- Perry Williams: Hello Dear, I am strongly agree with your point that the web analytics is associated with the social...
- Philip Sheldrake: Nice overview Marianina. I wanted to post a link to an article in Business Week from June about the...
- Luisa Woods: Hi Marianina, I think you make a very good point about the importance of segmentation. I like to carry...
- Eric T. Peterson: Marianina, Nice to have seen you Monday in London! I just got this post so perhaps something odd is...
- Marianina Manning: Hi Luisa, Thanks for your thought-provoking comment! I agree that new ways of looking at web...
- 7 Ways to make web analytics work better in companies
- Measuring social media, influence, debate, buzz monitoring
- Web analytics winners and losers? It’s the people that make the difference.
- Simple segmentation for your website and better web analytics understanding
- Web Analytics Wednesday in London – the future of web analytics
- Digital cream: revealing debating at econsultancy’s marketing event
- Google Analytics Tip: Ecommerce tracking set up, screenshots and why it’s useful
- Reliving my customer’s experience and some nice screenshots
- Internal site search part 2
- The best charts ever and food for thought for us web analysts
- August 2008 (1)
- May 2008 (1)
- April 2008 (1)
- March 2008 (3)
- February 2008 (2)
- January 2008 (3)
- December 2007 (3)
- November 2007 (5)
- October 2007 (4)
- September 2007 (5)
- August 2007 (4)
- July 2007 (6)
- June 2007 (3)
My Blogger Friends
I’m not going to presume to give the entire or perfect answer to the difference between reporting and analytics. But, here is a lovely start. Reporting could include creation of KPI dashboards, preparing results/reports to multiple data source integration. But analytics goes much deeper. Web analytics as a means to improve your site and company’s conversion rate and improve profitability and the use of web analytics to create action : ie, real strong sense of forward movement (ie, not just about ‘reporting’). Analytics services may include multi-variate testing, customer experience analysis, conversion analysis but the key is the actionable insights that emerge from these analyses.
The problem is that alot of the time, most people/companies/even some analytics consultancies (!) don’t differentiate clearly enough between reporting services and analytics services and in some cases don’t even see the difference between the two or assume that a KPI fashboard will give the insights as to why the site’s performance/conversion rate etc has changed. With web analytics we really need to get our hands dirty to find out why things are happening, come up with insights that are all about making things happen, taking action, testing and making things work better (increased conversion rate/other website holy grails such as increased engagement).
However here is an absolutely lovely analogy (courtesy of James Dutton): “So every day I get in my car and I drive to work; I have defined KPI”s for fuel, engine speed, speed (a dynamic metric), I have measures of success (for example averaging a certain mpg) and warning triggers (eg battery charge). In other words I have a dashboard, or in an analytics metaphor I have my site dashboard.
Now in the event that something goes wrong I may get an alert to tell me what has happened (eg “ABS failure, see technician”) but most of the time something will happen without warning – for example: overheating. Overheating is an awful problem; it happens very quickly, is disabling and has no direct signs of impending problems. Just like our site dashboard – if our conversion rate falls from 8.4% to 2.1% over the course of a month we may not realise until the next dashboard is due. The fix requires the diagnosis to proceed to rule out causator events, such as blocked radiator pipes.
Just as with web analytics the diagnosis process needs to be figured so that elements are ruled out as being a causator. The process may be simple, or may be a complex study, however it is still a process. Hence, our reporting services and our analysis services should be complementary, but without careful alignment in both process and definition will not be.”
I do love the idea of being website mechanic (not very glamourous but it’s actually quite exciting – you sort out the problem (overheating) and then you add on some bigger wheels, turbo-charge the engine and voila increase your car’s speed (website conversion rate). Oh wait, I am getting carried away.
If you agree or disagree or have any ideas then please do share.
I think facebook is the most influential social media network. There are apis and widgets springing up all over the place on facebook at the moment (since they opened up their api)and it’s interface is extremely addictive. And they are adding extra feature all the time. The degree of customisation is great as well – being able to define on a group and individual level what people can and can’t see on one’s own page or be notified about. For example, I hear 6000 siemens employees have joined facebook recently so could be replacing the corporate intranet too.
