Learning about relational data models and databases

The other day I did a mini-course online about relational databases with Stanford Online. The course is taught by Stanford Professor Jennifer Widom and is aimed at relative database novices like me.

Jennifer started off the course by providing some useful characteristics of a Database Management System (‘DBMS’):

“Efficient, reliable, , convenient and safe multi-user storage of and access to massive amounts of persistent data”

  • Massive scale – Jennifer talked about databases with the potential for terabytes of data.
  • Persistent – The database always outlives the programme that will operate on its data.
  • Safe – Power or hardware failures for examples shouldn’t affect the database.
  • Multi-user – Jennifer talked about concurrency control in this respect.
  • Convenient – The convenience of DBMS comes for a large part from ‘physical data independence’, i.e. operations on the data are independent of the way in which the data is laid out.
  • Efficient – This relates to database performance, e.g. thousands of queries/updates per second.
  • Reliable – Jennifer pointed out that with databases 99.999999% up-time is critical.

Jennifer then went on to break down the structure of a database into to its most basic elements, something which I actually found quite helpful:

  • Database is a set of named relations (or tables)
  • Each relation has a set of named attributes (or columns)
  • Each tuple (or row) has a value for each attribute
  • Each attribute has a type (or domain)
  • A database schema is a structural description of relations in a database
  • An instance contains the actual contents at a given point in time
  • “Null” is a special value for “unknown” or “undefined”
  • A key is an attribute whose value is unique in each tuple (e.g. a student ID) or a set of attributes whose combined values are unique
  • Having unique keys helps to (1) identify specific tuples (columns) (2) fast indexing of the database and (3) refers to tuples of another key

In the session, Jennifer gave a first flavour of how to create relations (table) in SQL, which is commonly used database programming language. I found below example which Jennifer gave to be very helpful:

Create Table Student (ID, name, GPA, photo)

Create Table College (name string*, state char*, enrolment integer*)  * These are all attribute types

She then went on to talk a bit about querying relational databases, outlining common steps in creating and using a (relational) database:

  1. Design schema – Create a schema which describes the relations within a database, using a Data Definition Language (‘DDL’)
  2. “Bulk load” initial data – Load the initial data into the database
  3. Repeat – Execute queries and modifications

Jennifer then finished off by giving an example of a SQL query that returns the follow relation: “IDs of students with GPA > 3.7 applying to Stanford”. In SQL this query would look something like this:

Select Student ID

From Student, Apply

Where Student ID = Apply ID

And GPA > 3.7 and college = ‘Stanford”

Main learning point: I’ve clearly got my work cut out for me, but I felt that this first mini-course on relational data models and databases was incredibly helpful and easy to understand. I’m hoping I can build on these first foundations and understand more about how to best structure SQL queries and interpret the relations that these queries return.


Mr & Mrs Smith’s personalisation drive

I’m pretty familiar with recommendations and personalisation in media; companies like Spotify and Netflix specialise in providing their users with relevant content based on their -implicit and explicit- preferences. The other day, I found out about luxury online hotel booking site Mr & Mrs Smith starting to provide more personalised content to its customers. I looked into what this means for a business like Mr & Mrs Smith and for its (target) customers:

  1. Why? – Mr & Mrs Smith wants to provide its customers personalised communications and experiences, tailored to the interests and preferences of the individual customer. The hotel booking site wants to create a single of view of its customers, across its multiple channels. Mr & Mrs Smith’s main channels are its website, its app, email and its call-centre. A more customised approach will help Mr & Mrs Smith in targeting its communications and in optimising the opportunities to sell hotel bookings to its customers.
  2. What? – I’m not a Mr & Mrs Smith customer myself, but I can think of a few scenarios where personalisation could be pretty useful to both the customer and to Mr & Mrs Smith as a business (see Fig. 1 below). I’d love to find out how Mr & Mr Smith currently go about aggregating all information for its individual customers into unique profiles.
  3. How? – Mr & Mrs Smith will use a solution by a company called Sailthru which will help the company to better understand customer interactions at multiple digital touch-points. Tamara Heber-Percy, co-founder and CTO of Mr & Mrs Smith, explains that “we appreciate that personalisation is about people, so we’re confident that our partnership with Sailthru will assist us in truly understanding our customers’ interests and ultimately in creating intelligent, tailored messages especially for each of them”.

