Heat map

From Wikipedia, the free encyclopedia

Heat map generated from DNA microarray data reflecting gene expression values in several conditions

A heat map is a graphical representation of data where the individual values contained in a matrix are represented as colors. Fractal maps and tree maps both often use a similar system of color-coding to represent the values taken by a variable in a hierarchy. The term is also used to mean its thematic application as a choropleth map. The term "Heatmap" was originally coined and trademarked by software designer Cormac Kinney in 1991, to describe a 2D display depicting real time financial market information.^[1]
Heat maps originated in 2D displays of the values in a data matrix. Larger values were represented by small dark gray or black squares (pixels) and smaller values by lighter squares. Sneath (1957) displayed the results of a cluster analysis by permuting the rows and the columns of a matrix to place similar values near each other according to the clustering. Jacques Bertin used a similar representation to display data that conformed to a Guttman scale. The idea for joining cluster trees to the rows and columns of the data matrix originated with Robert Ling in 1973. Ling used overstruck printer characters to represent different shades of gray, one character-width per pixel. Leland Wilkinson developed the first computer program in 1994 (SYSTAT) to produce cluster heat maps with high-resolution color graphics. The Eisen et al. display shown in the figure is a replication of the earlier SYSTAT design.
There are different kinds of heat maps:

• Web heat maps have been used for displaying areas of a Web page most frequently scanned by visitors. Web heatmaps are often used alongside other forms of web analytics and session replay tools.
• Biology heat maps are typically used in molecular biology to represent the level of expression of many genes across a number of comparable samples (e.g. cells in different states, samples from different patients) as they are obtained from DNA microarrays.
• The tree map is a 2D hierarchical partitioning of data that visually resembles a heat map.
• A mosaic plot is a tiled heat map for representing a two-way or higher-way table of data. As with treemaps, the rectangular regions in a mosaic plot are hierarchically organized. The means that the regions are rectangles instead of squares. Friendly (1994) surveys the history and usage of this graph.

There are many different color schemes that can be used to illustrate the heatmap, with perceptual advantages and disadvantages for each. Rainbow colormaps are often used, as humans can perceive more shades of color than they can of gray, and this would purportedly increase the amount of detail perceivable in the image. However, this is discouraged by many in the scientific community, for the following reasons:^[2][3][4][5]

• The colors lack the natural perceptual ordering found in grayscale or blackbody spectrum colormaps.
• Common colormaps (like the "jet" colormap used as the default in many visualization software packages) have uncontrolled changes in luminance that prevent meaningful conversion to grayscale for display or printing. This also distracts from the actual data, arbitrarily making yellow and cyan regions appear more prominent than the regions of the data that are actually most important.
• The changes between colors also lead to perception of gradients that aren't actually present, making actual gradients less prominent, meaning that rainbow colormaps can actually obscure detail in many cases rather than enhancing it.

Contents

• 1 Software implementations
• 2 Examples
• 3 References
• 4 External links

Software implementations

A sample heat map created using a Surface Chart in Microsoft Excel.

Several heat map software implementations are listed here (the list is not complete):

• PermutMatrix is a work space designed to graphically explore numerical datasets. It offers several methods for the optimal reorganization of rows and columns of a numerical dataset.^[6]
• NeoVision Hypersystems, Inc., a software firm founded by Cormac Kinney, and funded by Intel and Deutsche Bank, developed Heatmaps depicting real time financial data and calculations, which were licensed to over 50,000 users. NeoVision Heatmaps became a feature on nasdaq.com.^[7]
• R Statistics, a free software environment for statistical computing and graphics, contains several functions to trace heat maps [1]
• Gnuplot, a universal and free command-line plotting program, can trace 2D and 3D heat maps [2]
• The Google Docs spreadsheet application includes a Heat Map gadget, but for country-wise data only, not for general matrix data.
• Dave Green's 'cubehelix' colour scheme provides resources for a colour scheme that prints as a monotonically increasing greyscale on black and white postscript devices [3].
• Qlucore includes a heat map that is dynamically updated when filter parameters are changed.
• The ESPN Gamecast for soccer games uses heat maps to show where certain players have spent time on the field.
• By searching the List of bioinformatics companies more tools for heat maps can be found.
• Microsoft Excel can be used to generate heat maps using the Surface Chart. Though the default color range for Surface Charts in Excel is not conducive to heat maps, the colors can be edited to generate user-friendly and intuitive heat maps.
• Sightsmap is a sightseeing popularity heatmap overlaid on Google Maps, based on crowdsourcing: the number of Panoramio photos taken at each place in the world.
• Maptitude is business mapping software that includes a variety of customizable heat mapping tools and can use external data such as Excel files to show the results on geographic maps of your location.

