Archive by Author

February 2019 Recap – What Lies Beneath Sentiment Analysis

Despite the crappy weather, many in the group recognized this event would have been a terrible talk to miss out on!*

If you didn’t manage to make it to our February event with Dr. Marie-Catherine de Marneffe from the Ohio State linguistics department you might wonder why my writing is even (slightly) more convoluted than usual. Those who did attended will certainly recognize this as an example of a sentence that would be judged as positive in sentiment by a human, but perhaps negative by a computer.

Dr. de Marneffe provided the group with fascinating insights about how sentiment analysis engines really work, and also when they might fall down on the job. These systems can be incredibly powerful and useful, but before relying on their output for real-world decisions we should really understand some basics, including:

  • What data the model was trained with. If the test data is similar to the training model (for example determining the sentiment of a movie review with a system trained upon movie reviews), then we might hope and expect to get some pretty accurate results! But take that same classifier and apply it to all the tweets you find about your product and maybe not!
  • What kind of output does the model create? If you’re making real-world decisions based upon what the model tells you, maybe get some more details than happy face vs. sad face? It’s not an all-knowing magic black box and dangerous things can happen when we treat it as such.
  • What are the biases inherent the training data? The decisions made by the system are reflective of the data, warts and all! (Anyone remember the crazy Microsoft AI twitter bot?)

Real world examples of additional language data encoded in non-word form (yes, I mean their hands).

There will be no March meetup, but we encourage everyone to join us at the Women in Analytics conference at the convention center!

In April, please come back to Rev1 to hear Sharon Santino and Kevin Boehm talk about autonomous vehicles.


*So how did our first sentence do when run through the Stanford NLP lab sentiment analysis demo? Well, if you believe the computers this must have been a pretty lackluster event…, wrong again! Maybe that’s why they call it “machine learning”, not “machine knowing”?

January 2019 Recap – Dabbling in Data Science

For our first meetup of the new year we were in a new space (Hopewell) with a lot of new faces! Our speaker wasn’t new though, it was none other than cbuswaw co-founder and data science dabbleR Tim Wilson.

2019 is definitely the 11th year of Web Analytics Wednesdays (we counted), but what else is it? Is it the “Year of Mobile”? Or maybe the “Year of Linux on the Desktop”? Tim declared it to be the year of “Applied Data Science”! Sounds good to us. The mobile thing has already had a few years, and the Linux desktop thing doesn’t seem too likely… so let’s go with it!

But what does “applied data science” mean? Or “data science” for that matter? Rather than debating the definitions for the 100th time or drowning in Venn diagrams — Tim got to what it’s really all about, using data to answer questions.

Tim had somewhere between four and five examples of using R with Google Analytics to give new perspectives into the same old data that we all have and see. Which blog post was the most effective? What are the users on my site really interested in? When is my site most heavily used? Unlikely the definition of “data science”, all of these questions are eminently answerable with the right approach. You can still use a Venn diagram though if you’d like.

Tim’s approach stated no fancy paid GA or BigQuery required, no sites with millions of sessions, but a simple process:

  1. Have a question or idea about your data.
  2. Explore that data.
  3. Use R (and Shiny) to visualize and iterate on your ideas.

Tim also was brave enough to do a real-time demonstration of the R Shiny apps that he has created and made available. All of the code, slides, and links to the apps are available on github here:

https://github.com/SDITools/ga-and-r-examples

You can also watch the presentation itself!

Please join us again next month at Rev1 when Professor Marie-Catherine de Marneffe from Ohio State will cover sentiment analysis, which should fit in nicely as an extension of some of what Tim talked about this month with text analysis!

October 2018 Recap – Call Tracking and Analytics with Alain Stephan

At our October meet-up we learned a lot about an oft-neglected part of that tiny multi-function computational device we keep on us at all times: the phone. You know, that thing you use when you talk to people? Really, that does still happen… A lot.

Alain Stephan, SVP analytics services at call tracking and analytics company DialogTech, showed us how and why we might want to actually pay attention to what customers say when they call, no matter what size our business is.

162 billion calls driven by digital marketing!

Alain walked us through the basic mechanics of how this kind of system works: forwarding, recording, extracting the contents of a call, and then mining that content for points of interest in a customer journey.

We learned that the building blocks of this kind of system have come as far from the days of dial-up Compuserve (their contributions to text-to-speech technology notwithstanding) as a iPhone from an old rotary phone.

Maybe this doesn’t sound like “web” analytics, but those of us that are driving traffic and running marketing campaigns might want to think about what happens when the click we drove turns into an inbound customer call!

A digital analyst might think of the content of a phone call as an “offline” conversion, but if we can extract the contents and customer funnel interactions in that call — that doesn’t sound like something fundamentally much different a series of website interactions to me. And as we learned from Alain, one call can have a series of different interactions just like a site visit. If we’re listening to the interactions in a site visit shouldn’t we be listening (this time, more literally) to the phone call interactions as well?


