2018-1 Case Studies From Practice.fex

company tools
=====

challenges in companies:
	decide without complete facts
	meet impossible deadlines
	finish a half-our argument in one minute
	archive something with a team of disagreeing individuals
	development:
		ensure compatibility with other products
		maintain and extend a large code base
		deal with ambiguous/conflicting requirements
		question solutions which are "quick and easy"
	strategy:
		difficult to establish in large companies
		invest time to ensure all branches follow same strategy
		establish consensus about objectives & approach
	legacy:
		superseded SW/HW which is difficult to replace due to wide use
		old systems may be badly written, but not necessarily
		old systems battle-proven, therefore relatively stable
	assess reliability of statements:
		hearsay, alternative facts
		unjustified belief, opinion
		factually, statistically, justified opinion
		definition, proven theorem
	technical problem motivation:
		build a tree, from left to right
		start with technical problem
		describe business problem(s) it leads to
		describe possible solution(s) for each business problem

information formats:
	in general:
		present only short list of options, only structured data
		don't describe analysis approach / previous work
		answer "so what" / motivation as early as possible
	decision based presentation:
		provide base to make a decision
		present short list of options
		describe implications of these options
		provide single recommendation & its rationale
	take-away based presentation:
		max 3 key take aways
	
prepare presentation:
	find out about audience:
		talk to stakeholders in advance (find out view points, expectations)
		think about already brought up questions / issues
		consider conflicting interests from different stakeholders
	define objectives:
		work out what is expected to result from meeting
		define decisions to be taken, take aways to be remembered
		place all further material, math in appendix
	develop structure:
		possibly present initial situation
		top down (start with decision, then arguments, reasoning)
		or tell a story & align with examples
	create presentation:
		develop presentation based on storyline
		each slide one core statement with 3 min presenting time
		fast sketches by hand before spending time on perfecting them 
		not necessarily self-contained, focus on content, not design
	finalize:
		proofread on printouts, verify objectives are reached
	
decompose complex questions:
	issue tree (disaggregation):
		if given a problem ("why budget overrun")
		tree from problem statement to issues, sub-issues
		mutually exclusive, collectively exhaustive, relevant (MECERE)
		slow but steady approach, clearly see most important issues
		"budget overrun" -> unit cost too high; scope too big
		"any decision" -> benefits; cost; risk; strategic flexibility
		"MVP" -> full design/implementation -> functional, reliable, usable, UX
	option tree:
		if given a desired outcome ("deal with bad PM")
		adhere to CERE, ME is nice to have
		slash non-feasible, build business case for others
		easily compare available solutions
		"cost too high" -> "let people go", "stop business"
	hypothesis tree (hypothesis driven):
		if given an opinion ("budget overrun because bad PM")
		identify & collect arguments needed to prove hypothesis
		test hypothesis fast, for complex projects
		fast to do, but only if hypothesis is broadly correct
		"more budget bad" -> costs unclear; requirements impossible
		"reduce scope good" -> delivery time; cost; risk; easier to change
		
big projects:
	problems:
		incomplete, changing requirements
		insufficient user involvement with unrealistic expectations
		insufficient planning, resources & management support
	evaluate bad performing projects:
		base decision on how much more time / budget needed
		forget about sunk cost
	cost:
		solution deliver cost (design, realization, testing)
		ongoing activities (operation costs, maintenance)
	improve profitability:
		reduce cost (per unit / resource, of employees, of operations)
		increase revenue (more customers, sell more units, increase price)
	any (IT) decision:
		benefits (revenue increase, cost & risk reduction)
		cost (development, maintenance, operations)
		risk (technical, business)
		strategic flexibility (decision, production)
		implicit feasibility (higher cost & risk)
	
business tools
===

analyze company / products:
	SWOT:
		strengths (internal, helpful)
		weaknesses (internal, harmful)
		opportunities (external, helpful)
		threats (external, harmful)
	business model canvas results:
		key partners (interdependencies)
		key activities (focus of business)
		value propositions (customer value)
		customer relationships (communication, trust)
		channels (sell, communicate)
		customer segments (target group)
		revenue streams (billable services)
		cost structure (expenses)
		