Main learning point: Personalisation by Mr & Mrs Smith makes total sense. Targeting customers with personalised messages and offers is very logical from both a business and a customer experience perspective. This approach – and the subsequent customer insights that it will return – will no doubt give Mr & Mrs Smith the customer understanding required to tailor its marketing efforts and to predict future customer preferences.

Fig. 1 – Potential scenarios with regard to personalisation for Mr & Mrs Smith customers

  • A single customer profile and targeted campaigns- As a customer I don’t wanted to be spammed by Mr & Mrs Smith with big blast type mailing lists. I only want to receive communications about offers for hotels that could be of interest to me. I expect Mr & Mrs Smith to be smart about my previous bookings. For example, if I’ve previously booked child-friendly hotels through Mr & Mrs Smith, I don’t expect to be sent lost of info on Honeymoon Hotspots.
  • Personalised recommendations – If I opt-in to receive push notifications from Mr & Mrs Smith on my mobile, I expect these messages to be relevant and tailored to my interests. For example, Mr & Mrs Smith make it more attractive for me to engage with these messages by offering a window during which a discount for a specific hotel is available. I can imagine that Mr & Mrs will be particularly interested in the feedback loop related to its personalised recommendations. For example, if I don’t open the Mr & Mrs Smith email with “Discounts on all our Maldives hotels” what will Mr & Mrs Smith infer from this implicit behaviour?
  • Targeted messaging and editorial – I can imagine that Mr & Mrs Smith differentiate in the way they approach customers who maybe book 1 hotel per year compared to its “Goldmsith” members, who pay £400 per year for specific benefits. The type, frequency and copy of communications will vary per customer segment.
  • Personalised service – When I phone up Mr & Mrs Smith’s call centre with a question, I expect the person at the other end of the line to have all my information at hand. The person helping me with my query knows about my previous bookings through Mr & Mrs Smith, my preferences and the current booking that I’m calling about.

Related links for further learning:

  1. http://static2.thedrum.com/news/2014/02/26/mr-and-mrs-smith-looks-drive-personalised-marketing-communications-sailthru
  2. http://www.retailtechnology.co.uk/news/5336/boutique-hotel-booking-group-uses-automated-personalisation/
  3. http://ukhotelexperts.co.uk/2011/is-there-anyone-out-there-who-really-loves-the-customer
  4. http://www.luxurybranded.com/mr-and-mrs-smith-interview/

Mr & Mrs Smith



EDITD and applying big data analytics to the fashion industry

Since I went to a talk about visual search of fashion products, I’ve been keen to find out more about how the fashion industry uses big data analytics to make product decisions. I then came across EDITD, which is a real-time fashion analytics company based in London. Given my passion for both fashion and data, I thought I’d have a closer look into what EDITD do:

  1. EDITD’s mission – EDITD’s overall mission is to “help the world’s apparel retailers, brands, and suppliers deliver the right products at the right price and the right time”. Given the fast pace nature of the fashion industry decisions about product planning and the right amount of stock are absolutely critical. Julia Fowler, co-founder of EDITD, gives a good example when she explains that “today, an EDITD user can simply run a query on cardigans, for example, and receive results in under a second. More than 50 million SKU (Stock Keeping Units, MA) are tracked by the system.” I came across another good example in EDITD’s UK lingerie market retail calendar which aggregated data on new arrivals, discounts and sellouts can help merchandisers planning timing and location of their stock (see Fig. 1 below). It also helps navigate promotional activity and discounting.
  2. EDITD’s product – It was interesting to see what kind of features are included in EDITD’s product offering (see Fig. 2). I read in article in Fortune that EDITD’s dataset includes 53 billion data points on the fashion industry dating back more than 4 years. The Fortune article also mentioned that EDITD’s data covers more than 1,000 retailers across the globe. The way in which EDITD aggregates all this data through its different features (see Fig. 2) is where the main value of using EDITD’s services comes into play.
  3. Tangible benefits – Earlier this year, fashion retailer Asos said that using EDITD led to a 37% revenue increase in the last quarter of 2013. This was due to the data insights provided by EDITD which helped structure Asos’ pricing competitively. Geoff Watts, EDITD’s CEO, told The Guardian that the main value for Asos from using EDITD came from using their insights to make informed buying decisions grounded in data. “Retail on a basic level is all about buying the right things, so getting that right and making sure you’re selling the right product at the right price is really what dictates your success,” Watts said. Maria Hollins, Asos’ retail director, echoed this and stressed the importance of Asos making the right decisions faster than their competitors.“At ASOS, being first for fashion means being always competitive and having just the right assortment,” she said. “We’re using Editd every day to help us make critical buying and trading decisions”. If anything, EDITD saves retailers and brands from having to do so-called “comp shopping”, having to spend time going to competitor sites and stores to buy their products and compare prices. Instead, through EDITD, people can see at a glance what the competitive price points are (see Fig. 4 below).

