Dashboards are for analysts – Insights are for everyone else

21 Sep

I have been going through the Udemy Lean Startup Course in my “spare” time recently and one thing has resonated with me.  It is about defining the Customer vs the Product.  i.e. You and your facebook profile are not the customer, you are the product.  The customer is the advertiser, investor, perhaps even acquisition targets.  In our Digital Analytics world, analysts are the customer of Dashboards, the product are the insights and that is what people are “buying”.

Now I know this seems obvious at first, but before you say I am stating the obvious – how many canned reports did you send any Directors or above this month?   More than 1 — keep reading.  We are all guilty!

IMHO, this is the high-level breakdown of responsibility within any organization.

C-level – Vision   |   Directors – Strategy   |   Managers – Execution   |   Knowledge Workers – Tasks

Any CEO will tell you what they are concerned about is what is moving in respect to the vision and strategy they have put in place.  Not what is at the top, or what the current count is.  Because it can take time to construct the “perfect” dashboard, it rarely meets the needs of what is happening at that time.  Now, if you are not doing anything to influence the metrics, why are we looking at them?  Assuming this is untrue, shouldn’t the reporting reflect the things that are being done at that time to influence results?  TA DA, welcome to the number 1 reason those canned reports are not being read.  Do you read things at work that don’t influence your job?

Now this makes your job challenging.  Isn’t that wonderful!  Gone are the days of mundane dashboard work in lieu of building a semi-perfect automated report and delivering what I call an “Insight Report”.  This is tied directly to both initiatives that are happening and opportunities that are awaiting.  That is our job.  The dashboard is meant to make OUR jobs easier, not theirs.

In later posts I will detail the Insight Report as well as the process to assure systematic, data-driven change management.  Sounds groovy doesn’t it!

 

 

SEO yesterday vs PR today – any difference?

20 Aug

Since MashMetrics has been working with more  local and smaller organizations here in San Diego, I have noticed a common trend among their impression of what Search Engine Optimization (SEO) involves.  It seems to be a common impression that to build a “SEO friendly” website is a simple exercise of:

–          finding the appropriate keywords

–          matching with competition and volume in the marketplace

–          assuring META information

–          assuring the robots.txt is in place

That used to work, but now there is so much more.  Luckily, the above items have automated tools to bring the time involved down, however, the items that really matter now cannot easily (or effectively) be performed.  These include:

–          blog postings

–          external blog commenting

–          content creation (infographics, videos, etc)

–          backlink partnerships (people linking to your site)

The important thing to remember here is this is no longer just SEO, for good reason, Google and others put more weight in others impression of you and the depth of your content than anything else.  It seems that SEO now and days is quite simply just PR done digitally.  So how much is that worth to you?!

To clean or not to clean … how much integrity does your data have?

19 Mar

In building the #Marchmadness Twitter Timeline Dashboard I fell across an issue/insight that actually bugged me for days.  The Twitter hashtag #MarchMadness was being used, in high demand, by a couple spammers to send out deals and offers (no surprise).   The Top Tag Bubble seen below, solidified that immediately.

mm-spamandno

While a little disappointing, the data is completely legit (I mean why analyze if not to see the marketing use of #Marchmadness as well?). To us marketers, knowing the #hashtag ecosystem of a topic is super important (hence why I assured my data model accounted for this). Knowing that for the first week, nearly 20% of #MarchMadness mentions are accompanied by “#Moms” and “#giveaway” is the exact relationships I wanted to know.

Playing Devil’s Advocate- This dashboard is not just for marketers, it is also about the fun and excitement of #MarchMadness.  As time goes on, I am much more interested in hearing how Gonzaga and KU fans are trash talking each other rather than how I can save 20% on dishsoap.  Even more important, once the initial Spam Usage was identified, I was able to conclude which tweets were actually relevant to basketball.

So, #dataviz friends, I decided to provide the best of both worlds, mainly by creating a simple data filter which took about 5 minutes total to complete.