Examples

• 

  Geographical heat map of ocean salinity, using a rainbow colormap
  
• 

  Lake effect snow - weather radar information is usually shown using a heatmap.
  
• 

  Choropleth heatmap showing election results by municipality
  
• 

  Human voice visualized with a spectrogram; a heat map representing the magnitude of the STFT. An alternative visualization is the waterfall plot.
  
• 

  Example showing the relationships between a heat map, surface plot, and contour lines of the same data
  
• 

  Combination of surface plot and heatmap, where the surface height represents the amplitude of the function, and the color represents the phase angle.
  

References

1. "United States Patent and Trademark Office, registration #75263259". 1993-09-01.
2. Borland, D., & Taylor, M. R. (2007). Rainbow Color Map (Still) Considered Harmful. IEEE Computer Graphics and Applications, 27(2), 14-17. IEEE Computer Society. Retrieved from http://www.ncbi.nlm.nih.gov/pubmed/17388198
3. How NOT to Lie with Visualization - Bernice E. Rogowitz and Lloyd A. Treinish - IBM Thomas J. Watson Research Center, Yorktown Heights, NY
4. Mark Harrower1and Cynthia A. Brewer - ColorBrewer.org: An Online Tool for Selecting Colour Schemes for Maps, The Cartographic Journal Vol. 40 No. 1 pp. 27–37 June 2003
5. Green, D. A., 2011, `A colour scheme for the display of astronomical intensity images', Bulletin of the Astronomical Society of India, 39, 289. Dave Green's `cubehelix' colour scheme
6. Caraux, Gilles; Pinloche S. (2005). "PermutMatrix: a graphical environment to arrange gene expression profiles in optimal linear order". Bioinformatics. 7 21: 1280-1281. doi:10.1093/bioinformatics/bti141.
7. Sansoni, Silvia (1999-05-17). "Forbes Magazine Article on NeoVision Heatmaps".

• Bertin, J. (1967). Sémiologie Graphique. Les diagrammes, les réseaux, les cartes. Gauthier-Villars.
• Eisen, M.B., Spellman, P.T., Brown, P.O. & Botstein, D. (1998). "Cluster analysis and display of genome-wide expression patterns". Proc. Natl. Acad. Sci. USA 95 (25): 14863–14868. doi:10.1073/pnas.95.25.14863. PMC 24541. PMID 9843981.
• Friendly, M. (1994). "Mosaic displays for multi-way contingency tables". Journal of the American Statistical Association (American Statistical Association) 89 (425): 190–200. doi:10.2307/2291215. JSTOR 2291215.
• Ling, R.F. (1973). "A computer generated aid for cluster analysis". Communications of the ACM 16 (6): 355–361. doi:10.1145/362248.362263.
• Sneath, P.H.A. (1957). "The application of computers to taxonomy". Journal of General Microbiology 17 (1): 201–226. doi:10.1099/00221287-17-1-201. PMID 13475686.
• Wilkinson, L. (1994). Advanced Applications: Systat for DOS Version 6. SYSTAT Inc. ISBN 978-0-13-447285-0.

External links

• The History of the Cluster Heat Map. Leland Wilkinson and Michael Friendly.