 

September 2018 Recap – Experiments in Attribution with Katie Sasso

At our September meetup Katie Sasso from the Columbus Collaboratory took us again into that most dangerous of web analytics jungles — attribution. And by “dangerous” I of course mean working on it puts you in danger of straining your back after hurling your computer out the window in frustration.

Katie walked us through a fresh approach to this topic covering how she would (and has) attacked the problem from a more scientific perspective.

From the rules-based approaches most commonly used (as in Google Analytics) to some example R code for Markov Chain attribution Katie got quickly to puncturing some of the vendor-hyped conventional wisdom. She also encouraged the audience to take the same kind of experimental approach to other common problems like A/B testing.

 

Katie also was the first cbuswaw speaker to create & present her slides in R Markdown (code on her github), which you will likely have more success in downloading and compiling yourself than PowerPoint.

View Katie’s Slides Here

 

Upcoming Events:

September 21 (at the Columbus Metropolitan Library) — Market Research Exchange of Central Ohio’s fall conference.
September is Data Analytics Month at OSU, with events all month long.

See you next month for learn about using AI for call tracking.

August 2018 Recap – Storytelling With Ruth Milligan

“They want the Teddy Bear, not the stuffing”.

So what does that mean exactly?? No, it’s not a German translation of an old Steiff company slogan — it’s about putting your audience first and thinking about how to be an effective communicator. It doesn’t have to be a TED talk viewed by millions; any time we speak to a group it’s a chance to hone our communication skills and craft a story that puts our message clearly in minds of our audience.

Our guide and personal Obi-Wan for this excellent session was Ruth Milligan, founder of Articulation executive communication coaching and organizer of TEDxColumbus.

Ruth knows a thing or two about how to communicate with an audience and shared a lot of wisdom with a packed house at Rev1.

Ruth’s slides are available below (most of the videos she showed are included as links in the slides):


The big takeaways:

  • Make sure your presentation has a clear “Why?”
  • Boil down your presentation to one idea (if you truly have multiple ideas, you have multiple presentations)
  • Limit yourself to three supporting points
  • Kill the bullet points!

The talks that Ruth mentioned were:

Casey Brown – know your worth
Chip Kidd – designing books
Eli Pariser – beware filter bubbles
Sam King – infectious diseases
Yiem Sunbhanich – data is the new seat belt

Articulation’s YouTube channel (including the short videos from the talk).

Join us next month on September 12th when Katie Sasso from the Columbus Collaboratory will talk about experimental design and machine learning.

Also don’t forget about the Market Research Exchange’s Fall Conference downtown at the Columbus library on September 21st!

June 2018 Recap – Mastering Google PLAs with Samuel Johnson

At June’s meetup Samuel Johnson from Adept Marketing got deep into Google Product Listing Ads. He showed us that despite what it might look like at first, there really are a lot of ways to optimize PLAs. And considering the growing domination of PLAs in search results (76% of US retail search ad spend, and over 85% of clicks according to Adthena), if you are selling stuff online you should start learning more about PLAs! Certainly your competitors are.

Samuel showed the group a couple of interesting case studies including lots of great tactical details. This being cbuswaw, when the implementation details come out so do the questions — and I believe we may have set a record in terms of total questions asked during a talk. Samuel was there with the answers — as well as an additional magic trick of turning the content of his slides to snow with the wave of his hand:

 

Minor technical glitches aside I think everyone learned a lot about PLAs including some really interesting optimization details you won’t come across every day!

We are taking July off but will be back in August with Ruth Milligan.

April 2018 Recap – Winning Holidays in Paid Search with Rob Barto

It was Christmas in April at CBUSWAW! Not because it has been snowing again, but because Rob Barto from Search Discovery came down from the far north (Cleveland) with the gift of paid search knowledge for any big event on your business’ calendar. But really, it can also stop snowing now.

Speaking of calendars, before Tim introduced his fellow Discoverer Rob he ran down our local upcoming analytics events:

First Rob ran down some quick hits on overall strategy like: plan ahead as far as you reasonably can, but be prepared to pivot quickly once your campaigns actually get going (so is he Santa Claus or Clausewitz? I’m getting my metaphors mixed up). Then Rob ran through a set of 3 detailed case studies showing some pretty incredible holiday results in both paid search and SEO.

Rob pointed out that while indeed some of these results were pretty eye-popping (1,000% lift anyone?!?!) good results are not usually the result of a single magic marketing bullet, but a full-court press across all possible angles of paid search and SEO. Want to do great on Black Friday? Well, you better start thinking about how you might approach that this Friday, and start thinking about all the different things that might help you do so, like maybe:

  • Offline conversion tracking
  • AdWords dynamic search ads
  • Timers in ads showing how long until an event happens or how long an event has left
  • More logical and actionable account structures
  • Smarter retargeting (like not showing ads to someone who has already purchased, or bidding more on existing customers, etc.)

Getting fantastic results like the ones Rob showed might require getting all of those items right and then some, or it might be something totally different for you! It’s a question of taking best advantage of the tactics that might fit your campaign and product best and then testing those ideas and working with the client.

To complete the holiday theme Tim then proceeded to hand out gifts for good questions from the audience, though let’s face it, they all were pretty good questions!

