1
The Complete Transforma1on from Offline to Online Magento Implementa1on of a B2B Pla;orm – a Case Study. Tom Karwatka, Divante.co
This Case Study Shows How We Managed to Reach...
2
100.000.000 EUR online rev aOer Q3 of last year
B2B is The HoRest eCommerce Field Right Now
• Nearly 75% of B2B buyers say buying from an e-‐commerce site is more convenient that buying from a sales rep.
• 30% of today’s B2B buyers finalize at least half of their purchases online. By 2017, this will nearly have doubled to 56%.
3
• 60% of B2B companies -‐ their B2B buyers spend more when interac1ng with mul1ple channels.
• Omnichannel B2B customers are also more likely to become repeat and long-‐term customers.
hRp://divante.co/blog/b2b-‐ecommerce-‐2016/
About Client -‐ TIM
• TIM SA is the biggest electro-‐technical goods distributor in Poland and one of the biggest in the region.
• 30 years of experience in the market. It’s been on Warsaw Stock Exchange since 1998.
• Clients from the construc1on business area.
4
Growth Strategy: eCommerce
• Goal: diversifica1on of business and improvement of effec1veness – Expand customer base (new
industries, B2C clients) – Expand product range and add
new product categories.
5
It is difficult to convince the employees to introduce new ideas, especially such radical ones. Moreover, our vision and
strategy need to be clear-‐cut to the Board of Directors and owners.
We cannot begin without their support and -‐ as you can see -‐ the project is
planned to last for many years. -‐ Artur Piekarczyk, Vice-‐President
Start of coopera1on – 5.2011
Start of first B2C test project – 9.2011
Niche webshop with cables for B2C
clients Can TIM sell on-‐line ? Can we acquire new customer groups?
Start of B2B project -‐ 3.2012
Switching to SCRUM – 5.2013
B2B project going live – 7.2013
S.W.A.T – 11.2013
Major redesign – 8.2014
Moving hos1ng from TIM internal datacenter to outside
datacenter – 2. 2015
Advanced BigData project – 1.2016
Timeline of Coopera1on
6
Process
7
1. Business requirements. 2. Customer research.
Research
• Field research. • We had previously completed a B2C
project, so we had some interes1ng data about a new type of customers.
• Researches we did in this case: Compe1tors analysis, Surveys, Benchmarking, Trend analysis, industry reports etc.
8
In the picture you can see Kasia in one of our end-‐clients’ facili1es. During that field research we learned that sales reps prefer to call their call center and place orders directly to the operator.
All analy1cs packages provide key reports, such as most frequently visited pages, visitor demographics, bounce rate and more.
Personas
9
• A persona is a fic1onal iden1ty. • Personas help you to understand
what a client actually needs. • Usually, we create 3-‐4 personas
for each project. • A persona gives the customer
data a human face. • Personas also help prevent some
common design pi;alls, which may otherwise be easy to fall into.
Design Phase
• Prototyping – changes are much. • We use mockups -‐ the interac1ve prototypes. • We get our client involved in almost every phase of design.
10
Concep1on Selec1on of ideas
Prototype 1
Prototype 3
Prototype 2
Genera1ng Ideas Coherent
prototype
Workshop Findings Polishing the Prototype
Research Findings
Data from GA
6 clients of TIM
2 poten1al clients
4 itera1ons of
prototypying
User Research
Design Process
We work in teams. We generate ideas using personas.
Then we do paper prototyping
AOer that each team presents mockups and discusses them with other teams. Then we choose the best ideas and build one coherent mockup prototype.
Workshops
Paper Prototypes
12
Concept 1 Concept 2 Concept 3
Digital Prototypes
13
• Digital version of prototypes.
• Clickable and will allow us to conduct user tests.
User Tes1ng
• Usually 6 users are enough to determine 80% of usability problems.
• We do itera1ve tes1ng. AOer the tests tweak the mockups and test everything again.
14
In this case designing was a challenge: poten1al customers either had an experience with online stores or had no experience whatsoever.