Main learning point: I can see why brands and retailers are keen to use EDIT’s data tools and insights on a daily basis. The data on fashion and apparel has been aggregated and presented in such way that it accommodates fast decision making. I was particularly impressed with what I’ve seen of EDITD’s front-end dashboard and the way in which its purpose built product tracker provides real-time market visibility.

Fig. 1 – EDITD’s UK lingerie market retail calendar – Taken from: http://editd.com/blog/2014/10/timing-why-uk-lingerie-market-blooms-in-may/


 Fig. 2 – EDITD’s main product features – Taken from: http://editd.com/product/features/

  • Market Analytics – EDITD offers a product tracker built specifically for the fashion and apparel industry. The tool provides real-time market visibility, analysis of new stock and discount activity, entry and exit prices and number of options in stock which enables retailers to benchmark their performance against each brand or retailer, providing insights into market positioning.
  • Retail Reporting – EDITD offers daily and week reports on what’s selling the fastest and the latest trends in new arrivals.
  • Visual Merchandising – EDITD has an archive of newsletters, blogs and webpages with real-time updates for brands and retailers across the whole market, worldwide. The idea is that every communication with customers is captured, to help users find discount cycles, product trends and themes, and understand their retail cadence.
  • Trend Dashboard – EDITD’s real-time tracking monitors trend progress, and historic data shows performance and trajectory. This data is all captured in EDITD’s Trend Dashboard (see Fig. 3 below).
  • Runway & Street – If you want to get a better sense of emerging trends straight from runway shows or the ‘streets’, EDITD provides visual reports on new fashion and apparel trends to look out for.
  • Social Monitor – EDITD has a Social Monitor which combines the knowledge of over 800,000 thought-leaders, key influencers and fashion experts providing an instant source of inspiration and insight into the hottest trends and opinions.

Fig. 3 – Screenshot of EDITD’s Trend Dashboard – Taken from: http://editd.com/product/features/

EDITD trend-dashboard-1

Fig. 4 – Screenshot of EDITD’s front-end providing competitive insights – Taken from: http://menapparelonline.com/blog/fashion/fashion-data-tool-editd-helps-asos-push-revenues-up-37-the-guardian/


Related links for further learning:

  1. http://fortune.com/2014/09/22/fashion-industry-big-data-analytics/
  2. http://www.theguardian.com/technology/2014/jan/30/fashion-data-tool-editd-helps-asos-push-revenues-up-37
  3. http://www.wgsn.com/en/
  4. http://menapparelonline.com/blog/fashion/fashion-data-tool-editd-helps-asos-push-revenues-up-37-the-guardian/
  5. http://editd.com/blog/2014/10/timing-why-uk-lingerie-market-blooms-in-may/

Using data to inform product decisions

A few months ago, I delivered a talk to an audience of product managers about the importance of having a data informed approach (see Fig. 1 below). As a product person, using data optimally to help make product decisions is critical. It’s something that I try to work on every day. I’m constantly learning about things such as the best combination of quantitative and qualitative, measuring the right things and using data as an integral part of the product development process.