Mobile Analytics - Nuevas Oportunidades y Retos

Digital Analytics - University of Utah

Mobile Analytics - Nuevas Oportunidades y Retos

 La tecnología Web Analytics ha existido por muchos años, y seguramente este sistema de seguimiento para la retroalimentación y mejora es un sistema importante y funcional para el seguimiento y la personalización de los resultados de la presencia web de una empresa. Por otra parte, con la mejora continua en el rendimiento y personalización, que seguirá siendo una importante fuente de información para el futuro previsible. Otra área de monitoreo está emergiendo y deberá recibir la mayor cantidad - si no más - Atención: Mobile Analytics.
 Los dispositivos móviles son un fenómeno relativamente nuevo y sin duda creciente segmento del mercado de la web de visualización. Estos incluyen, por supuesto, los teléfonos inteligentes (Android, Iphone, Blackberry, Windows), dispositivos de tableta (nuevo Android y iPad), dedicado dispositivos móviles como el nuevo libro Chrome, e incluso los pocos teléfonos no inteligentes restantes que pueden ver el contenido web . Mientras que muchos de estos dispositivos son capaces de ver las versiones de contenido total de páginas, a menudo de forma predeterminada una versión móvil. Esto se suele hacer a los hechos simples que las pantallas tienden a ser más pequeñas, la interfaz táctil sobre la mayoría de estos dispositivos hacen pequeños eslabones más difícil interactuar con, y las consideraciones de ancho de banda (costo de los datos móviles, así como la velocidad). Por esta razón, Google Analytics con una mirada específica hacia el mercado de telefonía móvil deben ser tratados.
 Ciertamente Mobile Analytics ofrecen algunos desafíos en comparación con los más tradicionales de Web Analytics. Una de las funciones más céntricas de Analytics es la capacidad de identificar a un visitante del sitio como devolver nuevo y exclusivo. Hay muchas técnicas para el manejo de esto en Analytics tradicionales, pero algunos de ellos no pueden trabajar en el ámbito móvil. Muchos navegadores móviles no ofrecen funcionalidad javascript. Si una herramienta de análisis utiliza esta, puede resultar inútil en el ámbito móvil. Las cookies son ampliamente funcional ahora, pero que fácilmente se puede desactivar (esto no es realmente diferente de un navegador no móviles).
 Si una dirección IP de los dispositivos se utiliza como un medio de identificación del visitante, este presenta su propio conjunto de problemas. La IP de un dispositivo móvil puede ser asignada por el ISP puerta de ese dispositivo. Por otra parte, los dispositivos móviles pueden cambiar de proveedor en una sola sesión de navegación. Digamos, por ejemplo, que usted visita un sitio de comercio electrónico, mientras que en un restaurante. Usted no termine su sesión antes de volver a meterse en su coche. Su coche tiene su propio punto de acceso, y en interés de la seguridad, su cónyuge está conduciendo. El teléfono cambia al punto de acceso del coche. A continuación, entrar en su casa y su teléfono se entrega a su hogar wifi. Usted ha pasado ahora a través de una sola sesión de tres canales distintos (posiblemente cuatro si el restaurante tenía wifi). Es evidente que este es un ejemplo extremo. La ilustración más importante es completado una sesión en una red y otra en una red diferente. ¿Sus análisis ver al cliente como un nuevo visitante?
 Soluciones móviles Analytics tienen otras formas de seguimiento de los individuos. La más interesante de ellas es el uso de características específicas del propio dispositivo. Fabricante, modelo, memoria, resolución de pantalla, etc se puede seguir para ayudar en la identificación (por otra parte, estos criterios no desaparecen cuando se elimina una cookie). También los modelos de dispositivos que visitan el sitio es útil para realizar un seguimiento como el sitio puede ser optimizado para el hardware de los dispositivos que visitan el sitio más a menudo.
 Con el creciente número de dispositivos inteligentes, la calidad de contenido móvil es extremadamente importante. Se debe prestar atención a ajustar la experiencia en el sitio móvil. Debido a que hay páginas específicas para móviles, mientras que las páginas de contenido completo todavía están disponibles, los datos de la tasa de rebote pueden tomar un nuevo significado. Un visitante puede visitar un sitio, ya que el sitio detecta un navegador móvil, la página por defecto de destino para el contenido móvil. Después de un par de toques que el usuario puede llegar a ser frustrado con la experiencia y solicitar el sitio completo. Porque se ha producido clics, esto no se ha registrado como un "rebote", pero es muy importante hacer un seguimiento. Se representa un fracaso de la versión móvil. Por lo tanto se requiere una nueva métrica.
 Por último, existe también la posibilidad de aplicar la analítica para aplicaciones móviles. No son los navegadores de contenido web, pero muchos son los usuarios de contenido web. Uso en línea de una aplicación móvil puede ser seguido (por ejemplo, aplicación de ebay, amazon app Kindle aplicación, etc.) Además muchas de las aplicaciones móviles se pueden utilizar sin conexión. La actividad puede ser almacenada y transmitida al anfitrión la próxima vez contenido en línea se accede.
 Esto no es en absoluto una discusión exhaustiva o completa de análisis móvil, más que una introducción general al tema en su conjunto con reflejos de algunos de los retos y las nuevas recompensas potenciales de aplicar análisis específicos sobre el terreno. A medida que el campo de los dispositivos, la conectividad y la gama de tipos de nuevas aplicaciones se expande, este campo de los teléfonos Analytics sin duda crecer con ella.

