Deli XL (food service provider, NL, B2B, €750M/yr) decided to implement ATG, while replatforming the back office at the same time. In March 2012, the first ATG release was delivered with an exceptional TTM: multiple Production deployments can be released per week. This presentation, presented at the Oracle CRM Day 2012 in Etten-Leur (The Netherlands), explains how ATG is reshaping Deli XL and how our customers experience that.
20. Software Selection: E-commerce
> Long list: h o rt l i s t > Why
s
> ATG > merchandising
> Hybris capabilities
> IBM > personalization
> Intershop > customer focus
> Microsoft > leader (Gartner)
> SAP > growth platform
(multi-site)
21. This Presentation
1 The Prelude
2 The Plan
3 The Project
4 The Pilot
5 Conclusion
6 Questions
22. bol.com → Ernst → Deli XL 2010
Q1 Q3
Q2 Q4
> I thought I knew e-commerce
> …but that was B2C
> E.g.: conversion rate:
> bol.com: …
> Deli XL: 100%
23. Online Regular B2C B2B ordering site with
presence webshop (ATG) purchase lists (AS/400)
Payment Online, during checkout Invoices
Large catalog, Catalogs organized by
Volumes
millions of products logistical processes
Recommendations,
Personali- Different catalogs, prices,
targeted campaigns,
zation delivery schemes, etc.
etc.
Statistics, segmentation, We know each one.
CRM
service center, etc. Personally; we go there!
Recurring, e.g. same
Types of orders Random
products every week
24. Like the old days, before the Web kicked in:
first authenticate…
28. Example: Deli XL Customer
> offline
> when he calls, he says his name
> …then immediately hangs up
> …and expects us to call back
> …which we do.
Yes, this is Holland
29. Project Status @ Q4, 2010 2010
Q1 Q3
Q2 Q4
> Business readiness: very limited
> ATG expertise: very limited
> Project approach: waterfall
> Integration strategy: big bang
30. Translation 2010
Q1 Q3
Q2 Q4
> Business does not know what is coming
> …the scope is carved in stone
> …there are many dependencies
> …we are planning a big bang
> …we do not have local experts
31. Translation 2010
Q1 Q3
Q2 Q4
> Business does not know what is coming
> …the scope is carved in stone
> …there are many dependencies
> …we are planning a big bang
> …we do not have local experts
Ow… kay…
32. 2010
Q1 Q3
Q2 Q4
How often should we be able to do a release?
One a year should do.
That’s… not a lot. Are you sure?
Perhaps 2…
33. Analysis of Legacy System 2011
Q1 Q3
Q2 Q4
organisations, sub-orgs,
web accounts, products,
customer-specific
assortments, DC-specific
AS/400 assortments, prices,
promotions, ordering
schedules, delivery
schedules per temp.
stream, exclusive
suppliers, product
34. Analysis of Legacy System 2011
Q1 Q3
Q2 Q4
organisations, sub-orgs,
web accounts, products,
customer-specific
assortments, DC-specific
AS/400 assortments, prices,
promotions, ordering
schedules, delivery
schedules per temp.
Nope: that will not map to exclusive
stream, the ATG model
suppliers, product
35. Updated Plan 2011
Q1 Q3
Q2 Q4
> Agile approach
> Start development ASAP
> Closely involve the business
> Start building now
> Address integrations ASAP
> Local ATG expertise
> High quality standards
36. 2011
Q1 Q3
Q2 Q4
We could aim for releasing once every 3 weeks?
Nah, once every 6 weeks is more than enough.
37. Execution 2011
Q1 Q3
Q2 Q4
> Smart Architecture
> separation of concerns — embrace change
> Small team of experts
> started with 4 developers, scaled up to 8
> Continuous Delivery
> from developer to Production within an hour
> can do multiple releases a week
38. Smart Architecture
shop 1 shop 2 …
AS/400 (gap) ATG 10
data and old
business logic nice and shiny
SAP
some new data
39. Smart Architecture
shop 1 shop 2 …
Giga
AS/400 ATG 10
Spaces
data and old
business logic nice and shiny
SAP
some new data
40. Small Team of Experts
> Initially specialists only
> ATG, GigaSpaces, UNIX, test automation
> 10+ years
> Scale up with ‘juniors’ (4+ years)
> Moving to mixed on-/offshore team
41. Continuous Delivery (CD)
Without CD With CD
Production release time Days/weeks Minutes
Low
Release stress High
(business as usual)
Risk of a software release High Low
TTM Difficult & variable Good & consistent
Validate small
Slow & costly Fast & low cost
business case
Typically slow Fast
Quickly react to issues
& hence expensive & hence cheap
Level of automation Scattered at best Very high
42. Continuous Delivery is Not Magic
> It requires:
> time & money
> experts to set it up
> maintenance
> …and in return it brings:
> faster turn-around time
> options for the business
> severe stress reductions for IT
43. This Presentation
1 The Prelude
2 The Plan
3 The Project
4 The Pilot
5 Conclusion
6 Questions
44. 2012
Q1 Q3
Q2 Q4
Could we do a release, 1 week after the previous?
Uh, sure…
45. Initial Release Planning 2012
Q1 Q3
Q2 Q4
> Release 1:
> First Pilot release — 3 customers
> Full functionality
> Release 2:
> One week later — add 27 customers
> Release 3:
> Three weeks later — optimizations
> etc.
46. 2012
Q1 Q3
Q2 Q4
I know the first release is due tomorrow…
…but can we add something small ?
…that would be really nice!
No. But we could do an interim release 2 days later?
Excellent, thanks!
50. First Customer Reactions
It looks nice!
It looks similar to a webshop
Where is my purchase list?
I didn’t know you could offer all this!
What’s different? I will keep using the phone
51. Status @ now 2012
Q1 Q3
Q2 Q4
> Pilot customers
> Suppliers
> Business
> Incrementally adding customers
> Business leverage growing
> New initiatives – e.g.
52. Next Steps 2012
Q1 Q3
Q2 Q4
> Finetune storefront (×- & upsells, etc.)
> More business initiatives
> ATG Customer Service Center (CSC)
> Integrate ATG & marketing campaigns
> etc.
53. This Presentation
1 The Prelude
2 The Plan
3 The Project
4 The Pilot
5 Conclusion
6 Questions
54. Success Factors
> Mature technology (ATG & GigaSpaces)
> Local expertise (Mindcurv)
> Continuous Delivery
> Small team size
> Dedication & hard work
55. This Presentation
1 The Prelude
2 The Plan
3 The Project
4 The Pilot
5 Conclusion
6 Questions
56. New ATG site:
beta.delixl.nl
@delixl / @ernstdehaan
ernst-de.haan@delixl.nl