Using Tableau, it is quite easy:

1) By creating two data sets, and

sets2

2) Creating a Parameter Control to place in workbook

Param

filter

3) Finally a Calculated Measure to Filter the data across the Workbook
sets

Now I can switch back and forth, and so can everyone else!  This will also be much more scalable in creating future data sets (team sets, media users, etc).

Click here to view or download the full #Marchmadness Twitter Timeline Dashboard! 

twitterdashboard_filter

 

Making data talk – or at least speak the stakeholder language

30 May

Along with the other challenges of a web analyst job … One of the toughest is changing the confused yet somewhat intrigued look off people’s faces after a long review of the month’s web performance reports.

It makes this analyst ask?
How many reports that you deliver are getting acted upon?
Are you being insightful enough to make change?
What has been your bottom-line value?

You are good at what you do. Not only is the presentation completely pie-chart proof but you have:
Highlighted the dips in % of change broken down by channel
Mapped the funnel of users between your web and iPhone app
You even presented the tagging strategy for the next social campaign
…. People walked out happy, but weeks later no changes were made based on the data.

Our job can be like a misunderstood artist: watching people enjoy your work at a gallery all day but never buy a thing. Most likely it is not because of the quality of work … People just don’t relate to the paining or your data.

How to create a water cooler moment!

1) First and foremost instead of being the person that points out the issues (even if you sprinkle in the good) no one wants a whiner. What people want to know is how to get promoted, and you’re supposed to be their key. Use this.
What are they working on?
What are their KPI’s?
How are they impacting the bottom-line, and how could you make that impact more positive!

Use your domain expieriece to let them know some next steps and the numbers to back up the decision. When you go in with a plan, identifying a dip in conversion looks more like an opportunity than a problem.

2) Second, make data interesting in a way your audience can relate to. Data is pretty “cool” right now, but that doesn’t mean your audience gets it (or worse…they think they do!).
Is your site slow (and you know due to the new report in Google Analytics?), like a second slower? Sure you know that this can start to erode conversions, but people hear seconds and disregard? Now what if we compared a second to the time difference between first and third place in the latest Olympics? That thought leaves the room

Forget the top 10 keywords or your normal referral and campaign segments. Everyone is tired of that story, even as good or bad as it may be lets try something different how about:
– Split aggregated results into numerous categories. For example phrases with st.lukes, st. Lukes + hospital, just hospital and everything else. See something different?
– look up your keywords in the google Adwords keyword estimator. Should your top keywords be something different?

3) Third, get data faster. As mentioned in my previous post, and I am certainly biased, the job of an analyst is moving outside our trusted “tools” (WebTrends, onmiture, ga, MailChimp, SalesForce…) and into using more traditional business intelligence tools. (R, tableau, qlickview, etc). These are the tools for the future, the rest are merely data acquisition tools.

ETL – know it or wither away to report monkeydom

17 May

20110516-064310.jpg

The #measure community is finally starting to embrace ETL tools.
For the non data warehousing junkies, ETL stands for Extract, Transform and Load. The three things you as an analyst loathe the most, whether you know it or not.

Extract – This is the act of obtaining the data in it’s raw format to be played with in Excel, Google Spreadsheets or your tool de jur.
Transform – oh if only the data were ready to be used! Transforming is when you clean up URL’s, create calculated metrics like ROAS, or simply group pages or events together.
Load – While officially this is loading the data back into the analysis tool, for the sake of our roles I am going to stretch this into the final presentation of data. We will say loading the data into a charting tool or even the HiPPO’s mind.

Now from a show of Retweets… how many of you do this on a daily basis! This is why I feel the ETL process for marketing data especially will see a huge momentum in the next few years. The fundamentals have been there for years, however I will argue the data we in the digital marketing measurement industry have is kinda tough.  There are new models, new relationships to build within the data, new data sources (not SQL but another REST API).