Apparently facebook is still 99% under 35 but this may change – any thoughts? Facebook is such a good place to paddle one’s feet in the social media stream so to speak because it is the most popular (bar myspace) and fastest growing (definitely) social network around. Myspace doesn’t have the same kind of appeal. Interestingly, an ethnographer just released a report comparing the average demographics of myspace to facebook visitors. Facebook started off as a harvard thing, then college only thing and only recently has gone global but with its clean clear interface in the majority appears to attract wealthier (in the states anyway) visitors than myspace. Haven’t properly looked into this though and I also don’t see where they are getting the data from (sounds like an intepration fudge to me).
Business Week review of the facebook versus myspace class divisions August 7, 2007 edition http://www.msnbc.msn.com/id/20010695/site/newsweek/page/0/ Danah’s original paper: http://www.danah.org/papers/essays/ClassDivisions.html
Interestingly, as companies have begun to realise the marketing potential of social networks, a horde of companies providing white label social networks have sprung up.
From techcrunch “The news may overflow with stories about the social networking giants, such as Facebook and MySpace, but a horde of companies are doing their best to reduce the fundamental features of these websites to mere commodities. These up-and-coming companies provide so-called “white label” social networking platforms that enable their customers to build their own social networks (often from scratch) and to tailor those networks to a range of purposes.
The idea of white labeling a network is to make the platform provider as invisible as possible to the social network’s users and to brand the network with the builder’s identity or intent. While definitions of “social networking” may vary, social networks are primarily defined by member profiles and some sort of user generated content.
There are roughly three types of companies that have emerged in the space of white label social networking. The first provides hosted, do-it-yourself solutions with which customers can largely point and click their way to a brand new social network. Companies of this type interact minimally with their customers and rather focus on providing the network-building tools that they demand.” http://www.techcrunch.com/2007/07/24/9-ways-to-build-your-own-social-network/
Increasingly for the young switched on demographic the line between personal and business life is blurred and linkedin appears to be less used than facebook.
In the path to web analytical enlightenment (heaven), there is no one answer or even one path - a bit of a misnomer really. However, since we have to start somewhere here is a 5 step web analytics plan that I have developed after working with a number of clients that really helps me to get the perspective I need to really improve their marketing effectiveness and as a result conversion rate. First, look at your actual website (it’s design and layout etc) and then get started.
Here is my 5 step web analytics plan
1. Map out your websites’ goals
2. Assess clickstream data
3. Take ownership of your site’s goals
4. Ask your customers – surveys/testing
5. Benchmark against your competitors
1. What are the site’s goals?
Map out your websites’ goals and what is currently available to make these a reality – it is so important for key stakeholders to be involved at this stage to make sure you really nail your KPIs. At the end of the day, everything else that you find out about your site and your ultimate conversion rate are all completely dependent on making sure you know exactly what your goals are and what is currently available online to make them a reality. If you cannot decide the one important reason why your site exists, you have a big problem – it’s that classic “build it and they will come” which unfortunately does not work. Once your goals are firmly in place, brainstorm and drill-down into inordinate amounts of detail as whether the website supports these goals as successfully as it possibly could (there is no such thing as perfection but whatever you think is as perfect it could get). Already, a long list of action points (actionable insights) are jumping out where the site is not supporting it’s goals as completely as it could.
2. Assess clickstream data
Assess clickstream data with site overlay tools, which can also show which clicks are for example generating most goal conversions and also with heatmaps which give a wonderful visual perspective of what is actually happening on your site. Google Analytics or Clicktracks site overlay tools both can show total clicks and clicks generating conversions. Crazy Egg has a great heat map and it is so easy to set up. However, clickstream data is often missing the context, in that absence we overlay our own opinions / perspectives to make sense of it all – which is why it is only one part of the analytical process. For example, looking at a conversion funnel on your website will not always be as meaningful as we think it should – in most cases your visitors will have so many paths to a your site’s goal that a linear funnel is not helpful. Instead if we have more of a flow chart with the most popular pages in each step of the process grouped together a much more evocative illustration is depicted. As another example, it may be that over 20% of the clicks on your homepage are on non-clickable items which is telling you that your content is not as intuitively displayed as it could be. In any case, this new generation of heatmapping and overlay is a very visual way of understanding what people are actually doing.