Final Digital Mockup
Graphic Design
16
• It took us one month to finish the graphic work. • To shorten the 1me to market, in the mean1me, we carried on implementa1on.
Choosing the Technology
• Since 2008 so we have tried many eCommerce pla;orms.
• Why you should consider Magento: – Magento is currently the most
popular eCommerce soOware – Magento is the most popular
pla;orm to migrate to – You have the producer’s support
and an elas1c license model. – The soOware provides a rich set of
ready-‐to-‐use features. – Magento 2.0!
17 hRp://blog.aheadworks.com/2015/05/ecommerce-‐pla;orms-‐popularity-‐may-‐2015-‐two-‐pla;orms-‐take-‐half/
Alexa 1 million top sites survey unveiled leadership of Magento among world top eCommerce pla;orms.
Magento vs B2B Needs
You should consider these points: • Adding products faster (with photos and videos) -‐ PimCore or Akaneo. • Marke1ng automa1on that actually works and is easy to use -‐ Mau1c. • Recommenda1on engine -‐ Hadoop. • Search that works like Google -‐ Fact Finder • CRM allows your sales reps to work on your clients’ accounts – OroCRM • Custom B2B modules: a tree structure of the clients' account, payment
methods, flexible shopping cart.
18
Omnichannel Open Architecture
19 hRp://divante.co/blog/omnichannel-‐open-‐architecture/
Integra1ons
Nowadays the plaBorm is connected to the following systems:
• IntegraDon with CRM system: – Sending client’s data – Downloading client’s data – Cart synchronizaDon (price) – Cart downloading – Cart reservaDon – Sending orders – Sending RFPs – Downloading offers for a customer – Downloading product price – Downloading product price for a customer – Downloading invoices data and invoices in pdf – Downloading order’s history – Downloading order’s data
• Product Management Program – Downloading data on products and categories – Downloading catalogue cards of products and
pictures
• Warehouse IntegraDon – Downloading product availability – Downloading cable segments
• FactFinder IntegraDon
– Sending product database – Downloading search results
20
• Integra1ons are one of the hardest problems in eCommerce development. • Crea1ng failbacks -‐ if one of the systems is down we are able to operate -‐ s1ll • We use a dedicated soOware as a buffer layer.
CRM Live Integra1on
• CRM integra1on plays the key role in the system
• Bilateral integra1on in a real 1me -‐ the merchants in TIM’s central can edit the shopping carts of the customers that are currently logged on.
21
Reco engine Browsing
Call
Add to cart Client Consultant
Users Inspire Us
• One day a consultants finds an order for 100 light bulbs and a huge electric pole.
• To ship such a pole, you have to carefully arrange the delivery.
22
Dedicated Func1ons
23
Dynamic offer You want to buy 2 km of the cable. TIM has a 2.5-‐km-‐long piece in stock, they’d love to get rid of. So, they offer you their 2.5 km cheaper than your 2 km. It’s up to you to decide which length you will opt for.
The online cart is expanded. You have a bunch of specific func1ons, like communica1on with the merchant, dedicated payment methods, conver1ng cart to an offer inquiry and so on.
B2B Func1ons to Consider
The most common B2B func1ons we implement in our projects: ü List of contractors with verifica1on ü Individual or collec1ve nego1a1on of discounts ü Order files ü Offer crea1on ü Consultants who can accept the order on behalf of the customer by the
phone / email / during a mee1ng ü Saving and sharing the cart content ü Forms of payments that are typical of B2B (for instance: a trade credit)
In the cases like this one, all the dedicated func1ons are implemented as extra Magento modules. So you are s1ll able to update your Magento.
24
Development phase q Scalability and Performance q Quality q SLA q Effec1ve development q Data Driven Customer Experience
Implementa1on and Start-‐up
The implementa1on of the first version of the system took us over 14 months.