	business case:
		define business problem:
			about the project & key questions to be resolved
			deliverables, objectives, outcomes
		identify target customer needs:
			where does it fit in customer journey
			how does customer do it now, how will it change
		outline benefits:
			address issues, ensure alignment to overall strategy
			increases in revenues, reductions in cost
			customer loyalty & perception changes
		estimate cost:
			total investment (effort + expenses), timeline & different options
			assumptions & likelihood of estimates
		identify risks:
			risk when doing projects vs when not
			technology risks, technology changes
			how to cover if over budget, time
		calculate returns:
			potential return (best/worst estimates)
			resources needed to maximize returns
		define success metrics:
			implications on key performance indicators (best/worst estimates)
			metrics to top-line (revenue) & bottom-line (profit) business gains
		
create new product:
	startup vs internal development:
		startup can scale up/down fast with entirely new mindset 
		is free from legacy business, more pressure for fast MVP
		but higher cost, knowhow transfer difficult, ownership of staff unclear
		internal can provide fully staffed, interdisciplinary teams
		has connection to customers for early feedback, field tests
	establish vision:
		develop vision of what to build before starting other things
		may externalize to startup to enable new thinking models
	interview customers:
		not asking for needs (only improved versions of products are recommended)
		asking for pains to discover real problems, focus product on that
	define target customers:
		large customers for large cash flow from only a few customers
		but have high negotiation power, long requirements list
		small customers easier to get to use MVP without reputation
		are easier understood, accept outsourced, clouded application
	calculate price of product:
		higher than production cost
		lower than customer willingness to pay (perceived benefit, alternatives)
	allocate resources:
		there can also be too much, scaling up / scaling down needs time
		at first start with a few people
		develop as a big team, then scale down again
	build MVP:
		functional, reliable, usable, emotional design
		build good first impression
	

problem solving tools
==========

iterative approach (redefine problem after analysis if needed)
be critical to proposed problem statements

problem statement:
	agree on question (establish shared focus, assure actionable results)
	define underlying problem (describe as-is, find drivers, causes & effects)
	find decision makers (key stakeholders & the ones who decide in the end)
	establish decision criteria (provide foundation to make decision)
	set solution constrains (ethical, moral, strategic limits to solutions)
	set solution scope (define out of scope items, like international units)
	
investigate:
	potentially build hypothesis tree to correct arguments, focus investigations
	analyze data, collect experience, interview stakeholders, read reports
	be respectful (mistakes happen, don't judge by degree/background)
	
analyze & structure problem:
	build issue tree to understand components of problem (what, how)
	think from business perspective (cost, risk, revenue, flexibility)
	make each problem component intellectually manageable / solvable
	build a common understanding, clarify priorities / responsibilities
	summarize key learnings, start over if problem definition needs review
		
synthesize results:
	define evaluation/success criteria (feasibility, costs, risks, payoff)
	ask "so what" / "what must be done", assume company perspective
	build MECERE option tree (always include do nothing)
	analyze options using facts (stating any additional assumptions taken)
	check implementation feasibility & risk, slash options not worth pursuing
	analyze viable options detailed (using success criteria, cost analysis)
	do technical evaluation (compatibility, availability)
	do functional evaluation (functionality offered)
	do integration evaluation (integrations needed)
	
formulate recommendation:
	propose best option, may goes beyond problem statement
	propose next steps to be undertaken	
	
dormakabra
====

sells locks

SWOT results:
	strengths (internal, helpful):
		design portfolio, identity, long-lasting customers
		experience, use of market potential, premium technology
		complete business offering (technology, firmware, installation)
	weaknesses (internal, harmful):
		development costly & customer driven, inefficient software platform
		local focus, no global price, no multiplication of output
	opportunities (external, helpful):
		system integration, parter distribution, home automation
	threats (external, harmful):
		rising complexity, partners collaborate with competition
		market saturation, growing competition
	results:
		usecase specific UI, package based offering, health check (opportunities)
		solutions instead of products, consulting, design, UX (against threats)

business model canvas results:
	key partners:
		home automation
		data center, development partners
	key activities:
		sales, ordering, shipping, repair
		support, communication, training
		operations, delivery, production
	value propositions:
		planing, easy install, integrations
		complete system status, access control
		continuous product improvement
	customer relationships:
		business infrastructure
		trust in brand and solution
	channels:
		local partners, installers
	customer segments:
		residential, small/medium, partners
	revenue streams:
		planning, hardware sales, installation, configuration, support
		cloud usage, maintenance, credential management, special solutions
	cost structure:
		sales, support, marketing, cloud operations
		process adaptation, new products, material
		