Gordon Oremland

Measuring Social Media ROI With Google Analytics

Measuring Social Media ROI With Google Analytics

• inShare39

Social media is an integrated component of so many companies' marketing strategies, but very few of us are tracking the ROI of our social initiatives. Since Google Analytics has released social media reporting, it has streamlined the way we can report onsite social media conversions and value.

In this post we will cover best practices for ensuring you have accurate and actionable data inside Google Analytics and also step outside of the tool to complete the full picture and establish onsite and offsite metrics to help us report on our success and most importantly, drive action.

Below is a quick video where Justin Cutroni, Analytics Advocate at Google, and I share some ideas on Social Media measurement, just to open the appetite!

Measurement Challenges: Multiple Data Sources and Offline Impact

There are a number of challenges we encounter when trying to measure and report on our social media initiatives. Our first challenge is that the majority of interactions don't actually occur on our website and instead happen on the individual social network. For example, when somebody 'likes' our content on our Facebook page we can't get this data into our Google Analytics reports.

This leads to our next challenge which is aggregating data from all these different data sources in order to report on performance and perform analysis. For a small business this could mean manually grabbing data from the different social networks and working in Excel or a Google Spreadsheet, but for larger companies this would become a debilitating process.

Our third challenge is to understand the offline impact of our social media campaigns. What do we do if someone ends up purchasing in-store? How do we understand the value of a phone call? Or what if someone tells a friend or colleague about our brand after seeing a post on a social network? These are much harder questions to answer.

Data Collection: Tagging Social Campaigns

Before we look at reporting we need to start by taking a moment to ensure we have the best possible set of data available. You will need to ensure you are using Google Analytics campaign tags to correctly track people who are clicking through to your website from your social media campaigns.

For paid social ads or promoted posts, my recommendation is to always set the campaign source as the domain of the social network. For example, for Facebook we would use facebook.com and for LinkedIn we would use linkedin.com as this is what you will already see appearing inside your traffic sources reports. I would then always set the campaign medium as social. Next, you will need to define a campaign name to distinguish the particular paid social network campaign you are running and you can use content to report on the ad variation or call-to-action people are clicking on to then travel through to your website. For example:

It is best to leave the term blank when creating your campaign tagged URL as this can potentially show up in your keyword reports. You will also want to change the content for each ad variation you are running and the campaign name to suit your reporting needs.

You should also consider using campaign tagged URLs for your organic posts on social networks. This is because when someone clicks through to your website within a social network app, like the Facebook app on your phone, the visit will be seen as direct traffic instead of being a referral from Facebook. This means you could be missing a large portion of your social network traffic inside your Google Analytics reports. To tag your organic initiatives you can use a similar tagging method, but change 'social' to 'referral' in the example above.

Getting Your Data: Social Data Collection and Reporting Tools

Now we have tracking in place for our website and we can see data for all of our inbound social network traffic we need to look at pulling together data from our website and importantly from the individual social networks too. Lets look at a few options for getting all of our data.

The first option is going to be the most painful and that is to manually log into each social network and make use of the data they make available to us. For example, we can head over to our Twitter account to find our total number of followers and browse through our posts to find ones that were retweeted and received comments. We can also head over to Facebook Insights and export data for a particular time period. However this is not practical and with Twitter we are actually missing data about how many clicks our posts have received and we can't select a data range for reporting. So that isn't a very good option.

