So the time has finally come for eMetrics in San Francisco. eMetrics is one of the biggest gatherings of data and insight professionals across the globe, and we’re lucky enough to be speaking at the event. In this blog we’ll keep a live update of the most interesting quotes and stats that come out from the variety of speakers.

As there are three tracks going on simultaneously, we are unfortunately unable to cover everyone. If there’s someone you feel should be written up after seeing them speak then feel free to send anything through and we’ll get them included!

Avinash Kaushik, Google: Driving Radical Transformation, A Report From The Front

Avinash is somewhat of a rockstar in the data world.  He is an entrepreneur, speaker, author, and his current role is as Digital Marketing Evangelist at Google.

In this talk, Avinash highlighted how many analysts put data as the top priority, rather than the business (be it client or company that you work for).  He argued that we need to switch this balance, to ensure that the business always comes first.

This comes from only working on data that is actually going to be useful, don’t just use data for the sake of using data.  Avinash also argued that the real value in data work comes from human analysis and insight, and making this scalable and integrated throughout the business.

Avinash also showed some great data visualisation ideas, after arguing that presenting data in tables (particularly attribution data) can be misleading, and that visualising data instead is a much better way to get your point across fairly. 

John Lucker, Deloitte: Trends That CMOs Should Pay Attention To in 2014

John Lucker has a mass of experience from consulting with hundreds of large corporations with Deloitte.

In his presentation he argued that there are a number of trends that CMO’s need to be aware of in 2014.  Amongst these included the difficulty to hire real analytics talent and the profile of a good analyst - who shouldn’t just be a data scientist, but should also be curious, creative, and be able to display a mix of data analysis and instinct.

John also discussed big data, and came up with a definition of it as being data that has multiple structures, and is so large and complex that it would take too much time for traditional tools or people to process.  However, John argued that the  big data trend in organisations is largely irrelevant, and that we should be focussing on useful data, not big data for the sake of it to fit in.

Another element of this talk was that pricing is something that is often overlooked but is one of the most important element to analyse in detail and get right.

Mark Rudolph, Evidon: A Look Under The Hood

Mark Rudolph of Evidon gave a sponsor presentation about Ghostery, a tool that allows you to see all of the third party tracking software running on a site.  Mark argued that for top retail sites every second lag in load time can cost $50m, so knowing what is on your site is crucial.

The average large retail site has around 300 tags on it, whilst the average large site has 70.

Ian Lurie, Portent: Data That Persuades

Ian Lurie presented on the need to present your data in a sensible way.  If someone has to ask you to explain your data visualisation or graph then you have missed the point.

Data needs to be shown in a clean way to allow understanding.

Barbara Coll, Webmama.com: Maintaining Sanity, Measurement, and Visibility through Redesign

Barbara Coll of Webmama.com spoke about redesigns and how to get through them.  Step 1 is to prepare for your redesign by doing keyword discovery research, looking at page authority, and looking at content/SEO strategy.  Barbara argued that your SEO strategy is really your information architecture - how you structure your site.

Barbara claimed that the aim is to get keyword category domination - to that end when redesigning, don’t change anything about the top ranking pages.  Her strategy includes maintaining internal content links from the same pages.

Step 2 is the tedious part, and includes taking screenshots of queries, keywords and clicks on webmaster tools, screen shot the SERPS across different regions and different search terms.  You should also run ranking reports and set up redirect file for .htaccess.

Once you’ve made the change to your site there’s always a few things you’ve forgotten.  Have you put in place an error page, turned on the tracking code and filtered out the internal team from analytics?  Next check for redirect errors.  Expect major ranking changes over the first 2 weeks.

Comparing after a redesign is tricky.  You aren’t comparing apples to apples as the site has changed.  You can measure traffic and conversions, and traffic to pages where the URL has not changed.  You can’t compare content to content if you’ve made changes, the number of pages visited if you’ve changed the structure, and landing pages if they’ve changed.

Aaron Levy, Seer Interactive: Small Tests, Big Insights - Using PPC To Build Site Strategies

Aaron spoke on PPC as a testing space, rather than just an ROI platform.

At a basic level, you can direct your SEO strategy based on PPC keyword conversions.  SEER do this a lot as an SEO agency.  Aaron suggested slicing your AdWords data in a lot more ways to get more advanced data.

This can help you with feature development, faqs, product priorities and more.  One thing to consider is whether competition will be a hurdle based on PPC data.

Google AdWords auction insights show who is bidding on your terms, and their impression share - this is really useful for pulling out competitor insights.

Use the advantage of fair rotation in AdWords to optimise your creative, your brand positioning and more.  Breaking out your data by device and looking at conversion rate can allow you to prioritise development.

Looking at search queries over time and pulling this into word clouds can reveal real insight.  You may have to strip out the main head terms to find some real gems.

To analyse your offline marketing, look at your branded search volume.  Doing this over time versus activity, allows you to see the performance of offline in terms of brand awareness/search volume.

Test offers, free delivery and all aspects by A/B testing in your ad creative - and check conversion rate.  Use Paid Search as a testing tool to learn about new product ideas - even test fake products to see if there’s demand before you develop or purchase new items.

Jennifer Veesenmeyer, Merkle Analytics: How To Master The Art of Dashboard Design

Jennifer of Merkle Analytics spoke on dashboard design. A dashboard is not a dashboard unless it has strong visual elements, it’s not a good dashboard unless it fits on one screen and provides context on KPIs, substance and style are equally important, an automated dashboard is better than a perfect one, executive dashboards are highly visible so a well designed one can boost your career.

Select the most meaningful metrics - this may include combining other metrics to make useful metrics.  Don’t show both visits and visits per visitor for instance, show the relationship between the two.  Provide important context and normalise the data.