analysis results:
	established vision:
		provide parters & customers with access control management
		integrate planning, sales, ordering, CRM & support
	customer pains:
		needs to be AC expert, but high complexity & flawed technical support 
		fears proprietary solutions
		hates slow delivery time, low delivery quality
		fears breakdown of security

kaba exivo:
	shared vision to solve pains rather than detailed requirements
	develop new market:
		definition & setup of new market as difficult as development
		subscription plan for recurring revenue & high customer binding
		technology enabled new market, but internal restructuring needed
	develop new platform:		
		access control as a service
		for small practices (lawyers, doctors; need to comply with law)
		early & often usability tests with end users
		focus on business value rather than fancy technology
		develop internally instead in startup to avoid reintegration issues
	risks:
		high investment to until customer value generated
		need to balance brand value & early market release
		established business processes difficult to adapt
	timeline:
		envisioning (create core-team, establish vision, scope, requirements)
		start (scale up for MVP, agree on architecture, technical concepts)
		stabalize (scale down team, finish the project, freeze features / scope)
		operations (start production / maintenance / support)
	impact on stakeholders:
		employees need new skills (release cycles, security, user-centric design)
		partners need to adapt workflow to new platform
		customers need to be persuaded of new pricing, are more dependants
		managers must take new end-to-end responsibility, new procedures
		managers must manage touch points of new system with old business
		managers must support development approaches (fail early, new tools)	
		hr/recruiting must attract new employees from different industry
		sales needs to rethink measures of success

teralytics
=======

big data firm, collects usage of phone data

architecture:
	aggregate data:
		anonymize data (still in RAM)
		clean up noise
		estimate location based on events
		interpolate traces
		aggregate & extrapolate
		prepare results for dashboard
	dashboard:
		visualizes data
		records any query
		stop execution if too much info revealed
	VPN:
		to teralytics office (ISO certification)
		
provide data:
	aggregated by zip code, by weekday, by month, by day
	legal issues:
		set minimal bound according to regulations
		respect general data protection directive GDPR, country specific
		allow opt-out (usa with webpage, ignore foreigners)		
	hosting issues:
		AWS (very expensive due to constant, predictable load)
		therefore do hosting by yourself even if non-core business
		cluster at teralytics (easy physical access, faster connection)
		but legal issues EU vs CH, who pays cluster		
		cluster at peppermobile (responsible for security, maintenance)
		but slow processes, can't extend old machines
	pricing:
		calculate delivery of service costs (lower bound)
		calculate customer benefits, costs (higher bound)
		compare with other products
	trial concept:
		big enough to show benefit, but not to reveal everything 
		contracts to make usage illegal outside trial, be prepared to sue
	contract versions:
		fremium (more functions, services if customer pays)
		license based contract (pay per view)

hotelcard
====

hotel-halbtax
quite successful, offerings of all variety
customer buys member fee, hotels pay no commission

win-win:
	hotel:
		higher occupancy rate
		no booking portal commission
		more revenue from secondary services
		free publicity
		self-controllable availability 
	guest:
		overnights at half price
		large offering, high availability
		unlimited usage
		
market platform:
	supplier with overcapacity, clients with frequent usage
	sustainable high profit margin
	network effect:
		value for each participant increases
		competitors have hard start
	introduction strategies:
		letter of intent (promise to register if 150 parties found)
		partnerships with TCS, SBB, tamedia
		exclusivity (regional, limited in time)
	
sales strategy:
	sales partner like SBB, TCS
	offer:
		via company newsletter/magazine
		to clients, members, employees
		discount at first year, automatic renewal
	digital marketing:
		emails to clients/potential clients
		sales portals like DeinDeal
		SEO, SEA (+advertising)
	media 4 equity (M4E):
		tamedia takes 20%, pays off with spare advertising space
		various print & digital media
	product variants:
		personal hotelcard (double room booking)
		company hotelcard (transferable)
		vouchers (as a gift purchasable)	
	estimated market saturation:
		1 mio (+1 mio with surrounding countries)
		maximize renewal rate with newsletters, inspiration
		aquire new customers for less than 95 CHF

startups from ETH
=======

focus types:
	technology (AdNovum):
		programming languages, compilers
		software engineering, data modelling
		IT security
		information retrieval
	function (CodeCheck):
		marketing (big data, CRM)
		human resources (eLearning)
		finance (payments)
	industry (Avaloq):
		aviation, banking, clothing
		energy, health, construction		
	platform (Doodle):
		B2B, B2C
	