Now we need to look at tools that can pull all of this data for us. There are a huge number of options ranging from expensive to accessible. My preference is always tools that are easy to use and get the job done quickly and these tools are also generally cost effective too.

1. Hootsuite is a social management tool that also allows you to report on your social media initiatives. Hootsuite integrates with Google Analytics, but unfortunately the reporting on individual posts is limited and does not include the reach (total audience size) of your posts.
2. Simply Measured is a reporting tool that allows you to pull together data from different social networks and from your Google Analytics reports. You can access reports online or using Excel. The solution is flexible, but does start at $500 per month, so might be restrictive for smaller organizations.
3. Sprout Social is another social management tool with a more complete set of integrated reporting and also includes integration with your Google Analytics reports. Sprout Social allows you to quickly create a report for a given time period for all your posts for a particular social network. The report includes clicks, responses (including replies, comments and shares) and reach (audience size). It's certainly not perfect, but it is an accessible tool that will help you be in control of your social network data.
4. Buffer is a tool that allows you to schedule your social media posts. It is a simple tool and doesn't provide a full set of management features, but Buffer does make individual post analytics available. Unfortunately, you can't export the analytics to easy aggregate the numbers, but you will find the potential reach of each post and engagement metrics like clicks and retweets for each post you have published.

The options we have looked at so far give us access to our organic social media data, but what about our paid campaigns? You can look towards enterprise-level solutions for aggregating paid and organic data, but grabbing data from our paid campaigns is fairly straightforward and more importantly available via the major social networks. For example, you can download your advertising data from Facebook, LinkedIn and Twitter with very little effort.

Tip: Once you have downloaded your advertising data you should consider uploading the impression, click and cost data into your Google Analytics reports using Cost Data Upload, but this is a topic for another post as we will focus on pulling together all of our data outside of Google Analytics.

This of course brings me to Google Analytics. We need to extract our onsite data for our social network traffic together with our goal conversions and any ecommerce value we are tracking. Before we really dig into the data, lets look at the metrics that we can use for reporting and analysis.

Metrics That Matter: Actionable Social Reports

There are a number of different metrics you can use to report on various aspects of your social media initiatives. I encourage and urge you to not use every single metric I will cover here, but instead focus on the few that help you drive action to improve your campaigns and drive greater value for your website.

Branding Metrics

For purely branding objectives you might report on number of posts and audience size, but this is boring and definitely not actionable. Consider expanding on these metrics by including posting change rate where you calculate the percentage change based on the number of posts you are creating in a given month. This can then be compared to audience change rate where you calculate the percentage change in your audience size. This will allow you to quickly see if an increase in posts leads to an increase in audience.

Engagement Metrics

To report on engagement you might report on the total number of people liking your posts, but you should really focus on metrics that help you drive action. Start by reporting on the conversation rate for each post, this is where you divide the number of interactions by the audience reach for each of your posts. By identifying your posts that generate the highest conversation rates you can use these topics to guide your future posts and tailor content to what your audience enjoys. You can also report of the total number of conversations, mentions and shares you are starting to paint a more complete picture of how people are engaging with your content.

To better understand your overall engagement with your audience you can report on audience churn rate to see the percentage of your audience that is leaving your social networks. By segmenting your audience churn rate by individual social network and correlating churn to particular posts or post topics you can help shape your future posts to ensure they are appropriate for your different social network audiences.

To understand how social networks are leading people to your website you can report on the total number ofsocial network visitors to your website. Consider adding social network visitor growth rate to easily compare your other social media metrics and identify if changes are resulting in a trickle down effect to your website. For example, you will be able to compare your audience change rate to your social network visitor growth rate to see if you are having a positive impact in driving visitors to your website.

Conversion Metrics

To report on conversions resulting from your social media initiatives you need to ensure you are correctly tracking the inbound links directing people to your website. From here you can use Google Analytics to report on your social media conversions and by using the campaign tagging method we covered previously, you will be able to split your paid and organic efforts when reporting.