1)For the last few years we have seen open access via REST API’s from several vendors, none more utilized than Google Analytics.  The growth of API’ in just the last year has fundamentally increased:


2) The GA API has been so popular, and is such a good representation of obtaining raw data, that it has resurrected old-school data extraction tools such as NextAnalytics (formerly…), Analyst Canvas (formerly DataMartist).

This speaks to the power to having quick access to raw data. The analyst community must give a hand to all the developers contributing to making this job easier for all of us. However, the challenge has only just begun. If all we are getting is raw data quickly, that certainly won’t help us transform and then load it into the HiPPOs brain.

The Transform and Loading process is where we will be providing the most assistance. While many companies have been stuck in a rut normalizing things such as customer lists, market territories or regional sales cycles…we have a whole new challenge. Completely unstructured, mis-understood, at times ambiguous datasets. We all know the “transformation” this data takes to become actionable (or loadable).  MashableMetrics will  fix that!

Metrics into action – a simple framework

17 May

Break through the numbers into Action

There are many obstacles to getting to the right data at the right time, however, with a little bit of planning and a some more flexibility your team can be off to a running start.

Solution #1: Build an Enterprise Analytics Framework

While you can acquire all the reporting tools you want, none of them will tell you how you are performing against your business goals. In order to adopt analytics across your organizations you must adhere to the company’s corporate goals and objectives. Then, they can be translated into the activities your company plans on performing.

  • What projects are in the budget/road map this month, in 3 months and in 6 months? (adding a hero graphic with 3 highlights to the homepage, adding a new feature to the product detail page, launching a new post every week)
  • What are you trying to achieve this month, in 3 months and in 6 months? (increase click-throughs to event pages while decreasing homepage click-though, increasing overall Add to Cart rate from Produce Detail page, increasing New and Returning Visitors to the site)
  • What are the 3 things you hope happen? (Increase paid sign-ups to event classes, Increase Average Revenue Per Visitor, increase advertising inventory while also maintaining well targetted traffic)

A little more work and…

  • What do you want to prove/disprove/do? By promoting the events more heavily on the homepage we are able to increase deep level traffic towards conversion, adding new features engage the user thus sending them to purchase our products, if we write more they will come
  • What do you need to track?(NOTE oversimplified, need historic (few months and current data):: HP bounce rate, event pageviews and visits both altogether and from the old HomePage;  Conversion Rates from entire site, from product page, Average Order Size + Rev per Visitor, Time on Site, Return Visitors; Total PageViews, Ad Units Served, Posts Launched (Day of Week), comments, New vs Returning Vistors)

Once this is in place, stay with it, try not to stray.  Now, instead of just looking and presenting data that might appear to look interesting you are sticking with a plan, proving or disproving activities, and if you listen to the data, coming up with next steps to improve even further!

Instead of the analyst running around performing one off requests they have a solid regime of reports. In the fewer cases where that advanced, manual extraction, drill down reporting sheet is needed the fact that it will be acted upon lights the fire.

Solution #2: Define Accountability and Standardize data communication

Organizations that have a process in place on what happens under certain criteria (landing page bouce rate is low for a page with a high entry level) and who performs the change (Sally in design meets with Rich in Marketing) have a much greater chance of making changes that make the most impact.

Building out the framework further to include:

  • Who delivers the data? (Tom in Marketing, the analyst)
  • How often is data delivered?  (Daily via email, weekly in PPT)
  • What signals a change request? (Change in Bounce or Entry Rate by 5% over same day last week or average daily change)
  • Who is responsible for improving the data? (Sally in Design, but Rich and Tom both are heavily invested in coming up with strategies)

Now, when the data comes in as planned, everyone knows what to do. Those who simply use it as a sticker better start paying a little more attention to it. The HIPPO now has a direct arrow to point fingers to.

Solution #3: Automate report building and delivery

All this process without automatation just doesn’t make since. Some of the key obstacles to automation in the past is the one off reporting and ill defined reporting led to a negative cost\benefit to automate the reporting. It was much cheaper to buy your analyst lunch every week so they would stay late and finish your report that was a little different because you were curious about … something.