3. Taking ownership of goals for the site
Make sure that the relevant/right person at your company takes ownership of their goal for the site, such as Increase site engagement rate by posting relevant articles and blog posts and highlights them on the home page. If each goal owner has a KPI (key performance indicator) card they are answerable and responsible for improving on that particular goal – it has become part of their job description. This is a very simple idea that has been around for ages, I saw Nokia doing KPI cards recently but this really goes back to Kaplan’s balanced scorecard.
4. Ask your customers and listen- testing different page versions, online surveys, old fashioned questionnaires
Add a short “free text” 3 question survey to the site:
Definitely an Avinash ism but also a fundamental/traditional part of marketing as well. Ask questions such as: Why are you here today? Were you able to do what you wanted? How can we improve the site to make it easier for you to do what you want to do? One client consistently has 50% response rate to their survey on the confirmation page. And this is the quickest, cheapest and one of the best ways to find out what your customers/site visitors want. Let the customer/visitor talk (no drop menu choices), give them a chance to tell you in their own voice the reasons and let them provide you with suggestions. It works better than guessing what the answers might be and suggesting those. Categorise the responses into common themes and then rate the % of times each theme is occurring for those who are not able to complete their task. This is a simple and direct to-do list of issues directly from the horse’s mouth about what you should work on in order to improve your website experience for your customers and as a result increase your site’s effectiveness (marketing effectiveness dare I say it).
Test different versions of the home page/key conversion page:
Change specific parts of the page to assess the impact on defined goals, and assess which version is most effective – change the images, change the tone and messaging of copy from salesy to informative, change the layout – for example place only your two most popular products/services on a page versus in addition having all your significant products or services but without too much detail. The key is to test 2 page variants that are significantly different from the original with underlying marketing premise/understanding. Version 1 generates more short-term sales of product X, version 2 generates visitors with a better site experience, version 3 generates less sales of product X than version 1 but more sales overall via a better displayed overall product/service range.
Market questionnaires – old-fashioned but they work
These can be a goldmine of valuable customer preferences to one’s marketing. However, they can take longer to organise, are usually more expensive if one involves an agency and due to their qualititative nature take longer (than online surveys) to generate the same quantity of responses.
5. Expansive benchmarking against your competitors
Try comparing yourself/site against other successful online marketing resources. What can you learn from them and apply? How can we be better than them? Define where the benchmarks are (eg how often do other sites update their homepages with relevant content)? Who are the competitors we see ourselves competing against. N.B Remember that significant time and resources are required to achieve benchmarking successfully – this is not just an add-on.Benchmarking is not always easily quantifiable – but insights will begin to jump out at you once carried out. For example, if you work for a magazine – what is the customer subscription process delivering, how often is content updated in key areas and what is it’s quality, what level of services are offered to visitors versus registered users versus subscribers, how is the magazine presenting it’s brand/itself etc. Some of this is subjective and qualititative and best carried out if you have experience in usability for example. Others such as site popularity, seo presence etc can be easily found out either through Alexa the poor man’s competitive intelligence tool or paid-for services such as Comscore and Hitwise.
A list of areas (not definitive) to benchmark your company against your main competitors:
Usability: Navigation and tools, Content categorisation and quality.
Audience needs: Currency and relevancy to audience, Depth of services for subscribers.
Self representation: Brand communication, Innovation.
Content areas: Visitor interaction, Frequency of updated content.
Traffic: Incoming links, Site traffic.
It can be a bit daunting when faced with this 5 step plan to contemplate carrying out the whole thing within a short time frame (2 months), but it is achievable and the results can be unimaginably insightful and really can and will transform your site’s performance and marketing effectiveness overall.Each step in the analytics plan will have generated conclusions, action points and actionable insights and when placed in a categorised matrix by conclusion theme and in proximity to the site goals (betweenness), a natural order in which to carry out and deliver the action points and insights delivered by this plan will develop. I will write a thorough post on delivering change and insights as a follow-up to this 5 step analytics plan post. Hope that you enjoyed reading and please do comment whether you agree or completely disagree!
I think of web analytics like a puzzle. You need to get all the bits together and understand their context and importance in order to get a big picture overview.
For example, today if we look at the key conversion rate for your website (whatever that might be). Everytime there is a fluctuation in the conversion rate, upper management will be querying the reasons why and next actions which can begin to become unhelpful and also waste a lot of the web analyst’s time. And usually the more senior they are, the more suggestive (dictatorial) they get.