startup summary:
	leader:
		software guy, passion, stamina, fun
		is a role model for employees
		team builder & capable of learning
		focused and competencies in various areas 
	team:
		search for complementary skills & passion (even lawyers)
		discuss ideas openly with others
		define clear responsibilities
		team up engineering with sales & UX
		don't delegate recruiting to HR or keep under performing employees
	financial:
		we build it, they finance it
		we own it, they get the right to buy it
		preserve reputation of customer
		don't develop product without market
	process:
		have a stable vision but flexible strategy
		be patient, build your brand
		dare to make decisions
		pay attention to timing
		don't be shy or stubborn
	product:
		build high end software, platform, eShop, SAAS
		focus on specific product for specific, big market
		focus on urgently needed pain killer with highest value add
		find customers & smart plan for initial population (avoid investors)
		make it alternative-less, multipliable, high margin, scalable
		look for n:m market (even better if n:n, )
		multiply business model at other branches
		try to stay at service level
		don't be replaceable, too local, nice-to-have 
	sell:
		B2B easier than B2C, to business easier than to engineering
		peer2peer advertising
		don't spend unwisely on advertising

cost effectiveness and software metrics
======

productivity of software development:
	development:
		function points / person months inversed exponentials
		10 points / months at 100 points
		4 points / month at 2000 points
		2 points / month at 8000 points
		1 point / month at 16000 points, and decreasing further
		comparison applies within same company or controlled environment
	maintenance:
		fixed at 10%-15% of project development effort
		
function point (FP):
	metric to asses quality and productivity of software
	convert requirements to FP, then estimate cost with past metrics
	works well for comparable projects (same company, technology)
	usage:
		time (person months needed for large project)
		cost (total cost of project)
		schedule (time of completion)
		deliveries (intermediate releases)
	benefits:
		technology independence (high vs low power language accounted for)
		variability (to any kind of software)
		completeness (all activities measured)
		analysis (over multiple projects, technologies)
		estimate at start of project (unlike LOC)
	functional classes:
		external inputs (values in input mask)
		external outputs (report)
		internal logic files (states)
		inquiries (query to db)
		external interface file (like database)
	example external input:
		data element type (how many inputs)
		file type reference (how many groups)
		then use table to estimate FP
	
FP results:
	relations:
		1P is small code change, 100 LOC 
		10P is small application, 1k LOC, 5K
		100P is application, 12k LOC, 100K, 9months
		1k P is commercial, 100k LOC, 1m, >1y
		10k P for system, 1m LOC, 10m, 5y, 100people
		100k P for OS, 12m LOC, 1bio, 8y
	explanations why big is bad:
		implementation of new functionality more difficult
		large retesting efforts after each new change
		maintenance effort bigger (missing support, libraries stop working, ...)
		architecture degrades (unused code, now obsolete or unfit patterns)
		knowledge gets lost (employee leaves, forgets)
	low language levels:
		more logic mistakes per statement
		more LOC per function point
		more defects per function point
	effort changes with increasing FP (1 -> 10000):
		management effort (10 -> 16)
		error correction (15 -> 35)
		paper work (5 -> 31)
		coding (70 -> 18)
	risk changes with increasing (1 -> 10000):
		changing requirements (0 -> 45)
		poor project quality (0 -> 90)
		observed project delay (0 -> 80)
		project cancellation (0 -> 40)
	conclusions:
		productivity decreases with size of system
		synergies of projects must be high to be integratable in single product
		flexibility often bad (FP overhead but hard to predict)
		marginal requirements increase cost, risk exponentially

estimate new project:
	find efficiency at current FP level (FP/PM)
	calculate person months
	take square roots for team size / duration
	read out LOC / FP for the specific language
	if another language, compare LOC/FP and adapt needed PM accordingly
	if new language to learn, include performance reduction ~10%
	
AIP case study
====

want to replace old system, which has proven itself
currently most money spend on maintenance
	
AIP II:
	solves future problems with parametrization
	new architecture & new technologies
	but unrealistic:
		API was designed to be flexible as well
		huge risk & not doable in this time
		maintenance increases substantially		
		