Define all your data sources and find out how to automate the data collection week over week or month over month. (Beg your data provides or integrate with their API web service, ODBC, or easy extraction method, or switch to a provider that has one).

Sign-off on how the report will be displayed for everybody. Decrease the report customizations across the organization and start formulating standard data views.

Intranet analysis – measuring and managing your organizations content

10 Apr

The Intranet is the center of your organization. It defines the corporate culture, allows for collaborative learning as well as allows for unlimited workflow efficiencies.

Not much different from a normal site, however, your audience is COMPELTELY DIFFERENT, thus, the objectives you track will be different as well.

You sit on the segmented clickmap of your organizations workflow. Embrace it. Who is really engaging in that sales info. Is there a need to translate content. is the corporate balanced scorecard seen by all organizations? Is everyone pulling out of their 401K plan? Do you see excessive downloads coming from a “relocated” department?

While normally the CEO is trying to set this pace, you as an analyst have the power to do more!

WHY YOU SHOULD TRACK YOUR INTRANET.

Reason #1 – You are spending time, effort, money, ideas on it. Why not!

Understood, depending upon the size of your company this may or may not be true. Let’s say it is true. Intranet’s cost a lot of money. Expand this idea to your extranet and then you are talking true budget greenbacks, none of that internal IT money.
You purchase and worry about Intranet search, the help content, your HR information/collaboration, sales information. Now wouldn’t it be nice to offer the same analytics on that content as you do on the product page of your website? Do bounce rates, newsletter tracking (you get corporate newsletters, don’t lie), search terms not found reports ring a bell?

Reason #2 – Remember when there was low hanging fruit?

That might be gone. Analytics is hard work now. Unless you just launched your site you are now looking into behavior targeting, A/B testing, customer segmentation by now. Make those same changes on your Intranet (remember the improvement margins!!), and you will have a solid work plan for IT for quite some time. (think people will mind being busy in 2010 … think again!)

Reason #3 – Did I mention our how big my rack is?

Let’s face it. We love our large rackspace. No matter who you are, you know you get excited when Google announces another Gig of email space, even though you have only used 27% of what you have. Samething is happening across your Intranet servers.

“The Intranet must be up” and so each service has redundancy, a load balancer, and tons of space. Employees are saving everything on the Intranet to share. Several times over. Content is ALL OVER THE PLACE but only used twice and trashing search results. Depending upon your organization, a pretty large savings in a time where a penny saved is a penny earned.

There are several challenges though-

1) Start with the basics, you will make more friends. You are no longer working neck to neck with the marketing guy on raising SEO conversions. YOu are trying to convince sales departmerns or content editors to listen to insights from the intranet data. Very different conversation!

2) get ready to use hits again, and log files- and then defend their valued use. Remember, intranets thrive on file dls, video, and untagged areas.

3) you have the holy grail (100% registered users), and will notice the tools aren’t built for this. Warehouse is the way to go. Translate Visitors, Pages, etc with your corp. taxonomy

4) you have no baseline. No alexa, no ad metrics, no googleanalytics. When data gets strange, you better know the tool.

Quick wins

Now, while there are a few challenges, luckily there will be some easy quick wins. Remember the first time you reviewed yoru analytics, and the insights you were able to collect? The good times are back.

1) Hook up the Unique Vistors to your corporate database. Start to segment by role, discipline, etc.

2) improve search / purge ‘inactive areas’. This is probably some fairly dull (if not dead) content out there. If u want to improve your internal search (and trust me, everyone hates the corporate search engine), get rid of this content

3) define app usage – chances are hr, finance, facilities all have apps online now. See if they should be promoting it more or going back to paper.

4) influence the intranet homepage – chances are once u connect the data u have the best idea on the key areas.

