Airbnb & the art of analytics storytelling!

Analytics and data are transforming companies around the world. Yet one of the great difficulties with analytics is that it can be difficult to explain and understand; it is widely held that analytics specialists don’t communicate well with decision makers, and vice-versa. As a result, analytics adoption is still not easy within companies.

Analysts, at one end, are busy learning more specialised & deeply technical methods of analysing data & at the same time they are finding it difficult to get them “heard” within organisations. Influencing ultimate decision makers is similar to selling products or services to external customers.

Analysts need to understand that when they present ideas to decision makers, it is their responsibility to sell – not the decision maker’s responsibility to buy. Rudyard Kipling once wrote that if History was taught in the form of stories, it would never be forgotten.” In her persuasion & power of story video, Stanford University Professor of Marketing Jennifer L. Aaker explains that stories are meaningful when they are memorable, impactful and personal. Have a look at this wonderful story told by Jennifer.

Stories are the best way to influence! But we don’t see them being used so often. Analytics doesn’t need you to solve only a technical problem but a “social” one. Analytics is sexy but for it to make an impact, it needs to be embedded into the fabric of the company. This calls for analysts to become more social & in fact better presenters & story tellers.

They need to learn to demystify analytics & link it to practical ways for the business to make money! And analysts need to learn to link their work to “the last mile”. Analytics should not be expected to deliver a “Aha moment”, instead it should be a “factory approach to improved decisions”. So analytics is not just a planning tool as much as it is an Execution tool to improve the customer experience & business impact. Start with a decision in mind & work backwards, not with the data in mind & working forward. And today with reams of external data available to most marketers, analytics can even mash up different kinds of data & improve the Customer experience.

Compare the analytics industry with the world of journalism. One of the most deadline filled industries in the world is getting it right with what it calls precision journalism! Despite crazy deadlines, I am amazed at the powerful stories journalists write using data. I wish the analytics industry was half way as good!!The corporate world needs to learn from this & use data to tell stories better! Journalists are coping with the rising information flood by borrowing data visualization techniques from computer scientists, researchers and artists. Some newsrooms are already beginning to retool their staffs and systems to prepare for a future in which data becomes a medium.

Analysts are often tempted to communicate how they did the analysis: “First we removed the outliers from the data, then we did a logarithmic transformation; that created high autocorrelation, so we created a one-year lag variable”—& the typical business user is already yawning! The audiences for analytical results don’t really care what process you followed; they only care about results and implications

Here is an example of a master storyteller. Many people employ static charts, but visual analytics are increasingly becoming dynamic and interactive. Hans Rosling, a Swedish professor, popularized this approach with his frequently viewed TED Talk that used visual analytics to show the changing population health relationships between developed and developing nations over time. Rosling has created a website called Gapminder ( that displays many of these types of interactive visual analytics

In early 2010, The New York Times was given access to Netflix’s normally private records of what areas rent which movies the most often. While Netflix declined to disclose raw numbers, The Times created an engaging interactive database that let users browse the top 100-ranked rentals in 12 US metro areas, broken down to the postal code level. A colour-graded “heatmap” overlaid on each community enabled users to quickly scan and see where a particular title was most popular.

See more at:

Brent Dykes has this wonderful take in a Forbes article & I quote:

“It’s important to understand how these different elements combine and work together in data storytelling. When narrative is coupled with data, it helps to explain to your audience what’s happening in the data and why a particular insight is important. Ample context and commentary is often needed to fully appreciate an insight. When visuals are applied to data, they can enlighten the audience to insights that they wouldn’t see without charts or graphs. Many interesting patterns and outliers in the data would remain hidden in the rows and columns of data tables without the help of data visualizations.

data analytics

storytelling with data

Finally, when narrative and visuals are merged together, they can engage or even entertain an audience. It’s no surprise we collectively spend billions of dollars each year at the movies to immerse ourselves in different lives, worlds, and adventures. When you combine the right visuals and narrative with the right data, you have a data story that can influence and drive change”.

change management

Creating organisation changes through storytelling

Also today Marketers have access to a lot of external data. How they mash this up creatively with their own data & produce features that are of value to consumers is going to become very important in the days to come.