AIP II shared:
	collaboration with another company
	productivity decrease through larger project
	but collaboration difficult:
		scope creep (incentive to add more of own functionality)
		difference in company culture (decision taking harder)
		system integrator in strong position (can play off companies)
	
AIP renovated:
	renovate AIP to remove duplication, architecture issues
	remove unnecessary functionality
	increase flexibility & parametrization at selected times
	include new business functionality
	
bad behaviour:
	from user:
		marginal additional features cause cost, time overruns
		requesting flexible solution instead of fixing requirements
		forgetting about increased complexity while trying to use synergies
		bias towards description of requirements by other people	
	from staff:
		forgetting about increased complexity while trying to use synergies
		building flexible solution, but not good in anticipating future requirements
		blindly fulfilling all requests (not in company's best interest)
		requesting rewrite instead of maintaining properly
	
conclusions:
	as few features as possible:
		all have to be maintained and increase complexity
		synergies with other projects have to be high to be useful
		flexible requirement connected to large overhead
		marginal requirements increase cost, time overproportionally
	communication:	
		failures due to lack of user involvement, business buy-in
		plan well, make assumptions explicit, keep up to date
	legacy systems:
		can often still fulfil business needs
		should run as short as possible at the same time as replacement 
		but avoid big bang replacements due to high risk
	requirement engineering hard:
		old staff left the firm
		documentation incomplete, misleading, motivation low
		source code only real requirement, but hard to analyze

digital strategy
==========

attack points:
	customer facing to increase revenue
	non-customer facing to reduce cost
	new products & services (like new possible products)
	new business models (like online shop)
	new customer needs (like cloud services) 

objectives:
	value created/delivered with digital channels
	key performance indicators KPI to measure success of strategy
	
roadmap of initiatives:
	potential digital initiatives with their respective impact & complexity
	conclude coordination needed, prioritization or projects
	
organization:
	governance structure to enable timely delivery
		
workflow:
	framework & inventory:
		analyze digital portfolio (past, present, future)
		gather end-user expectations & pain-points
		understand the value chain & customer journeys
		do interviews with employees & clients
		use as-is assessments, value chains, customer journeys		
		value chain:
			how company creates value to its customers
			analyses departments such as communications, marketing, product, customer care
		detailed persona:			
			personification of a typical customer
			socio-demographic information (age, marriage, location, occupation, salary)
			personal quote (describing individual situation)
			motivations (work, learning, social, shopping, fun)
			expectations (list of needs, pains), the core information!
			behaviours (usages of devices, apps, websites, spare time)
			influences (of other products, services, peoples)
		customer journey:
			analyze customer touchpoints with brand/product before/after purchase using persona
			discover (push info, ads / friends), explore (pull info, call / webpage), buy (purchase), engage (maintenance)
			collected touchpoints describing user actions & channels, emotional response
			emotional responses can be frustration, anxiety, satisfaction, happiness
	digital benchmark:
		market analysis & trend forecasting
		benchmark current experiences & get ideas from competitors
		assess real & perceived limitations
		do workshops, project team research
		use benchmark templates
		benchmark template:
			describe idea & solution, assess benefits for target group				
			describe quantified benefits for company (internal) & customers (external)
			estimate cost & feasibility
			place project in framework (impacted steps for target group)
	ideation:
		organize, participate in ideas generation effort
		classify ideas based in defined framework
		select ideas based on strategy, feasibility, expected ROI
		do workshops, project team research
		use priorization quadrants, business cases
		solution targets:
			web, eCommerce, mobile app, sale points, call centers
			collect initiatives for all solution targets
		priorization quadrants:
			grid with (technical complexity, customer value) as axis
			foundational (low, low) critical for business
			quick wins (low, high) for first priorities
			optimizations (high, low) for probably not valuable ideas
			future end state (high, high) for long-term initiatives
	roadmap:
		assess, organize selected ideas into reasonable, feasible projects
		quantify projects
		create actionable roadmap
		do final workshops, project team analysis
		use prioritized portfolio
				
