Making Big Data Actionable.
Big Data. Your Business.
Where do you fit in and what does it all mean?
Metrics. Measurability. Trackability. Statistics and Analysis.
Discussion points for the presentation will include:
- Big data and it's usage by worldwide companies and brands.
- Showing you that adopting a "data first" methodology for marketing is easier than you may think.
- Show you how you can group data together to get achieve a useful amount (a statistical quorum).
- Help you to identify when you have the right data to make marketing decisions for you and/or your clients.
- Show you how to look at the data in multiple ways to make the best and most informed marketing decision.
- Explain how to adapt and reevaluate your marketing decisions after each decision is made and new data is available.
About the Speaker:
Daniel Scala uses big data at the core of his marketing decision making process. This "data first" methodology enables him to drive excellent results for advertisers of all sizes. These advertisers include worldwide brands such as: Coca-Cola, Williams Sonoma Group, NBC Universal, the Orlando Magic, Ann Taylor, and many others.
Daniel was a Senior Marketing Strategist at Channel Intelligence which was acquired by Google in February, 2013.
Currently Daniel is a managing partner at agency THE, a Performance-Based Creative Agency based in Orlando, FL.
1. Big Data in the Driver’s Seat:
Leverage data best to improve your direct marketing efforts
2. Intro to our Speaker
Daniel Scala
Introducing Daniel Scala – Today’s speaker!
3. Big Data in the Driver’s Seat:
Leverage data best to improve your direct marketing efforts
4. Goals of Today’s Presentation
Goals of Today’s Big Data Presentation:
Help you to:
• Understand Big Data
• Teach and help you to develop an
action plan to make data the core of
your marketing decisions
5. Intro to Big Data – Let’s Define It
Big Data
True “Big Data” is usually used by large companies, and could be
considered a unreachable resource for smaller businesses.
Big Data – Fundamentally the term is referring to the collection of
huge datasets that can be very complex. The collection of “big
data” is beyond the ability of traditional database management
tools. A number of algorithms have as such been developed to
capture, store, and index data.
This may seem daunting, but not to worry we’ll make sure you can
use data in your marketing decisions.
6. Big Data Usage by Worldwide Companies
Worldwide Brands and Companies that are Big Data Focused
Two of the biggest brands in the world and
names in advertising are Big Data centric
companies.
Both Google and facebook – leverage their big
data platforms to maximize their ad
revenues.
8. How Can You Make Data BIG for You?
Go Big with Data!
Few businesses will have access to huge volumes of
data like a facebook or Google.
So – What can you do?
Not just facebook and Google can have their
businesses benefit from data. Why not make your
marketing decisions based on the data that IS
available to you?
9. Using a Data-First Marketing Methodology
The Truth Lies in the Data!
Here are the 6 steps to a Data-First Methodology:
1. Determine a high-level trackable marketing goal
2. Determine what data you need to have to assist
you in achieving this goal.
3. Identify where you acquire the data.
4. Plan for your data storage.
5. Remove anomalies in your data.
6. Group similar data together – until you have
enough data to make a decision.
10. Step 1 - Goal Identification
Pick a Goal
The choice is yours to pick your goal.
Just make sure that it is something that
can be tracked and measured.
• Increased sales?
• New customer acquisition?
11. Step 2 - Define Your Needed Data Elements
Data Points
Based on your goal what data do you
need?
• Conversion Rate?
• Total Sales?
• Open Rate?
• Interaction/Response Rate?
12. Step 3 - Go Get It! - Acquire your data
Data Acquisition
Where can you get these data points
from?
If you can get all of them from digital
campaigns – great.
If you can’t then think creatively!
13. Step 4 - Put Me Away! – or so your data is telling you
Store Your Data
Few can afford or are willing to incur the
expense of a complex data storage and
indexing system.
Don’t shake your head at Excel – it’s
more powerful than you may realize!