We should all be reporting on the total conversions occurring onsite and the value of these conversions. If you haven't already established values for your conversions, then you will need to take the time to do this, I recommend reading Avinash's post on identifying values. This will allow you to report on the direct dollar value of your social channels.

There are also methods for tracking offline conversions that are being driven by social media, including coupons and dedicated phone numbers. To keep this post on track, we are going to focus our attention to online conversions, but once you have your offline tracking in place you can simply replace or supplement your offline conversions and value with your online conversions.

Crunching The Numbers: Establishing Social Value

In order to calculate the dollar value of each of our audience members we are going to use Dan Zarella's formula for calculating the value of a like. This allows us to use our conversion data to calculate an average potential value for each of our audience members. Here is Dan's formula:

L / Upm x (LpD x 30) x (C / L) x CR x ACV = Value of a Like

Where:

• L is Total Likes
• UpM is Unlikes per Month
• LpD is Links per Day
• C is Average Clicks per Post
• CR is Conversion Rate
• ACV is Average Conversion Value

In essence we take our onsite conversion data and combine this with data about our social network audience. This allows us to define a value even if people don't click through to our website. Then using this value we can calculate the ROI of our social media channels.

Social Media ROI Reporting Template

So you don't have to create your own spreadsheet you can use the one I have developed to speed up your social media ROI calculations: http://goo.gl/EByA4. You will need to enter some data for each of your social networks and from there it will automatically calculate your ROI.

Tip: If you are struggling to fill in the data, head back to 'getting your data' in this post for some tips on getting your social network data.

Now you can see the ROI of each of your social networks. Great! The spreadsheet includes a 'Total ROI' section where you will see ROI calculated based on your total audience. There is also an 'Incremental ROI' section where ROI is only calculated based on the new audience members you have acquired for the period. This gives you some flexibility on how your report, some of us might consider our investment as part of social media retention, while others might be more concerned about driving audience growth. Choose a section that best fits with your strategy and style of investment.

Get the Social Media ROI Reporting Template

Factoring in 'Friends of Friends'

If you are running ads on social networks you can also use this data to help factor in the effects of your content being shared and your brand being mentioned by your audience. This allows you to establish an approximate value of your 'friends of friends'.

Start by identifying the total audience size of a recent paid campaign you have run. For example, if you ran a Facebook campaign you will be able to find the campaign reach, which is the number of people who saw your ads. You will also have the number of page engagements. By dividing the number of page engagements by the campaign reach you can calculate the engagement rate, which is the percentage of the audience that engages with your page and page content. We can then apply our paid engagement rate to our 'friends of friends' to roughly calculate the number of people we can add to our organic audience.

This is certainly not perfect, but it is a good starting point to help us establish a value for our extended organic audience. Ideally the paid audience data you use for the calculations would have similar characteristics as your organic audience, for example their interests and other demographics. In a perfect world we would also only use page engagement data from our paid campaign that results in likes. If you don't have a large enough campaign, then go with the data you have available, this might include people viewing photos and other less valuable page interactions.

Let's say you have calculated your paid engagement rate and find that it is 0.052%. You can then head over to Facebook insights (or your tool of choice) to find your total number of friends of fans. If you have 500,000 friends of fans and an engagement rate of 0.052%, then you have the potential to add an additional 260 fans to your audience. Now you can take the average value of an individual audience member and multiply it by the number of potential fans to find the additional audience value of your 'friends of friends'.

Further Analysis

Now that we have established an ROI figure for each of our social channels you can begin to focus on improving your results by allocating your resources appropriately. In order to turn your analysis into action you should also begin to explore your social media efforts in more detail by performing content analysis for your individual posts.

Content analysis will enable you to understand the post that are most engaging and most successful at driving value, so you can repeat what works to more effectively leverage and grow your audience. Consider analysing your content by aggregating posts by their theme, tone and content, you can also consider grouping posts that contain links, questions and even different capitalization and punctuation. This will help speed up your analysis and help define the best content strategy for your social campaigns.

Now it’s over to you to start reporting on your social media ROI! I would love to hear your feedback, so please add your thoughts in the comments below.

