon -‐ post-‐purchase saNsfacNon surveys -‐ random check outs Tools: Onsite survey, Chat, UGC, emails, Social Media, telephone Want to hear the Voice? Make Sure You offer them value in return

(NET)ECONOMY

A/B Tes(ng ongoing (4a) What? Web Traffic split between 2 versions of the site. Goal? Establish the objecNve eg. increase sales, Nme spent on the site, returning visitors You can test: Content, usablity, accessability, product performance, funnel, errors

Old

New

… increased revenue by 56 %

Removing Drop Down increased conversion on e-‐commerce website, Case Study, 2013 Source

A/B/n, Mul(variate Tes(ng (4b) web traffic is randomly split between exisNng web and tesNng sites

Google Trends

Using Sojware (5)

How to evaluate market demand for the product?

(1) Search Interest for Coconut Oil, 2004-‐2015 in Google

(2) Regions in Google Trend

(3) Evalua(on of keywords in AdWords

(4) SERP Compe((on

(5) Social Media analysis of the Sen(ment via Topsy

How to create an ideal forecas(ng model for online shopping? What to consider? A) Qualita(ve B) Quan(ta(ve C) ?

Sources Nian Shing Chen, E-‐commerce ForecasNng Demand for New Products, hlp://www.academia.edu/2957971/E-‐COMMERCE_FORECASTING_DEMAND_FOR_NEW_PRODUCTS 9 E-‐commerce trends in 2015, hlps://www.shipwire.com/w/blog/9-‐e-‐commerce-‐trends-‐2015-‐influence-‐buyer-‐experience/ 20 E-‐commerce Trends and PredicNons in 2015, hlps://econsultancy.com/blog/65929-‐20-‐ecommerce-‐trends-‐and-‐predicNons-‐for-‐2015/ Types of E-‐commerce, hlp://bloomidea.com/en/blog/types-‐e-‐commerce How To evaluate Market Demand for Your New Product Idea, hlp://www.shopify.com/blog/13444793-‐how-‐to-‐evaluate-‐market-‐demand-‐for-‐your-‐new-‐product-‐idea US 2015: Digital Future in Focus?http://www.comscore.com/Insights/Presentations-and-Whitepapers/2015/2015-US-Digital-Future-in-Focus Nielsen Global Survey on e-‐commerce, March 2014, http://www.marketingcharts.com/online/which-e-commerce-product-categories-are-seeing-increasing-purchase-intent-45462/ Sales Forecast Techniques for Emerging Industries, MonarchControls, http://monarchcontrols.trustpass.alibaba.com/product/125897063-103353399/Sales_Forecasting_Techniques_for_emerging_industries_on_the_web_like_eCommerce_Search_Engine.html Removing Drop Down increased conversion on e-‐commerce website, Case Study, 2013 Source E-‐commerce Physical and Logical ConnecNon Methods, White Paper, NX Trend 2001 2015 Forecast: E-‐commerce, MulNchannel Merchant, 2015


How to predict trends for e-commerce?

Internet

ecommerceforecast challenges

e x p e c t e d t o

d w i d e t h i s y

n d approximately13

casestudy source

b2c ecommerce sales

meseries analysis

funnelasa levelof loyalty