This is a quick summary of the key things that I talked about in my talk in May of this year:

  1. Why do we need data? – I’m always keen to stress that as a product manager I don’t have all the answers. Product managers aren’t the holy grail and I believe it would be silly to pretend otherwise. I’m never afraid to say that I don’t know when people ask me which product idea we should go for or how people will use our product. Instead, I’ve learned to use data to draft and test assumptions which can help to inform product decisions (see Fig. 2 below).
  2. What can quantitative data tell us? – Quantitative data can really help to get stats on how people actually use a product and measure whether our product improvements or new features have the desired impact. I’ve learned a lot from the book Lean Analytics in which its authors, Alistair Croll and Benjamin Yoskovitz, expand on when and how to best use quantitative data.
  3. What can qualitative data tell us? – Qualitative data can be very valuable if you want to find out about the “why” behind quantitative data and to get a better insight into what users think and feel. Also, in cases where you don’t have much quantitative data at your disposal, qualitative data can help you to get some quick input into a product idea or prototype.
  4. Data driven approach – There are quite a number of game companies such as Zynga and Wooga that apply a data driven approach to product development. This typically means that a company will pick a single or a set of key metrics to concentrate on (see Fig. 3 below). With a purely data driven approach, its data that determine the faith of a product; based on data outcomes businesses can optimise continuously for the biggest impact on their key metric.
  5. Data informed approach – A few years ago, I came across an inspiring talk by Adam Mosseri, Director of Product at Facebook, who introduced the term “data informed” product development. The main rationale for this data informed approach is that, in reality, data is only one of the factors to consider when making product decisions. I make the point that typically, data is most likely to play a role alongside other decision-making factors such as strategic considerations, user experience, intuition, resources, regulation and competition (see Fig. 4 below). This doesn’t diminish the critical nature of data, but it does into take account a reality where other factors need considering when making product decisions.
  6. Where and why to use data to help inform product decisions – Also, there are a number of cases where a purely data driven approach falls short. For example, when there’s a strategic decision to be made or when you’re assessing a new product idea (see Fig. 5 below), looking at data in isolation may be insufficient. In my talk, I also gave some examples of how and why I use data at set points of the product lifecycle to help inform decision making (see Fig. 6 below).

Main learning point: I’m looking forward to a book by Rochelle King about data-informed design, hoping to learn more about how and when one can use data to inform product design decisions. I feel like I’ve already learned a lot from people like Adam Mosseri, Alistair Croll and Benjamin Yoskovitz and the way in which they use data as a key factor in making product decisions. Ultimately, I believe that using data continuously – in whichever way or form – is the best way to figure out how to best utilise it.

Fig. 1 – My presentation “Use data to inform product decisions” at ProductTank Hamburg on 16 May 2014 

Fig. 2 – Why do we need data?

  • To learn about products and users
  • To measure success
  • To choose between options
  • To understand user interaction
  • To decide on our products

Fig. 3 – Key components of a data driven approach

  • Focus on the “One Metric That Matters”
  • Build hypothesis around key KPI
  • A/B or multivariate test continuously
  • Optimise your product based on data
  • Are we making a noticeable difference?

Fig. 4 – Factors that can play a role in a product decision-making process (this is by no means an exhaustive list!)

  • Data
  • Strategy
  • Intuition
  • Competition
  • Regulation
  • Business
  • Brand
  • Time
  • Technology

Fig. 5 – Reasons a data informed approach might be better suited to your decision-making process

  • Data is one of the factors to consider
  • Focus on the questions that you want answered
  • You can’t replace intuition or creative ideas with data
  • Assess impact on relevant areas

Fig. 6 – Examples of how I use to data to inform product decisions

What do we want to do?

Assumptions, hypotheses, assessments and prototypes

How should it work?

User testing, user stories, A/B testing and prototypes

How is it working?