Related Content

1. The Definitive Guide to Google+ Analytics
2. Measuring Social Media Impact with Web Analytics
3. Social Media Measurement with Google Analytics [video]

Submitted by Benjamin Mangold on April 2013

About Benjamin Mangold

Benjamin Mangold is the Search & Analytics Director at Loves Data, a digital analytics and online marketing agency based in Australia. He lives and breathes Google Analytics. Follow him on Google+ or Twitter.

51 Tips To Succeed With Web Analytics.

51 Tips To Succeed With Web Analytics

• inShare45

Investments in Web Analytics can pay off big, but the degree of success depends on many factors. Simply said you need the right set of people, measurement tools and a clearly defined processto succeed.

It is impossible to provide you with all the advice you need in one article. In my experience, it comes down to consuming and sharing as much relevant information as you can. This, combined with many years of practical learning experiences in this magnificent field.

One tip beforehand: invest a few bucks and build a website yourself. Write about something you love. This will greatly enhance your learning curve! It is not limited to improving your Web Analytics skills, but touches on all the other online marketing disciplines as well.

Many different Web Analytics frameworks are out there; for the purpose of this article, I use the following 'five phases' framework:

Sit down and relax, here we go!

Phase 1: Pre-Implementation

You can't just start with implementing web analytics tags on your website and think great success is lying ahead of you. A lot of things need to be in place first. I will give you 10 tips to take into account in this first phase.

Tip 1: Define clear objectives for your specific web business first; without a clear set of online business objectives you are doomed to fail.

Tip 2: Set the scope of the project; does it involve one domain or 50 different domains?

Tip 3: Build a roadmap; what and when do you want to achieve something? Place milestones when needed.

Tip 4: Identify the stakeholders in the process; usually a lot of people need to get involved. Work on a clear overview of all the stakeholders.

Tip 5: Divide the responsibilities; who is responsible for what and when? It is crucially important to get this right.

Tip 6: Define your goals, KPI's, segments and targets right from the start; remember, this is an important, evolutionary process. It is never perfect, but you need to know where to focus on.

Tip 7: Define specific reporting requirements; don't become a reporting squirrel, but know right from the start what kind of data people in your organization are looking for.

Tip 8: Make wise budget decisions; spend enough money on people and not only on tools. Great, expensive tools won't do if there is no one there to make the data actionable. Start with free or cheap tools first.

Tip 9: Define micro and macro conversions; you may have one major conversion on your website, but there are more conversions tied to your business goals.

Tip 10: Decide together on the KPI's; your KPI's need to be widely agreed upon.

Let's continue with the second phase: the implementation.

Phase 2: Implementation

The technical phase is probably not the sexiest one for you, but you don't want to pass this one too quickly. These tips will guide you through the implementation phase.

Tip 11: Always reserve extra time; things never go exactly as planned.

Tip 12: Take into account the release planning; know whether tags can be placed within a few days or a few months.

Tip 13: Reserve a small budget for implementation testing tools; especially on larger sites tooling can be very useful.

Tip 14: Don't just handover a document, stay close to the implementation traject; IT and marketing need to work closely together.

Tip 15: Tag all your pages; you can't measure what you don't tag. "Is it OK if I only tag this part of my website?" It won't be the first time someone asks this question.

Tip 16: Customers are more important than tags; tagging is crucial, but make sure it doesn't negatively impact user experiences on your website.

Tip 17: Don't forget rich media experiences (Flash, Flex, RSS, Videos etc.); measuring in-page interactions is increasingly important.

Tip 18: Setup a testing environment; test in a different environment whether the tags are working or not.

Tip 19: Always triple-test implemented codes; it's better to take one extra day for testing than to go live with a bug.

Tip 20: Schedule maintenance periods; tags can easily disappear from your pages. Make sure you monitor this automatically or schedule periodic manual checks.

Phase 3: Configuration

By now you have clearly identified what you want to measure and all the tags are in place. Now it's time to setup your Web Analytics package on the admin side. What is important to consider? Phase three and four are mainly focused around Google Analytics, since it is the most widely used tool.

Tip 21: Limit the number of administrators to a minimum; setup your Google Analytics permissions in the right way.