The thing is that a conversion rate, like any other rate, will always naturally fluctuate over time. So what is meaningful and what isn’t? This is where we call statistics into the web analyst’s arsenal to enable us to make more informed conclusions and decisions.
For example, here is my fluctuating conversion rate for a website – without any use of statistics. There appear to be a number of worrying/exciting fluctuations. Where do I start?
This is not about needing to become a qualified statistician but knowing enough about statistics to be able use to them effectively in one’s web analytics data and have an intelligent conversation with a statistician, “statistics without tears” in other words. Don’t look up standard deviation in wikipedia or speak to a professional dataminer/statistician and begin to feel statistically challenged/mathematically innumerate, please.
My introductory guide to statistics without tears
Here is my guide to statistics without tears to enable you to make the informed decisions you need to make with your web analytics data.
In statistics, here are some things that help me to make more informed decisions:
1. the mean – or the average or central location of my data
2. the trend – when we look at data over time, we apply statistics to give us a trend
3. the standard deviation – describes the spread on either side of the mean
4. upper and lower control limits – setting limits on my data to a specified standard deviation to see if anything stands out or is statistically significant
5. statistical significance – a result is called significant in statistics if it is unlikely that it happened by chance.
6. probability modelling – to determine how long until something happens, how many goal conversions will we see over a given period of time and given an opportunity to do something, how many people will choose to do it.
Conversion rate analysis using control limits, standard deviation and a trend
Here is my analysis of my conversion rate with the use of standard deviation, upper and lower control limits and a trend line. The brilliant thing about this is that it took me less than 5 minutes to do in excel.
Instantly we can see numbers that are standing out on either the top end or bottom end of the control limits, which are set to a specified standard deviation.
To do this in excel have two columns, one with the daily conversion rate and the other with each days date. Then insert a line chart which shows the conversion rate over time. Right click on the line in the chart and choose format data series. Choose “error bar”, “display both” and “standard deviation” and specify your standard deviation, for example I specified 1.7 and set values for those data points that stand out. Then add a trend line to your data, again by using excel. This error bar gives us an upper and lower control limit based on a standard deviation from the mean that helps us to see the data that stands out and has a statistically higher chance of being significant/meaningful. The trend line (which I set to be linear) gives a good idea of where our data is going, which we can’t get merely by looking at the data itself.
This is the equation that excel that uses to work out the standard deviation (which is why it is great when excel can do the job for us):
s = series number
i = point number in series s
m = number of series for point y in chart
n = number of points in each series
yis = data value of series s and the ith point
ny = total number of data values in all series
M = arithmetic mean
The result is we can instantly see which data points are standing out from the average conversion rate and correlate these backwards to company activity. For example, was there additional marketing activity such as SEO or direct marketing. Or were changes to the actual website design or copy primarily responsible? Were there periods where we would have expected the conversion rate to stand out, but it underperformed? What could the underlying reasons be? What action can we take as a result of this analysis – for example to examine all marketing activity and changes to the site during a period of a higher than normal conversion rate to see what was contributing to this success and to apply this to the future. And similarly, to see if any company activity can explain the lower than average conversion rate and highlight these as problem areas.
In my next post on statistics, I will be using probability to make projections about company performance that can really help with one’s marketing.
As usual, I very much welcome your feedback, agreement or complete disagreement and most importantly THANKS FOR READING!!
Social networks online, such as Facebook and Myspace are becoming more and more important. Increasingly, marketing through these online social networks will become ever more prevalent. For example, a friend of mine has launched his “socialight” application onto Facebook’s open API (which means external programmers can add programs and applications to facebook – not just facebook employees) – where you subscribe to an online and mobile phone GPS “stickies” service. Another example might be someone with a lot of social capital (ie contacts and influence online), in turn influencing their friends, contacts and others exponentially in theory due to the fluid nature of a social network to buy a product or service. Related to this but slightly different, Blogvertise is a service which pays bloggers to promote a product or service related to the blogger’s field of expertise. You get the gist. This is all seriously fascinating marketing.
The 2 things that absolutely fascinate me are:
1. The major challenge facing marketing strategists in how to increase the effectiveness of social network based marketing strategies.
2. The future marrying of web analytics and social network analysis and resulting improved marketing effectiveness and business intelligence.