Here is an example:

How Airbnb can add more value to its consumers?

Airbnb is making travel easier for its consumers & today they have access to a lot of data that can make the consumer’s buying process easier!There is a lot of data available about city neighbourhoods.I thought of this particular example because of  Ben Wellington’s article in The New Yorker. He used data points from New York City noise complaints not only to map out which neighbourhoods were noisiest, but why they were noisy.

Noise data

New York data

From the screenshot above, you can see that you’ll definitely want to steer clear of two neighbourhood near the Bronx if you hate the sound of ice cream trucks.
How can this help a Marketer?: Imagine if this led Airbnb to import this data & use it to help you in selecting a place to stay. I am fresh from staying in Singapore in an Airbnb apartment which was in a noisy neighbourhood. If this can be created into an index which pops up as I view an Airbnb apartment, it adds another data based layer to my decision of which apartment to choose. You can enhance this with other data like Crime in the neighbourhood etc & suddenly data is actually adding much more value to the AirBnB platform.
So if data based storytelling can be linked to “How customers buy” , that can hugely enhance a customer’s experience & value. Think about how you can do this in your business & use storytelling to impact key decisions in your company & also your customer experience.

4 thoughts on “Airbnb & the art of analytics storytelling!

  1. Hi Ajay,

    My favorite topic so a long reply. Hope it helps you…

    Nice post and a perennial issue….tech does not understand business and vice versa. One of the biggest issue I see with the present day “digital” noise is the tendency to forget that technology is for business and not the other way around.

    In Analytics though, it should be simpler. Simpler because the very premise of data science/analytics/machine learning (whatever name you give it) is making “predictions” on the basis of a well trained model. When you make a prediction, you can back test it (on data which was not used to create your model) and show that it works – and when you can demonstrate that your prediction works, you don’t need to sell or explain. A very important feature of data science (which in a way makes it hard to explain) is that it does not explain causality whereas a human mind is conditioned to understand “how”. Analytics just spits out the “what” as is and that is its beauty – revealing a pattern which a human mind cannot decipher from volumes of data.

    And I agree that the best way to present any data or a model or its predictions (whether a numerical output or a classification) is through rich visualization. Let us say that I am talking to a marketeer for consumer goods. What if I tell him that based on your data and the statistically significant features within them that explain consumer buying patterns, there are 3 clusters of customers each with unique spending habits and product needs, thereby requiring tailor made marketing outreach and product bundles. And then go on to explain how I arrived at this…I scrubbed your data, applied feature scaling, then log transformation and finally a K-Means algorithm to create distinct clusters. Phew.. sounds esoteric and a meeting killer! …..what if I instead showed a 3D or an n-D plot that depicts the 3 clusters in space, listing key characteristics of each…and then pick 2 unique customer samples and assign them to their respective predicted clusters on the same plot. It is very easy for a human mind to comprehend characteristics of these two unique customers and then correlate it with the salient features of the cluster that it has been predicted to belong to. When mind sees an association there, credibility is established and one does not need to understand how did the algorithm conceive those clusters or how did it figure which customer segment (cluster) did the chosen customer samples belong to.

    I am training a stock price predictor using advanced machine learning libraries. I dont need to explain to anyone bollinger bands, rolling means, rolling standard deviations, or alpha and beta values for a symbol , which among many other features, will determine a stock’s price. I just need to say that my model can predict the price of stock ‘X’ five days in to the future with a confidence interval of 90%. Then I can just backtest my model on past unseen data and plot the actual vs predicted price along with a coefficient of correlation. If my model is good, the correlation will show on the plot and it will not require any explanation. Well, there will be questions on how well will the model work on future data but that is in the realm of objection handling not convincing.

    When we don’t care about how Amazon assigns computing resources to create virtual machines we use or how does the operating system of our computer enables the application we use, why should be care about what goes inside analytics so long it can spit the desired output for a defined business problem in a way that the linkage is lucid.

Leave a Reply

Your email address will not be published. Required fields are marked *