5) Don’t – ab test, or beat on path analysis, and influence seo. The intranet folks don’t want to hear that yet. Don’t be an internal clog🙂

No cookies, 99% accurate visitor counts, easy data integration. This is your dream! Chase it.

Full disclosure:

1) I manage an analytics service primarily for Intranet websites

2) My primarily client is Microsoft, we have a couple of collaborate products you might of heard of

3) I have not always worked on the IT / Intranet side of the fence. But enough to talk about it beyond the last year.

4) I like WebTrends. A lot.

CREDITS: to two people who work at the WAA Genius Bar. Jim Sterne and Angie Brown. Jim, thanks for giving me the inspiration during Xchange to keep going on this niche Intranet tracking ‘thang. Angie, thanks for writing a great article on Intranet Analytics … that I disagree with (see her blog comments). That drives passion!

Angie’s article, If you pay your visitors’ salaries, then it pays to optimize, was dead on. She makes another dead on statement. Organizations are luckly if someone “slapped a tag on it” referring of course to the Intranet site.

WHY YOU SHOULD TRACK YOUR INTRANET.

Reason #1 – You are spending time, effort, money, ideas on it. Why not!

Understood, depending upon the size of your company this may or may not be true. Let’s say it is true. Intranet’s cost a lot of money. Expand this idea to your extranet and then you are talking true budget greenbacks, none of that internal IT money.
You purchase and worry about Intranet search, the help content, your HR information/collaboration, sales information. Now wouldn’t it be nice to offer the same analytics on that content as you do on the product page of your website? Do bounce rates, newsletter tracking (you get corporate newsletters, don’t lie), search terms not found reports ring a bell?

Reason #2 – Remember when there was low hanging fruit?

That might be gone. Analytics is hard work now. Unless you just launched your site you are now looking into behavior targeting, A/B testing, customer segmentation by now. Make those same changes on your Intranet (remember the improvement margins!!), and you will have a solid work plan for IT for quite some time. (think people will mind being busy in 2010 … think again!)

Reason #3 – Did I mention our how big my rack is?

Let’s face it. We love our large rackspace. No matter who you are, you know you get excited when Google announces another Gig of email space, even though you have only used 27% of what you have. Samething is happening across your Intranet servers.

“The Intranet must be up” and so each service has redundancy, a load balancer, and tons of space. Employees are saving everything on the Intranet to share. Several times over. Content is ALL OVER THE PLACE but only used twice and trashing search results. Depending upon your organization, a pretty large savings in a time where a penny saved is a penny earned.

eMetrics Contest – msNBC data analysis with Tableau

9 Mar

SITE: http://www.mashablemetrics.com
BLOG (barely worth mentioning): bosilytics.wordpress.com and mashablemetrics.wordpress.com
TWITTER: bosilytics
LINKEDIN: http://www.linkedin.com/in/webmetrics/

The Question:
Based upon weekday traffic leading up to a Holiday, how can we capitalize the most on our registered (Returning) visitors? This analysis will be used by the digital media team to display offers and profitable content around Valentine’s Day. Furthermore, they will use this analysis for future emails to determine which day of the week has the most positive activity.
(Full interactive dashboard can be downloaded at: http://public.tableausoftware.com/views/eMetricsContest/Dashboard1)

Condensed Version:

Overall

Chart setup:
In order to complete this analysis, I took the overall traffic metrics and opened within Tableau. This allowed me to automatically segment the traffic by Weekday. (This saved the need to create a separate translation sheet mapping Data to Day of Week.) I then show how each date (excluding weekends, as is a requirement from the business team) performed week over week. This is a perfect data set for our needs because we can see the disparity of two large holidays (Thanksgiving and Christmas) and they are highlighted accordingly. While it is quite evident to see the traffic per week, we also overlay the average for the complete data source for context. While highlighting the MIN and MAX band for each week, we highlighted the AVG Bounce Rate within the Returning Segment, as this is what we are truly analyzing.