Tip 22: Setup a master profile with raw, unfiltered data; in case something goes wrong you always have a backup profile.

Tip 23: Use Google Analytics profile filters for long term segmentation purposes; it helps you to optimize user experiences and conversions for different segments.

Tip 24: Build advanced segments for ad hoc data analysis; it helps you to uncover great segments that need to be targeted in a unique way.

Tip 25: Setup goals and funnels; essential to optimize your traffic sources, campaigns, keywords etc. on a certain outcome.

Tip 26: Setup goal values for non-transactional conversions; this will help to get a complete picture of the value per visitor for a specific segment.

Tip 27: Tag your Marketing campaigns carefully; ignore this phase and your data becomes meaningless or even worse, you make the wrong decisions.

Tip 28: Integrate a set of KPIs in a custom report; combine acquisition, behavior and outcome metrics.

Tip 29: Connect Google Analytics to external tools; this is very useful to derive insights more easily. For example, think about connecting Google Analytics to Next Analytics to automate reporting and free your time for analysis and optimization.

Tip 30: Setup intelligence alerts; uncover hidden correlations and causalities.

Phase 4: Analysis

Great, you have everything in place and two months of data are right there. It's time to earn some money. Start doing great analysis on your data!

Tip 31: Start every analysis with a question; know what you want to solve or improve first.

Tip 32: Don't focus on aggregates, always segment the data; one solution doesn't fit the need of all your visitors.

Tip 33: Don't focus on averages, always look at distributions; what if the average customer satisfaction is rated with a seven, but 25% of your customers are highly unsatisfied?

Tip 34: If you have time for only one analysis focus on the "All Traffic Sources" report; you will immediately get a clear overview on how your website and channels are performing.

Tip 35: Enhance quantitative analysis with qualitative analysis; solve for the 'what', the 'why' and the 'how'. The 'what' is not telling the complete story.

Tip 36: Visualize your Web Analytics data; it may help you to get your message across.

Tip 37: Use custom variables for multi-session analysis; how does a brochure download impact my hotel booking rate?

Tip 38: Know the difference between goal conversions and transactions; goal conversions can only happen once in a visit, transactions can happen multiple times.

Tip 39: Don't focus on raw, absolute numbers; put your data in perspective.

Tip 40: Accept that data is never perfect; the truth lies in your back-office.

Phase 5: Testing and Optimization

You have identified a few major issues on your website and some of your landing pages have a very low conversion rate. You might wonder, how can I improve? Read on and apply the following 10 tips to your specific situation.

Tip 41: Don't think you know better than your website visitors; the best optimization specialist in the worldnever beats your visitors.

Tip 42: Setup a wordclass testing team; you may need developers, designers, analysts and usability consultants to succeed.

Tip 43: Identify the most crucial pages to start testing with; optimize highly trafficked landing pages and funnel pages first.

Tip 44: Clearly define your test, hypothesis and goals; what are you testing, what is the expected outcome and what needs to be improved? Answer those questions first.

Tip 45: Spend healthy budgets on conversion optimization as compared to acquisition; driving lots of untargeted traffic doesn't make sense.

Tip 46: Select a testing idea on expected effect, duration and available resources; this will help you to get the highest ROI on your testing efforts.

Tip 47: Add the HIPPO's opinion as one of the testing variations; beat the Highest Paid Person's Opinion with numbers.

Tip 48: Apply the Conversion Trinity rules to your landing pages; think about relevance, value and the right call to action.

Tip 49: Use the right set of tools; automate what you can automate.

Tip 50: Set your confidence interval limit at around 95%; don't choose a winner too early!

Closing Thoughts

One more, my first and last tip: start a website and play in the real world! Apply these tips in your daily activities and I am 100% sure you will grow in the Web Analytics field.

Any great tips or experiences to share? We are happy to publish them. If you like the article, we very much appreciate a comment or share!

Submitted by Paul Koks on April 2013

About Paul Koks

Paul Koks leads the Analytics and SEO division of a Search Marketing Agency in The Netherlands. He is the owner ofOnline-Metrics.com, a blog about Web Analytics and Conversion Optimization. You can find him onTwitter, Google+ or LinkedIn.