Include insights, not narration. Don’t provide reporting, provide insights.

When designing a dashboard, place your most important metrics at the very top, with least important in the bottom right.  The most important element should be the biggest.  If you don’t know which are most important, go back to the site.  Font size should also follow a hierarchy.

Select the right visualisation for the job.  It depends if you are looking at time series or point in time data.  Column chart for comparisons, line graphs for trends.  Jennifer also recommend stacked area charts over stacked column charts.

If you want to compare data on a static point in time, use a bar graph.  Humans are wired to compare distance rather than area which is why bar charts are better than pie charts for comparisons.

Sponsor Lightning Round

Sweetspot, the University of California, Evidon, OpinionLab, University of British Colombia, Comscore and Tracking First all presented for 2 minutes on their products and services.

Nico Miceli, SEER Interactive: Tracking The Untrackable

Next, Nico Miceli of SEER interactive spoke about tracking the untrackable with Google Analytics.

Nico covered the Obama campaign’s use of data, and then went on to discuss Universal Analytics and the ability to send any data into the platform.  SEER connected up a Raspberry PI and sensors to the bar at an event to track activity in real time.

For as little as $60 you can set up systems to track any data, and send it into Google Analytics to allow you to track activity.

Example use cases could be barcode scans, conference booth activity and more.  You can use this technique to track customer lifespan in more detail - in a conference for example tracking normally stops at the purchase of a ticket. Instead, you can track the users experience of the conference (barcode scans, activity tracking, presentations viewed), and then whether they buy a ticket for next year.  It all comes back to thinking about customers rather than visits.

Susan Etlinger, Altimeter Group: Social Data Intelligence: Integrating Social And Enterprise Data For Competitive Advantage

Susan argued that thirteen departments are actively engaged in social media.  Organisations are beginning to integrate social and enterprise data and look at things more holistically.

If social data is now business critical, it needs to be enterprise class.  Users have to be able to use the data efficiently and at scale to make decisions.

There are a lot of challenges in integrating social data.  To make social data actionable you need to identify your business goals.  ROI is a key metric, but focus tends to be on the R rather than the I.  Key social media measurement compass includes brand health, marketing optimisation, revenue generation, operational efficiency, customer experience, and innovation.

Prioritise your metrics with a metrics scorecard.   CTR is objective, sentiment is subjective - think about the strength of your metric when prioritising them.

Daniel Mintz, Upworthy: Teaching Less Nerdy People How To Win Through Smarter Testing

Daniel went through how leaders need to change their thinking in a data led organisation.  He argued that leaders need to be prepared to have a culture of data whereby they are happy to be overruled by the data and not rely on their opinion as the highest paid person (HiPPO).

In terms of co-workers who are non-nerds, they need to have the right expectations in terms of data.  Data is a tool, it’s not the solution to everything everytime.

Co-workers need to tell you what they’re seeing as they are often much closer to the business than you are.  Let them create your hypothesis for testing.

Unknown “knowns” are a major threat as if someone believes a hypothesis - whether it’s from an old or inaccurate test, they plough on, even if it may not be true.

Beware of caveats with your data - they often get lost as data is passed down the command chain.

Andrew Janis, General Mills: Evolving General Mill’s Data Driven Culture

Key problems at General Mills three years ago were multiple data sources, antiquated tools, no clear owner, and minimal data governance.  Nobody believed their data.  At first they needed to get the small data right to build trust in the data.

The second issue was that the business didn’t see how the data could help sell more goods.  Andrew’s team did a pilot to prove the value of the data.

At General Mills the focus is on answering questions by using food trend analysis.  They have over 500 websites which allow them to pull data by lots of different user segments.

Phil Mui, Axciom; Aurelie Pols, Mind Your Privacy; Adi Kamdar, Electronic Frontier Foundation: What Do ‘They’ Know About You

Adi Kamdar posed a series of questions around privacy to provoke the thoughts of the audience.  Aurelie spoke on the differences between European and American views on privacy law.  Privacy fines are increasing a lot at the moment, and knowing the laws around data usage is crucial.

The session then opened up into questions from the audience.

Alhan Keser, WiderFunnel Marketing Optimization: Developing An Effective Conversion Optimisation Strategy

Alhan of WiderFunnel, discussed the concept of F.A.T - frivolous A/B testing.  This includes relying on things you read online, continually testing when it doesn’t count, and not taking long-term learnings from your tests.

Put tips and tactics you’ve heard toward the end of your testing strategy.  Think about your own problems without looking at other peoples solutions - as they solved their problems, not yours.

Focus on meeting your audience (make conversion tests based on what their needs are) & testing where it hurts.  Don’t rely on user tests, they’re an artificial environment and therefore the data is skewed.  Don’t let human stories overpower the data when planning conversion tests.

Alhan argued that we need to use data to find true tests, rather than test the colour of buttons and trust icons etc.  Ask the audience what their issues are, understand the business, and solve problems by using A/B testing rather than potentially creating problems.

Ryan Sleeper, Evolytics: Data Driven Storytelling

Ryan’s talk was entitled data driven storytelling - tips from the Tableau iron viz champion.  Ryan went through 15 data visualisation tips using tableau.

Gary Angel, EY: Big Social Media Data

Gary argued that digital analytics is done very different to traditional customer analytics in approach - whereas it shouldn’t be.

Big data isn’t about data, it’s about order, time and pattern - that’s the difference.  Meaning comes from order, time and pattern when viewed in detail.  This data didn’t exist properly before, but now it is available and much easier for people to access.

Ben Harper, Datify: Using Data To Lower Your Facebook CPA

A full transcript of Ben’s article, including slides, can be found on our blog.