DATA ASSUMPTIONS:

  • Data shows that the day of a holiday, a decrease in traffic is seen. We are assuming this is due to the fact that both Thanksgiving and Christmas are traditionally non-work days and are spent with family. We assume this is very different than Valentine’s Day.
  • While this analysis would be more concrete with actual offer and email stats, they were not available at this time. The data shown are high-level aggregates broken down by day and hour.
  • The data is representative of the last 9 weeks of the year, which include the Holiday Season. It is quite possible that yearly trends may show a slightly different picture, however, the business question does call for analysis of this disparity.
  • Weekends were stricken from both the chart and AVERAGES
  • .

ANALYSIS:
Holiday Traffic – Within both holiday cases, traffic actually increases POST event not prior. This is in direct argument to what the HiPPo assumed (traffic increases to the event, then decreases). In fact, the analysis shows that traffic increases post-event.

CONCULSION:
While the team should continue to run offers relating to V-day before and during the actual event, the former strategy to send high-paying offers 1,2, and 3 days prior to the day of the event should be reversed. It may be best to piggy-back on increased traffic the days afterward.

Upon average, Tuesday’s are the best day to send emails. While Wednesday offers about the same level of traffic, Tuesday shows a better performance from Returning Visitors.

WHY I WANT TO GO:
This by far is the easy part. I went to the second and third eMetrics in Santa Barbara, while working at Caterpillar. We had deep pockets, and it was an easy sell. It was awe-inspiring and is what kept me in this industry. That was in 2004-05, and we were still talking about issues of collecting data. Then I went to work for a vendor (and as a “consultant” you rarely get the luxury of attending anything but your own conferences), then as an independent it was hard to fork over the cash. Waking up from a 6 year slumber, the challenges are so much different (not to say the previous ones were solved!). We now have an abundance of data, coming from disparate data-sources, and we may very well be the ones to not only solve but own the Holy Grail of BI! I want to be there as we discuss how.

eMetrics is like going to a family reunion. There will be so many familiar faces, yet I look forward to establishing new contacts in the BI and big data space. And if anything else, I am also sure Jim Novo will appreciate having his smoking buddy back! (as mentioned in Twitter).

Now that was the inspirational, motivation, soft-sell. Now considering I am my own boss, I need to justify this in the monetary sense. eMetrics costs $2500.
My education goal will be to attend at least 4 workshops where I gain immediate actionable advise. I will factor a 4 hour savings in time per learning, worth $500 ($125 x4)* = $2000
But this is still soft, and doesn’t really hit my bottom line.
My primary goal will be to establish at least 8 hot leads from agencies or larger organizations. This core clientele, with a $12,000 per year potential per sale, this is almost a 1000% ROI, and exactly what I need to start funding the dream. This is just as much a funding trip as a career necessity. Even if this entry doesn’t win, I will surely be there for the social events, but that greatly decreases the chances of success. Factoring literally, the networking events are 10% of the time of the full conf, not to mention everyone is not there. That would mean that my lead potential could be 1! Not good. Considering my dream is to revolutionize the pain of collecting, transforming and presenting marketing data…we all win if I win!

*NOTE, this is not my billable rate, but my opportunity cost per hour

The #measure community in 2010. A little #dataviz to start it off right!

1 Jan

This year we have seen some great changes in our growing web analytics community. From Mergers and Acquisitions to another Privacy bash. Who helped spread these messages to us? Twitter has become my firehose of information, and these are the folks helping spread the word.
Data collected from Twitalyzer, normalized and aggregated by MashableMetrics and presented online through Tableau.

Click on Dashboard in order to interact
#measure Twitter Dashboard

MashableMetrics featured at SuperMath San Diego

22 Nov

MashableMetrics gave the public (data heads) quite a show at the SuperMath conference held 11/2010 in San Diego, california, the up and coming analytics community of the us of a! In it’s first public display, we blew away attendees by taking the normally arcane web marketing data and visualize it in a way that makes it Immediately ationable to analysts and execs alike.