Measuring the influence of myspace visitors
MIT Media Lab / Social Media Lab designed a flexible tool for the content driven exploration and visualization of a social network. Building upon a traditional force-directed network layout consisting of nodes (profiles) and ties (friend-links), the system shows the activity and the information exchange (postings in the comment box) between nodes, taking the sequence and age of the messages into account.
In the myspace service network-only visualization methods are no longer sufficient to meaningfully represent the community structure. Numerous commercial profiles, fake/spam/celebrity profiles and tools such as automated friend adders result in a huge numbers of connections, many of which carry little information about a person’s actual social ties and behavior. The average myspace user has more than 130 friends, but there are also profiles with over a million “friends”. By going beyond the “skeleton” of network connectivity and looking at the flow of information between the individual actors, the authors hope to create a far more accurate portrait of online social life.
Visualisation of the true influence of comment flow of myspace visitors:
What is social networking analysis:
Social network analysis has emerged as a key technique over the last century in modern sociology, anthropology (good thing I am a qualified anthropologist), sociolinguistics, geography, social psychology, information science to measure what individuals do and the many types of relationships between one another. And now social network analysis becomes important in web analytics. Social network analysis sees social relationships in terms of nodes and ties. Nodes are the individuals (visitors) within the networks, and ties are the many types of relationships between the individuals.
Two Social networking analysis metrics (an introduction merely):
Read Judah’s thought provoking post for more social networking analysis metrics and opinions.
Degree an individual lies between other individuals in the network i.e it’s the number of people who a person is connected to indirectly through their direct links.
The degree an individual is near all other individuals in a network (directly or indirectly). It reflects the ability to access information through the “grapevine” of network members. Thus, closeness is the inverse of the sum of the shortest distances between each individual and every other person in the network.
Marketing in social networks
Jason Ethier has written a good paper on social networks. He tells us that the main questions for researchers in social network theory are which types of social networks can be used as a basis for marketing strategy, how to identify and measure social networks, how to mobilise and manage social networks, and which marketing decisions can benefit the most from social network concepts and methods? Researchers have found that consumer networks that are not under the control of a corporation work best for marketing purposes which is why networks such as facebook and myspace are so successful. Corporations identify and measure social networks by collecting information from their customers. One method of doing this is by distributing loyalty/discount cards (large retailers do this) in exchange for customer information.
Social network theory and analysis and their marrying with web analytics is certain to become ever more prevalent as more companies learn of the marketing potential of social networks. And hopefully will become more mainstreamed into web analytics as a whole and into the web analytics solutions themselves. (hopefully anyway)
I welcome your feedback, thoughts or complete disagreement – so please share your thoughts and most importantly, THANKS FOR READING!
This is more of an addendum to my other post on social networking analysis meeting web analytics and marketing.
Social Network Analysis (SNA) is a good foundation, but another way of seeing the full picture of the true influence of social network is to look at value networks as these include traditional financial values within the analysis as well the importance of of comment flow for example. Value networks and VNA offer complete and thorough visualizations and understanding of a company’s performance.
“Value network analysis is a business modeling methodology for understanding internal and external value. The methods include visualising business activities and sets of relationships from a dynamic whole systems perspective and several unique analysis approaches for understanding value conversion of financial and non-financial assets, such as intellectual capital, into other forms of value. The value conversion question is critical in both social exchange theory that considers the cost/benefit returns of informal exchanges and more classical views of exchange value where there is concern with conversion of value into financial value or price.”
Courtesy of JHeuristic, see: http://kmblogs.com/public/item/166268
Open methods, tools see: http://www.value-networks.com/
The truth is that none of us know what will happen in this space of measuring the growing and true impact of social networking and media on businesses (although MIT’s work in this area on the true influence of comment is very interesting)…
For a more indepth explanation read my full post on social networking analysis meets web analytics.
Thanks for reading and as always and if you agree or disagree, please do comment.
I’ve just got back from 10 days holiday in a farm 5 hours drive from London, with plenty of hens, goats, horses, ducks, rabbits, cows and sheep (toddler paradise really – well my toddler thought so anyway) but no internet access. So please excuse the tardiness of, this, my third post!