14. Step 5 - Get Outta There! – get rid of data anomalies
Purge Data Anomalies
Remember, that one day, you had 4x your
normal sales because you ran a 50% off
promotion…
15. Step 6 - Group the Data! – 1st Grouping
Grouping Data by Geography
Pick a way to group similar data together
and create a pivot table based on it. In
our case we summed:
• Number of Visitors
• Number of Conversions
• Value of Conversions
16. Step 6 - Group the Data! – 2nd Grouping
Grouping Data by Hour of the Day
It’s time to create another view into
your/our data. This time we’ll group the
data together by hour of the day….
• Number of Visitors
• Number of Conversions
• Value of Conversions
17. Step 6 - Group the Data! – 3rd Grouping
Grouping Data by Hour of the Day
It’s time to create another view into
your/our data. This time we’ll group the
data together by hour of the day….
• Number of Visitors
• Number of Conversions
• Value of Conversions
18. Step 6 (part 2)- When is enough really enough? – Determine how much
you need for statistical relevance
Relevant Data Grouping Sizes
Now that you’ve group your data
together in several ways are you ready
to make your marketing decisions yet?
First we need to ensure you have enough
data to be statistically relevant.
19. Analyze your data and make your decisions
Time to Decide What Actions to Take
Use your grouped together data to
find areas in the data where you
have enough data to make definitive
decisions.
20. Summary and Key Takeaways
Summary
A data-first marketing methodology if executed properly WILL
increase your return and your revenue.
Stick to the relevant steps weekly to make your marketing
decisions:
- Acquire data
- Store data
- Group the data
- Make decisions based on the data
21. Questions?
Big Data in the Driver’s
Seat:
Leverage data best to improve your
direct marketing efforts
Daniel Scala
Daniel@agencythe.net
321.356.4934
Notas do Editor
I'd like to introduce Daniel Scala our speaker for today's event. Daniel has close to a decade of experience using big data at the core of his marketing decision making process. He worked as a senior marketing consultant atChannel Intelligence, which tracked over 30% of worldwide eCommerce. CI is now a division of Google. At Channel Intelligence he led some of the worlds’ largest retailers to success in their digital marketing campaigns. Throughout his career he has worked with advertisers including: Coca-Cola, Williams Sonoma Group, NBC Universal, the Orlando Magic, Walgreens, Ann Taylor, and many others. Currently Daniel is a managing partner at agency THE a performance-based creative agency, based in Orlando, FL, and has a B.S. in Computer Science from Rollins College.
"Big Data" is a term thrown around an awful lot these days. Fundamentally the term is referring to the collection of huge datasets that can be very complex. The collection of "big data" is beyond the ability of traditional database management tools. A number of algorithms have as such been developed to capture, store, and index this vast sea of information.
Yes - the biggest of the big out their use "big data" to assist them in their business. Google and Facebook. In fact... Facebook helped to develop one of the most widely used acquisition and storage algorithms for big data in the world. What do these companies use big data for as related to revenue generation? Well, Facebook uses it to store information and profile every single user on their platform. This allows them to serve ads that are based on things you "like", are suited towards your age, gender, and geography, etc... Google uses information stored in their big data repository to ALSO serve the best ads, however with Google it's based on geography, searches you've done, what's in your mail, what website you're currently viewing (adsense), etc....
Available to you are 3rd party software and tools that can make THIS level of data accessible, but are typically very expensive. So now you're asking, what can I do?NOW - Let's talk about making data (BIG) for you. Welcome to adopting a data-first marketing methodology.
These steps are the foundation of data-driven marketing.To best illustrate the implementation of this process – We will use a sample client …. A large law firm. We’ll call the law firm Scala & Scala.
In Scala & Scala’s (our sample large law firm) case the goal is to acquire new customers.
Every advertiser will have different data available, and based on differing goals this step is never a one size fits all. It's important to determine based on YOUR goal what data elements you need to collect. In the case of Scala & Scala we'll be focusing on grabbing data from digital marketing campaigns and getting the following data points:- Date- Day of the Week- Hour of the day- Region and/or State- City- Number of Visitors Number of conversions Value of conversionsThese data points are important for Scala & Scalabecause of how marketing at large is done for the law firm.