Product retrospectives, tracking and goal-oriented planning

Related links for further learning:

  1. http://andrewchen.co/2012/05/29/know-the-difference-between-data-informed-and-versus-data-driven/
  2. http://vimeo.com/14999991
  3. http://venturebeat.com/2013/12/04/the-4-keys-to-running-a-data-driven-business/
  4. http://www.realityisagame.com/archives/390/wooga-follows-zynga-in-metrics-driven-game-design/
  5. http://insideintercom.io/the-problem-with-data-driven-decisions/
  6. http://www.webdesignerdepot.com/2013/05/the-perils-of-ab-testing/
  7. http://www.fastcolabs.com/3033885/ebay-is-running-its-own-sociology-experiments
  8. http://www.beyondwebanalytics.com/2013/07/24/episode-67/
  9. http://www.slideshare.net/tgwilson/a-process-for-being-data-driven

Book review: “Inspired: How To Create Products Customers Love”

Marty Cagan is a legend. Period. He’s a legend in the world of product management anyway. I’ve been to a number of events over the past few years where Marty either was a keynote speaker or acted as a ‘judge’, reviewing the product pitches of fellow product people who nervously presented to this ‘product guru’.

What makes Marty Cagan such a phenomenon? The answers can be found in his book “Inspired: How To Create Products Customers Love” which he first published in 2008. This book was probably one of the first ones in its league; providing product people with practical tips on how to create products that customers love. Since then there have been an awful lot of books on product management and developing engaging products, but a lot of that started with “Inspired: How To Create Products Customers Love”.

These are some of the things that I’ve learned from this book and which I’ve been able to put in practice over the past couple of years:

  1. Great products don’t happen by accident – Even if you stop reading Marty Cagan’s book after about 10 pages, you’ll have already picked up on Marty’s ten learnings about how to create inspiring products. His main point is that great products never happen by accident, but ‘by design’ and he’s listed some great points which underpin this point (see Fig. 1 below). According to Marty, the product manager has two key responsibilities in this respect: assessing product opportunities, and defining the product to be built.
  2. Building the right product vs building the product right: Marty makes a great point when he states that “the product manager is responsible for defining the solution, but the engineering team knows best what’s possible, and they must ultimately deliver that solution.” In Chapter 5 of “Inspired: How To Create Products Customers Love” Marty suggests three practical ways of working closely with engineers to develop a better product (see Fig. 2 below). Equally, as a product person I can help engineers by keeping their focus on a ‘minimal’ product; don’t define the ultimate product but the smallest possible product that will meet your goals (you can find an example here where I created an MVP version of my own product last year).
  3. Recruiting good product managers – In Chapter 6 of the book, Marty outlines key qualities to look for when hiring a product manager. I compared my list of important qualities to Marty’s list of personal traits and attitude, which naturally is quite a bit longer (see Fig. 3 below). The one thing that I’d like to hear more from Marty about is about communication and related soft skills. In his brief section on communications, Marty mostly talks about writing clear ‘prose’ and about more generic presentation skills. I would, however, love to know how Marty typically deals with difficult stakeholders (beyond ‘managing up’ as described in Chapter 10) getting people to buy into an idea or investment. Also, how does Marty approach people in coffee shops or on the streets for some ‘gorilla’ user interviews? It would be great to have Marty zoom in on these (soft) skills in more detail.
  4. Assessing product opportunities – One of the main things that I learned through Marty a few years ago was the practical value of his so-called ‘opportunity assessment’ (see Fig. 4 below). Even if you don’t ask yourself all the questions raised in this template, it provides a very efficient way to quickly assess a product opportunity. The thing I like most about Cagan’s opportunity assessment is that it focuses on the user or business problem that you’re trying to solve and not the particular solution that you have in mind. For example, when I assessed the opportunity for my own product last year I concentrated on the problem that my app was trying to solve and not so much (initially) on the actual shape that my app was going to take.
  5. Frame your product decisions – In Chapter 13 of the book, Marty stresses the importance of properly framing the (product) decision to be made, and to get everyone on the same page in terms of the decision frame. I’ve learned how easy it can be to lose sight of key decision factors or to think that everyone is on the same wavelength (whilst they’re not). The points outlined in Fig. 5 below can help massively in identifying the different aspects of the decision that you’re looking to make and in figuring out the data required to help make the decision.