breitling:
	setting:
		product:
			price to different socio economic background of clients
			product features important
			price compromised of value associated & material value
		market:
			informed buying, decision emotional
			complex market with different brands, models
			customers often online, susceptible for input from environment
	non-digital customer journey:
		discover:
			advertisement seen (not personalized)
			talks to friends (small amount if people reached)
			visits shops (only can try out part of collection)
		explore:
			tries on watch (found by chance)
			talks to wife (printed, not personalized catalogue)
			studies brochure (no filtering)
		buy:
			buy model (purchase rethought if model not available)
		engage:
			maintenance (no news from brand)
	digital solutions:
		online catalogue:
			filtering & recommendation system
			responsive design
			availability of watches at the retailer stores
			benefits:
				personalized catalogue to customer
				insight on customers preferences
				better control on authorized retailers
			cost drivers:
				high technical feasibility of catalogue
				recommendation system
				integration with retail systems
		online community:
			sharing of information & opinions platform for customers
			co-creation process participation
			direct link to customer care for support / news
			benefits:
				informed, structured customers under the influence of the brand
				improved customer loyalty
				direct link to online community
			cost drivers:
				moderation, customer relationship management
				integration with customer care & product development
		augmented reality:
			mobile app to try on watches
			connected to online catalogue
			benefits:
				try-on of models without having to visit local shop
				more visibility of products for customer
			cost drivers:
				new technology (make or buy)
				3d high definition models of all watches needed
	digital journey:
		discover:
			online advertisement (personalized)
			forum visit (sees comments, opinions, live chat)
			schedules meeting online (guided to breitling, model available)
		explore:
			sales app (clerk shows not available models)
			breitling app (oriented to different decision makers)
			augmented reality (helps visualize)
		buy:
			online reservation (model ready, end-to-end support)
		engage:
			continuous usage of breitling app (interactions, continuous news)
			
deliotte
====

engagement approach:
	gather data
	incident containment and recovery
	analysis
	prevention & planning
	post-incident review/delivery

incident prepare:
	incidence will happen, better be prepared for it
	escalation criteria:
		scope (#users, #customers)
		criteria (tier 4-1 assets, recoverability, impact)
		score / criteria determine step to escalate on
	escalation ladder:
		IT help desk
		local/operational response
		cyber incident response team
		emergency management team
		executive committee
		board of directors (are informed, but don't act)
	cyper security for C level:
		CxO is held accountable
		CEO faces risk of complete business disruption
		CFO faces risk of significant losses & high recovery costs
		CIO needs to ensure IT runs smoothly
		chief strategist may loses strategic plans (aborted acquisitions, ...)
		head of marketing has to ensure brand is not abused
		general council concerned with lawsuits, IP protection, prosecution

incident response:
	gather emergency management team:
		physical security (law enforcement)
		information technology (internal IT)
		information security (security specialist)
		communications (PR)
		legal & compliance (lawer)
		chief of staff
		human resources
	containment/eradication:
		short term (kill threat asap, preserve evidence, assess/contain impact)
		long term (learn from incident, mitigate risk of recurrence)
		prepare for breach notification
		e.g. disconnect compromised computers, block malicious communication
	recovery:
		get assets operational again, monitor closely
		evaluate containment plan, possibly refine
		inform internal/external, compile report
		coordinate post-breach actions
		e.g. verify, update, restore systems, document decisions
	post-breach:
		incident triage compromise 10% of impact
		stop attack, communicate, assess impact within days
		fix infrastructure, legal issues, manage relationships within a year
		repair business processes, invest in defence within years
		
incidence good practice:
	preparation:
		develop standardized approach to incident response (policies)
		ensure consistent results prior, during an incident (playbook)
		training, awareness, preemptive / reactive controls (discipline)
	detection:
		identify, validate, report incidents
		confirm incident type & initial classification
	analysis:
		determine mechanism, root cause, scope, scale, impact
		develop incident remediation plan
	containment/eradication:
		contain, eradicate, recover from incident, notify stakeholders
		reduce effects, prevent escalation, ensure business as usual/service level
		preserve evidence, document actions & timeline
	reporting:
		produce post incident report
		include all activities undertaken, lessons learned
		
cyper-attack:
	visible impacts:
		deterioration of public relation
		costs of technical, legal cleanup
		costs caused by disruption of services
		post-breach protection/cyber security improvements
	invisible impacts:
		costs of regulatory changes (GDPR)
		loss of IP
		loss of potential clients, devaluation of brand
		increase of insurance fees
		organisational changes when executives change
	
credit suisse
====

about:
	global wealth manager, big investment banking sector
	want to be more flexible, global
	