Online marketing programs/campaigns are easy to get data for. Pick your analytics package of choice whether that be: Omniture (now Adobe analytics), Coremetrics, Webtrends, Google Analytics are the most popular but the list goes on. Offline marketing programs are tougher to track, but unique phone numbers on fleet vehicles, and micro-site urls in direct mail campaigns can be very effective. With these offline examples you can count the number of referrals to the main website or forwards from one phone number to the other.With Scala & Scala we will be using Google Analytics and will stay focused on digital marketing.
Since true data repositories like advanced Business Intelligence packages can be very expensive we, for Scala & Scala, use Microsoft Excel to store our data. We also simplify the import from Google Analytics by using a tool called: Excellent Analytics - which is a simple Excel plug-in that's free. If you choose to follow this storage methodology note that each worksheet in Excel stores just over 1 million rows. As such we recommend limiting the data in each worksheet in the spreadsheet to keep one month worth of performance data.
Large data anomalies like this have to go to have some sanity in your dataset. The issue is that a high volume of good, bad, or indifferent performance data can outweigh all other data in your data pool. As such you need to have it take a hike (or archive it elsewhere).
Geotargeting is a staple across online and offline marketing. As such we start with city/state. We then therefore go back to our trusty spreadsheet and add a new worksheet to it and create a pivot table that sums up the following data points:- Number of visitors- Number of Conversions- Value of ConversionsThis data is now all grouped together based on matching city/state. This is one way you can look at your new data to make decisions.
Many advertisers find that particular hours of the day don't perform well for them. Typically, for example, in B2B businesses non-business hours and days perform poorly. Specific times of the day that are or are not performing well unfortunately aren't particularly actionable in offline marketing. Short of paying your sign spinner only during peak hours....With Scala & Scala law firm though, we go to our handydandy pivot table in Excel and change it to group based on hour of the day for our key datapoints:- Number of visitors- Number of Conversions- Value of ConversionsThis is another way you can look at your new data to make decisions. Now you know what times of the day your campaigns perform the best and worst.
Many advertisers find that particular hours of the day don't perform well for them. Typically, for example, in B2B businesses non-business hours and days perform poorly. Specific times of the day that are or are not performing well unfortunately aren't particularly actionable in offline marketing. Short of paying your sign spinner only during peak hours....With Scala & Scala law firm though, we go to our handy dandy pivot table in Excel and change it to group based on hour of the day for our key datapoints:- Number of visitors- Number of Conversions- Value of ConversionsThis is another way you can look at your new data to make decisions. Now you know what times of the day your campaigns perform the best and worst.
Having enough data to make educated decisions is important. You should define a threshold based on numbers that make sense... for example at Scala & Scala they have a goal conversion rate of .8% as such we need at least 125 visits before we should expect to see 1 conversion. In looking back at your data - now with this lense you can determine which of the near infinite ways you should look at it to make the best decision.
In looking through your data groupings sort the data from highest traffic and exposure to focus on the things that will allow you to most dramatically impact your marketing programs. With Scala & Scala we see that goal conversion rates are highest in Cities that are near branch offices. We also see that there's a significant "long tail" with the data where there's very little traffic and no conversions from a significant amount of geographic locations.For Scala & Scala we therefore choose to invest more in several digital campaigns that can be geotargeted in areas near their offices. We also choose to follow the old adage that rules are meant to be broken and we blacklist a substantial list of geographies (cities in this case) that were a part of the non-performing longtail.
QuestionsHow to push product for holidays?Good data first. Bids, products you push must also garner popularity through clicks from consumer interest. Promo text highlighting your USP on this items. Do price points effect ROAS and how do you handle price swings? Bet practise for good ROAS is to eliminate low priced items – unless the CR rate is above 5%. We tier suppressions based on price