Main learning point: “Inspired: How To Create Products Customers Love” contains a great deal of specific pointers on how to develop engaging products and how to best create a product-oriented culture. Whether you’re completely new to product management or have been doing it for a number of years, I’m pretty sure you’ll be able to learn something from the absolute master that is Marty Cagan!

Fig. 1 – Creating ‘Great Products by Design’ (from: Marty Cagan – Inspired: How To Create Products Customers Love)

  1. The job of the product manager is to discover a product that is valuable, usable and feasible.
  2. Product discovery is a collaboration between the product manager, interaction designer, and software architect.
  3. Engineering is important and difficult, but user experience design is even more important, and usually more difficult.
  4. Engineers are typically very poor at user experience design – engineers think in terms of implementation models, but users think in terms of conceptual models.
  5. User experience design means both interaction design and visual design (and for hardware-based devices, industrial design).
  6. Functionality (product requirements) and user experience design are inherently intertwined.
  7. Product ideas must be tested – early and often – on actual target users in order to come up with a product that is valuable and usable.
  8. We need a high-fidelity prototype so we can quickly, easily, and frequently test our ideas on real users using a realistic user experience.
  9. The job of the product manager is to identify the minimal possible product that meets the objectives – valuable, usable, and feasible – minimising time to market and user complexity.
  10. Once this minimal successful product has been discovered and validated, it is not something that can be piecemealed and expect the same results.

Fig. 2 – Three ways to use your engineers to come up with a better product (from: Marty Cagan – Inspired: How To Create Products Customers Love)

  1. Get your engineers in front of users and customers
  2. Enlist the help of your engineers in exploring what’s becoming possible as technology develops
  3. Involve your engineers (or at least a lead engineer) or architect from the very beginning of the product discovery process to get very early assessments of relative costs of the different ideas, and to help identify better solutions

Fig. 3 – Personal traits and attitude of great product managers (from: Marty Cagan – Inspired: How To Create Products Customers Love)

  • Product passion – Having a love and passion for good products
  • Customer empathy – Being able to understand the problems and needs from your target audience
  • Intelligence – Intelligence is hard to measure, but problem solving is definitely an important trait to look out for
  • Work ethic – The focus here is on the level of responsibility that comes with being a product manager
  • Integrity – Being able to build up a relationship of trust and respect with the people that you work with
  • Confidence – Confidence is a critical ingredient when it comes to convincing others to invest their time, money or effort into a product
  • Focus – Have the ability to keep the focus on the key problem to be solved at any given moment
  • Time management – Being able to distinguish between what’s important (and why) and what’s urgent
  • Communication skills – Writing clear and concise prose, get ideas or points across clearly and succinctly
  • Business skills – Understand the business aspects of your proposition, market, etc. Being able to understand and speak business language and concepts

Fig. 4 – Assessing product opportunities (from: Marty Cagan – Inspired: How To Create Products Customers Love) 

  1. Exactly what problem will this solve? (value proposition)
  2. For whom do we solve that problem? (target market)
  3. How big is the opportunity? (market size)
  4. How will we measure success? (metrics/revenue strategy)
  5. What alternatives are out there now? (competitive landscape)
  6. Why are we best suited to pursue this? (our differentiator)
  7. Why now? (market window)
  8. How will we get this product to market? (go-to market strategy)
  9. What factors are critical to success? (solution requirements)
  10. Given the above, what’s the recommendation? (go or no-go)

Fig. 5 – Framing product decisions properly and getting everyone on the same page in terms of (from: Marty Cagan – Inspired: How To Create Products Customers Love) 

  • What problem exactly are you trying to solve?
  • Who exactly are you trying to solve this problem for – which persona?
  • What are the goals you are trying to satisfy with this product?
  • What is the relative priority of each goal?

Related links for further learning:

  1. http://www.svproduct.com/assessing-product-opportunities/
  2. http://www.svpg.com/the-power-of-milestones


What to expect from Amazon’s anticipatory shipping patent?

Earlier this year Amazon patented its “anticipatory shipping” solution, which basically comes down to Amazon shipping products even before customers have ordered them. Amazon will use an algorithm to pre-determine things that people want to buy, based on a mix of data including previous purchases, questionnaires, wish lists, browsing data, demographic data, etc.