history of banking:
	1960:
		use of electronic booking machines in branches, ATM's
	1970:
		centralized computing, automation of processes
	1980:
		widespread ATM, new products like LSV
	1990:
		electronic exchanges, internet based products
	2000:
		online banking, 7x24 processing, large # of products
	2010:
		mobile banking, algo trading, blockchain technologies
		
history of banking systems:
	1980:
		several thousand users sharing platform
		single computer handles all transactions
		batch processing to distribute payload
		highly tuned applications enable high transaction volume
	evolution:
		no more staff for manual transactions
		more products, automation, integration, requirements
		explosive growth in IT staff, unknown with whole architecture
		continuous tweaks of architecture, experienced staff leaves
		first attempts to replace systems fail
		new complex interfaces, must features, failed migrations
		can't find new staff
	current status:
		legacy applications both huge asset & burden
		40 year old applications running on newest mainframe technologies
		fit for purpose, highly scalable, rich applications
		but difficult to manage (high complexity, lack of qualified staff)
		lot of lost knowledge (business processes, requirements, code)
		not fit for real-time processing, micro service
	
credit suisse setup:
	swiss banking IT platform (SBIP)
	shared with other divisions of credit suisse group
	large set of functionalities, old
	history:
		2003 SOA architecture
		2006 One Bank (consolidation in SBIP)
		2015 decoupling of security processing from SBIP
		since 2016 redesign
	problems with migration:
		analysis takes to long (results already outdated)
		limited people with strong business / IT skills to architect complex systems
		reverse engineering impossible through continuous changes
		design deadlock because of too many dependencies
		persistence on detailed upfront analysis
		too high risk of implementation, migration
		focus lost after restructuring of organisation
		unable to obtain budget, too low priority
		market events driven cost/investment rationalization
	optimized mainframe:
		rich business functionality, still considered "fit for purpose"
		high scalability (optimized code, timely distributed load)
		highly integrated, single instance system
		but decreasing supply of engineers, high platform cost
	successor of mainframe:
		must have same rich business functionality
		but be better platform
	parallel distributed system approach:
		high scalability (distribution, parallelization)
		business agility through segregated micro services
		adapt new technologies gradually
		low platform cost

migration:
	to stay relevant & competitive in fintech
	rewrite (for new language, platform, apis; but not future ready)
	replace (off-the-shelve with customizations, redesign business)
	transform (green-field, reinvent business)
	drivers:
		expensive hardware, software licenses
		not aligned to IT or business perspective anymore
		high operational risk, production stability, change efforts
		end of life of architecture, technology
		skills & people risks
		new approaches (DLT, ML, micro services)
	challenges:
		people (psychological bias against change, office push/pulls)
		complexity (high volume, complex interfaces/integrations/maintenance)
		financial (significant cost increase, multi-year commitments)
		legacy knowledge (lost & forgotten, old philosophy & technology)
		project management (different stakeholders, moving target, large project)
	new technical approaches:
		utilities (existing standard products)
		distributed ledger technology (DLT) to ease communication
		machine learning, artificial intelligence for big data problems
		micro service architectures to replace monoliths
		agile development for faster execution, validation of requirements

security settlement engine (SE):
	manages delivery, receipt of securities (dematerialized & physical)
	35 years old, some automation, very efficient
	mission critical application, handling large volume of trades & events
	problems:
		multifunctional components with minimal logging (design)
		dead code, monolithic synchronous batch processes (technical)
		high time to market, it just works, high post-incidence efforts (business)
	future DLT platform proposal:
		replace central security depository CSD with DLT
		CORDA as shared infrastructure with banks, CSD, other users
		need consensus on ownership, governance, standards, investments, timeline
		need business case with benefits, cost, transition risk
	general approach:
		can't replace big bang, too risky
		therefore introduce interface for new SE into other applications & old SE
	approach:
		consolidate, standardize existing interfaces, solve bottlenecks
		migrate database from IMS (hierarchical) to DB2 (relational)
		decommission end-of-day processing, old business processes
		start real-time processing
	vision:
		settlement as a service, easy integration of new clients
	next steps:			
		micro-service architecture to replace mainframe step by step
		but needs clearer interfaces, capability to onboard new regions / users
		utilities for customization, integration into existing systems
		but needs fast, cheap way to build integration layer with current system
		DLT to replace middle man, messaging
		but needs sync between DLT & existing database
	challenges:
		business challenges (network effect, get & align goals of participants)
		technical side (new & old coexist, DB synchronization, many interfaces)