25
Start of coopera1on – 5.2011
Start of first B2C test project – 9.2011
Start of B2B project -‐ 3.2012
Switching to SCRUM – 5.2013
B2B project going live – 7.2013
S.W.A.T – 11.2013
Major redesign – 8.2014
Moving hos1ng from TIM internal datacenter to outside
datacenter – 2. 2015
Advanced BigData project – 1.2016
Scalability and Performance
• Replica1on of the database to speed
up reading opera1on and balance the traffic to database
• Using a cache system – in this case Memcached was replaced by Redis
• Using a scaRered file system (GlusterFS)
• Using an hRp-‐reverse-‐proxy system Varnish that speeds up loading pages and sta1c files
• Using PerconaDB instead of MySQL for the database
26
New Relic Performance Report, ELK Stack + Kibana
Ensuring Quality
• The extra risk factor: a big number of integra1ons.
• AOer every round of changes, we do the performance test to verify that the applica1on works beRer.
• To keep the quality high, we use – 2 phase Code Review – version control system, – 1cke1ng system (where all communica1on with
the client happens) – monitoring system – Incinga, Jenkins + Selenium
+ ELK Stack to monitor integra1ons
27
We use separate environments for developers, tests, and produc1on.
SLA
• 250.000 EURO daily rev. • We must be able to react under 1 hour and
prepare a workaround for any problem. • “Divante S.W.A.T.” – a specially formed unit
working 24/7. • The membership is op1onal but to sign up
every member has to pass an official exam. • Members get special bonuses added to
monthly salary. • The group responsibility -‐ if SLA is violated
on any website, nobody gets their bonus.
28
Divante S.W.A.T. is a unit, that covers security of the pla;orm and react ASAP when the system is down or an emergency happens.
SLA > 99.9%
Effec1ve Development
• To maintain effec1ve development of project we created a dedicated team
• SCRUM gives TIM the control and quality they need.
• SCRUM team – Product Owner – 4 Developers – 1 Front End Developer – 1 Tester
• Con1nuous delivery -‐ Ansible, Phing
29
Source: hRps://en.wikipedia.org/wiki/Con1nuous_delivery
To ensure con1nuous delivery we automate the tests and deployment.
Deployment time <5 min
Data Driven Customer Experience
We use data on a large scale to improve User Experience for TIM’s clients. To inspire you -‐ on TIM’s examaple: • 97% of customers preffer one
specific view of product list – „miniatures”
Business Results
31
The share of online sales in total sales in terms of income
The share of online sales in total sales in terms of No. of products
The number of the products ordered online is growing faster than the total value of online sales. That means that the online sales have bigger poten1al to diversify. On top of that, smaller but more frequent purchases build customer loyalty.
B2B project going live – 7.2013
Q3, 2015
• AOer the first year: – 50% of online sales – Over 50.000.000 EUR
online rev – 25.000 products
• AOer 3 years – Q1-‐Q3 – 67% of online sales
(85% of products) – Over 100.000.000 EUR
online rev aOer Q1-‐Q3
– 125.000 products
Business Results
• This project has been a huge success for TIM.
• They achieved their business goals: – Diversity of business – A lot of new customers – Improved business effec1veness
32
Clients tes1monials: hRps://www.youtube.com/watch?v=-‐QumO63G_7M&list=PLuXGEsfZdwIUuaIJTEd3PDSUQLilYVu9a&index=3
TIM client’s really like the B2B e-‐Commerce website
Harvard Business Review has wriRen an ar1cle about TIM recently. The Project was shown as a true digital transforma1on.
What’s Next?
• Improving the reten1on of clients through BigData analysis and personalized recommenda1ons.
• Automated pricing op1miza1on -‐ the next big thing we want to pursue.
33
In the Forrester research from 2014 -‐ Building The B2B OmniChannel Commerce Pla;orm Of The Future
• Automated pricing op1miza1on – 62% • Personalized recommenda1ons – 56% • Use of wearable compu1ng in distribu1on
centers – 49% • Sensors in shipments/products – 46% • Robots/automa1on in distribu1on center/
supply chain – 34% • Shopping cart abandoment analysis – 32%
B2B e-‐Commerce technologies to invest
Thank You!
Tom Karwatka Divante.co [email protected]
34
We base on the exper1se we’ve got from a number of B2B projects and generalize it to find the gold rules of B2B eCommerce.
Top Related