[Big]Data meets the start-up’s – 10 fundamental questions (Part I)
1) But seriously… should I care about data?
Do you analyze on a regular basis the social media channels to understand your potential customers? Do you regularly try new ways of engaging customers based on measurable results? Do you think of defining different models to quantify the satisfaction of your customers? Are you tracking your competitors‘ campaigns and the reaction of their customers to learn from their mistakes or from the things that worked for them? Do you put in place a system to measure the response to your marketing actions together with a process to leverage the findings and incorporate them into the next action? Do you know where your demand is geo-localized to apply differential targeting? Are you monetizing the data assets you got create a new revenue source for your business?… Well, the list doesn’t stop here… and be sure of one thing: if you are not doing it all of that, your competitors are certainly doing it and more.
The sooner you get aware of following points, the better:
- You live in a digital era, where the information is the most critical asset
- You can‘t let your competitors have this advantage => you need to make the most of it now
- You should consider both sides of the data-coin: your own core business and making a business out of your data
2) Where is the value in the data?
The following diagram explains how the value increases as you move from the left (Data as a commodity) to the right in the data value chain.
Logs, GPS traces, temperature values, social media posts, etc… per-se of no (much) value. You can sell it, so that others can turn it into insights and take it further up in the value chain, or you can do it yourself.
The Data Science’s main goal is generating value out of the data, turning it into actionable insights in form of a report, a set of charts, metrics and KPI’s, behavioral patterns, etc.
Creating a report for the sake of generating insights is pointless, unless the insights are sufficient to support the knowledge in driving meaningful actions that lead to the desired results. The lack of actionability is a common disease in bigger corporations with quite powerful Business Intelligence (BI) departments, where the decisions take place pretty far away from where the insights are created
You know your insights are actionable when: – they allow you to leverage the learnings from the past to understand effects and plan future actions accordingly. – they help you discovering what you weren’t aware of, so that you boost your business with new enlighting new findings. – you can put in place a (near) real time monitoring system based on them and steer on-going actions (like an AdWords campaign, etc) – they enable you to understand the consequences of a decision even before having taken the decision (predictive analytics)
At the end, the only thing that really matters for a company, is the results. Yet you cannot stop the cycle here: you need to make your results so measurable, that you can retro-feed them as data and start generating insights and driving better actions to get even better results.
In a nutshell, the value is not (only) in the data:
- Data is only valuable to create insights
- Insights are only valuable to help the Knowledge drive actions
- Actions are only valuable when they lead to the desired results
- Results can drive improvement if they can be measured, for which we need data to start the cycle again
3) How about data protection? What do I need to know?
Data Protection: common base but different everywhere
Terms and conditions and Opt-in handling
Anonymization vs. Pseudo-anonymization
In a few countries, there’s a distinction between anonymization and pseudo-anonymization: even if you have encrypted the PII information in a data record, you could out other records together over the time and work-out who the particular person is. Anonymization in these countries is more challenging, as you might need to change the encryption keys to shorten the amount of history available about a customer making much harder the identity disclosure.
Managing the PII History
You might also face cases where your customer wants to get the information related to her activity deleted (typically after churning, etc). You might need to offer a mechanism to enable that
- Make sure you are aware of the Data Protection Law in the countries you want to work with customers data
- Get legal assistance to review the Terms and Conditions and opt-in‘s/opt-out‘s
- Make sure you comply with anonymization, storage and deletion request requirements
- If you are thinking of selling the data you collect, get legal advice
4) How much value is in my data? What can I do to increase it?
Likewise, the quality of your insights typically increases as the volume of data you exploit grows
- Don‘t delete any logs, etc => Collect all you can
- Save your data in a handy way
- Keep doors open to matching attributes
- Whatever you save, always add the geo-coordinates and time-stamp