March 2018 Recap – Facebook Offline Measurement with Monish Datta

To celebrate 10 years of Columbus Web Analytics Wednesdays we realized there could be no better speaker than CBUSWAW legend Monish Datta. Sure, Monish might now be the Monetization Product Marketing Team Manager at Facebook, but we knew him way back when — and he hasn’t forgotten his Ohio roots. Managing to escape from New York just ahead of yet another late season snowstorm on the east coast, Monish stopped by to drop a whole bunch of knowledge about how Facebook is dealing with offline measurements such as store visits and offline conversions.

Despite our feelings that online commerce is taking over everything, online retail only accounts currently for about 10% of all transactions! This is why understanding what happens in “the real world” and bringing that gap between offline and online activity is so important. This is tricky stuff — not because there is any one piece of the technology puzzle that is so complex, but because there so many different data sources, touch points, and privacy concerns.

An engaged crowd asked some tough questions about how this type of measurement works and Monish was ready with answers. As a group many of us had experience working with Facebook Ads of multiple types, but very few had ventured into this area of offline activity. Despite threats to pivot to talking about himself — there were many answers to be had, and only a couple of “we can’t disclose that”s.

Ultimately, Monish described this sort of real-world activity as an “offline pixel” where, with the right data integrations, we can start to think about offline activity in some of the same ways we think about online activities — as those two different arenas continue to overlap more and more.

We hope to see everyone next time at Rev1 in April when Rob Barto from Search Discovery will talk about holiday paid search planning! It’ll be Christmas in April, with hopefully some less Christmas-like temperatures.

 

February 2018 Recap – Segmentation with CBUSWAW

For February we did something a little different. No speakers, but instead a bit of group introspection… in the form of Dave Culbertson asking questions and the audience going into different corners of the room depending on their answers. No dunce hats, not even for those who are clearly using the wrong type of phone while eating the wrong type of candy. It was a fun time, we swear!

But first, a quick run-down of upcoming analytics events:

We’ve done this kind of segmentation report before, back in May of 2014, so we decided to ask the same questions again and see how things have changed. Turns out it’s a pretty hard thing to keep asking the exact same questions the same way each time (props to Gallup, asking the same questions in the same way for over 80 years), but we did our best.

Some interesting findings:

  • Analysts continue to come from a pretty mixed background! “None of the above” remained our top first career job (chart #1 below).
  • We were split 50%/50% on iPhone vs. Android, whereas last time we were significantly more iPhone, 65%/33% (#4).
  • LinkedIn is where it’s at for professional social networking, at 92%! (#6).
  • For tools – our percentage of Google Analytics users was down, but “Other” was way up. This “Other” includes people who considered analysis tools like Tableau or R to be their primary rather than a traditional web analytics data platform like GA (#8).
  • A “What’s the deal with Nutella…?” bit is part of Dave’s standup routine he’s currently workshopping (#18).
  • Email reigns supreme for our desert island marketing channel (#17). Not sure how we’ll be receiving that email on that desert island, but we’re working on it (literal mail chimps?).
  • The distribution of our birthdays (#15) show either analysts tend to be Geminis at very high rates — or possibly one needs to be careful about drawing too many conclusions from small sample sizes.

So without further ado, let the data puking commence!

2018 vs 2014 CBUSWAW Audience Segmentation

 

See everyone next time, when Facebook’s Monish Datta returns to town to talk about how Facebook is exploring offline signals that consumers generate.

January 2018 Recap – Truth in Attribution with Clarivoy

We welcomed in the new year at Web Analytics Wednesday with a further exploration of one of the toughest topics around, attribution. In spite of the cold temperatures, the crowd gave Chris Kerr and Rick Otten from Clarivoy a warm welcome — and were rewarded with a wide array of information both about attribution in general and the challenges of marketing in the automobile space.

Chris picked up from where Jason Packer left off at our November event with a further discussion of the need to move past single-point attribution models to multi-touch. Chris also discussed the some of the details of the user journey in the car purchase process and stepped through some real-world examples with data.

Knowing the effectiveness of your different marketing channels is always a good idea — and car dealerships show some great examples of why this is so important. Many dealerships spend significant amounts in a wide variety of different channels to reach a buyer for a “high consideration” product that is bought offline — so these issues that we all face can be magnified in challenging ways.

Rick then took to the stage to present his 3 “W”s to WAW:

  1. Who is using the site?
  2. Where did they come from?
  3. What are they doing on the site?

Knock-knock jokes aside, Rick hit us with a lot of specifics of how tracking can work from the legit and well-established (you know, cookies) — to the scarily sketchy (hidden auto-completes, etc.).  Then took us through some analyst catnip with a few interesting data visualizations, including a crowd favorite dandelion chart crated by Gephi.

Video courtesy of Mixt Media.

We also welcomed two new sponsors for 2018. Continuing sponsors Clarivoy and TeamBuilder Search were joined by the Wittenberg University M.S. in Analytics Program and Search Discovery.

Please join us next month when we will do some segmentation experiments on the group itself!