Even though this “anticipatory shipping” concept is in its early stages, these are the more interesting aspects worth looking into:

  1. Anticipating demand – This algorithm will help Amazon to anticipate demand and, critically, the location where that demand is most likely to arise. Predicting user demand is the holy grail for most eCommerce or content businesses and Amazon is no exception. Amazon is looking to utilise its treasure chest of customer data – behavioural and demographic – to the fullest.
  2. Impact on distribution – It all depends on how exactly Amazon will implement its patent, but one can imagine that their whole distribution and supply chain system will alter dramatically. For example, the patent includes an outline of “speculatively shipping” packages to destinations and how to re-route items based on proximity to potential customers for those items. One of the diagrams included in the patent (see Fig.1 below) gives a clear indication of what such a scenario could look like. As a result, goods might stop being stored in huge warehouses but instead be continuously on the move in trucks.
  3. Improving Amazon’s recommendations even further – Similar to the point I raised about anticipating demand (see point 1. above), it will be interesting to see how Amazon will further improve its algorithms to predict customer demand accurately. Amazon is already doing a pretty good job at figuring out user profiles (Darius Kazemi’s “Random Shopper” provides a good case in point) and providing content-based recommendations, i.e. looking at similar products that you have purchased previously. In order to get anticipatory shopping right, Amazon will base its predictions on additional data such as product searches, wish lists, shopping-cart contents, returns and more random aspects such as how long a user’s cursor has hovered over a specific item.

Main learning point: Anticipatory shipping sounds a bit futuristic, but its underlying rationale and benefit is clear to see. What intrigues me most is to see how the customer will benefit from a predictive, as-and-when delivery approach. One of the key factors in this equation is the quality and accuracy of Amazon’s predictions. I guess that’s the thing I will be most interested; how will Amazon accurately predict what I need, where and when?

Fig. 1 – “Speculatively shipping” by Amazon (taken from: US Patent 24 December 2013 No. 8,615,473 B2)


Amazon anticipatory shipping


Related links for further learning:

  1. http://www.forbes.com/sites/onmarketing/2014/01/28/why-amazons-anticipatory-shipping-is-pure-genius/
  2. http://randomshopper.tumblr.com/post/35454415921/randomized-consumerism
  3. http://www.forbes.com/sites/stevebanker/2014/01/24/amazon-and-anticipatory-shipping-a-dubious-patent/
  4. http://techcrunch.com/2014/01/18/amazon-pre-ships/
  5. http://www.theverge.com/2014/1/18/5320636/amazon-plans-to-ship-your-packages-before-you-even-buy-them
  6. http://blogs.wsj.com/digits/2014/01/17/amazon-wants-to-ship-your-package-before-you-buy-it/

Book review: “The Lean Entrepreneur”

At first, when I picked up “The Lean Entrepreneur” I was slightly skeptical. I wondered how much new stuff I was going to learn from yet another book on “lean” practices. However, this book by Patrick Vlaskovits and Brant Cooper does provide a lot of interesting case studies and practical tools to implement. “The Lean Entrepreneur” can provide value to both people completely new to lean practices and to those who have applied lean principles previously.

For example, the book explains that “flow” in lean terms means that manufacturing is pulled through the product development process only when demanded by the next step in the product development process. I can imagine that practitioners familiar with lean principles will know this kind of stuff but will nevertheless find chapters and case studies around topics such as “minimum viable audience” and “viability experiments” of (practical) interest. These are the things that I found most helpful in “The Lean Entrepreneur”:

  1. Opportunity Matrix  The Opportunity Matrix is in essence a simple way of looking at one’s different (potential) user segments and their relevant characteristics. This tool helps in comparing aspects such as “depth of pain”, “budget” and “size of market” of each individual target segment.
  2. Value-Stream Discovery – A key aspect of lean thinking is to identify and optimise the value of your product. For instance, the authors suggest that a high level value stream might look something like this: validate product idea -> validate product -> validate marketing and sales -> validate growth engine. Validating the growth engine is about determining how to convert satisfied customers into passionate and loyal customers, enabling the business to scale. Dave McClure, who introduced “Pirate Metrics” suggests the following steps as part of this specific value discovery process: acquisition -> activation -> retention -> revenue -> referral (see Fig. 1 below).
  3. A Minimum Viable Product is about nailing the specific – In the book, Patrick and Brant offer a good reminder of what the Minimum Viable Product (‘MVP’) is and what it isn’t. An MVP isn’t about trying to capture all the general requirements in your product, throwing it against the wall and seeing what sticks. One could argue that this isn’t necessarily a bad approach and that it might even be successful in some cases. However, I liked how the authors stress the importance of “nailing the specific, not the general”. This means figuring out what your customers do with your product to complete the job your customer hired your product for.
  4. MVP Testing – I’ve written before about the importance of validating assumptions. The book does also talk about using validation as a core focus for your product focused activities. For instance, when you do interviews with users to find out if there’s a “product-solution fit” it’s all about setting a clear goal for these interviews; “we will talk to 10 customers and validate that they have problem x.” Another example is MVP testing, where the focus is on validating your MVP and validating the value that you expect your MVP to deliver (I’ve included an example in Fig. 2 below).
  5. Funnel vision – I SO want to have a ‘funnel vision’! One of the final chapters in The Lean Entrepreneur is dedicated to moving customers through the funnel. As I mentioned in point 2. above, the challenge here is to effectively move your customers through the funnel, in a way that’s appropriate for your business or product. For example, I wondered what a customer funnel can look like for a B-2-B-C business which creates Software-as-a-Service (‘SaaS’) products (see the related “B2B SaaS example” in fig. 3 below).

Main learning point: if you’re looking for more practical tools on how to best implement a ‘lean’ product development process, then “The Lean Entrepreneur” is your book. If you’re a ‘lean veteran’ you might already know quite a lot of stuff that’s in the book, but I think you’ll still benefit from both the practical suggestions and the real-life case studies which Patrick Vlaskovits and Brant Cooper have outlined in this great book.

Fig. 1 – Sample user acquisition and conversion funnel following Dave McClure “Pirate Metrics” (taken from: http://freemindtraining.com/blog/startup-saturday-delhi-june-2013-iit-data-journalism-data-analytics-for-startups/)

User acquisition & conversion funnel Fig. 2 – Web App example of MVP testing (taken from: Patrick Vlaskovits and Brant Cooper – The Lean Entrepreneur, p. 184)

  • Segment: First-time startup founder needing the ability to read market signal for their product
  • Minimum functionality: Assumption wizard, survey tool and dashboard
  • Satisfaction: Use of online wizard: one per week; sends survey: once per week and views dashboard: daily
  • Passion: Resubscribes, acts as reference and a Net Promoter for the product

Fig. 3 – B2B Software-as-a-Service (SaaS) example (taken from: Patrick Vlaskovits and Brant Cooper – The Lean Entrepreneur, p. 199)

  • Funnel State 1: Aware
  • How do you know? Comes to landing page from tweet
  • Customer now wants to: Learn more about the product
  • So, what do you do now? Messaging and positioning; special offer
  • Funnel State 2: Intrigued
  • How do you know? Reads product description, benefits, requirements, testimonials
  • Customer now wants to: Equate price with value
  • So, what do you do now? Online return on investment calculator
  • Funnel State 3: Trusting
  • How do you know? Signs up for 30-day trial
  • Customer now wants to: Test product
  • So, what do you do now? Thank customer, send “getting started” package, offer personalised customer support
  • Funnel State 4: Convinced
  • How do you know? Clicks on “check out now”
  • Customer now wants to: Buy; needs boss approval
  • So, what do you do now? email sales for references,; provide competitive analysis
  • Funnel State 5: Magic
  • How do you know? Demo was successful, pricing page studied and competitive analysis studied
  • Customer now wants to: Buy, but is a bit reluctant
  • So, what do you do now? “Buy now” discount
  • Funnel State 6: Customer
  • How do you know? Input credit card number and clicked “complete purchase”
  • Customer now wants to: Realise value proposition
  • So, what do you do now? Get them into product experience (go outside funnel)

The Lean Entrepreneur