For today’s post, I’ll focus on understanding better how internal search works and how important it is that it works effectively so that you don’t end up losing business and visitors and sending them on a galaxy quest. We don’t want our visitors to be wondering how to get to what they want or if they are there yet. The Eisenberg brothers tll us that 50% oif nternal searches result in a failed search which is a considerable number of dissatified visitors and customers to your site. So, this really is of crucial importance if you want to improve site experience and ultimately the effectiveness of your site and turning visitors into customers.
Internal search refers to the keywords that people use while exploring your site (not the keywords they use on the search engines such as Google).
For many websites, in particular holidays, recruitment, publishing and large retailers, internal search can be the most important and used feature on the website and can account for 50% of all pageviews on the site. Obviously, that isn’t the case for all sites as small sites don’t need or have internal search in most cases (so we’ll ignore these today).
In order to be able to analyse internal search keywords at all, you’ll need a paid-for analytics solutions – though not necessarily an expensive one, as both Clicktracks and Indextools can combine internal search parameters with visitor segments. Google Analytics does not currently provide any data at all on internal search keywords (don’t be confused with the “keywords” heading under traffic sources, these are the words visitors used on search engines to arrive at your site).
How do we assess how well (or not) internal search is working?
1. What do people search for and do any keywords stand out?
2. What searches results in failed searches and what proportion are failed searches?
3. Let’s do some segmenting
1. What do people search for?
Rather than getting too bogged down with the exact words and words and common misspellings that visitors use (although interesting and at times surprising), it is better to start with much broader strokes and then drill down and do segmenting later. By which I mean, how many visitors as a percentage of all visitors to your site use internal search and which are the most and least popular keywords over a representative time period, 2 months for example.
If for example, less than 1% of visitors use internal search and this is large recruitment site, then make sure that search button is placed in an intuitive and easy to find place for visitors. If the internal search area was moved, would it make a difference to the number of visitors who search? On the other side of the scale, if for example too many visitors (over 10% – it really does depend on the site) are searching on the site, then your site’s products and/or services are not easy enough to find on the site.
If the most popular keyword search is a noticable percentage of all searches, then this clearly signals that this should be clearly displayed on the site or on a site navigation bar. If a popular keyword is not even an item or service offered by your site, this is a clear signal that this should be something you either should be offering or linking up with someone who does (affiliate marketing perhaps?). Again, this seems simplistic but it is so easy not to notice how important popular keyword searches can be! When internal search is working effectively, we should not expect to see any one search keyword as a noticable proportion of the total – expect to see more of a long, long tail (lots and lots and lots of different slightly obscure keyword searches).
Here is how to get your long tail of internal search keywords:
First, create a report of all the internal search keywords and unique visits for a 2 month period and upload to excel. Grab all the keywords and visits into smoothed line chart with data points (so that you can easily see the keywords that stand out). See my chart below:
Then try making a list of the keywords that stand out. These are words that need to be looked at carefully as they will be benefit from being presented on the site in an easy to find way so that your visitors do not always need to search for them.
2. Which searches result in failed searches
A failed search is when a visitor doesn’t find what they are looking for. For example, keyword searches on products or services that you do not offer would be a failed search as would a time sensitive product or service that is not available within the results of the keyword search made. For example, visitors that click on the back button after making a search would be classified as failed/frustrated searchers. To reduce failed searches make sure the site reflects at the minimum easy to find information on the more significant failed search keywords (this is just a quick fix and not the solution if only information is presented but is a needed first step until a good solution is found).
Then look at the percentage of all visitors that have a failed search. In addition, you can create a visitor segment where the search results page is also the exit page and compare this against all failed searches to see how many “failed search” visitors, leave the site immediately.
3. Then we segment, to confirm our suspicions and insights
Assuming we are using an analytics solution where we can label visitors segments with specific keyword searches, we begin to drill-down further. If on a recruitment agency site, a noticable search is for “web analyst”, we can see which were the most popular pages they visited before searching. From this, we could learn that they visited the “marketing jobs” page and the “Web jobs” page and as a result of not being to find what they were looking for, searched for the term “web analyst” and subsequently left the site. Therefore, it would appear that both of these jobs page would benefit from having information about web analyst jobs on them (until the recruitment site started posting web analyst jobs that is).
We can segment against new versus returning visitors, time spent on site, by navigation path etc. For example, we can see how these visitors came to the site in the first place by looking at search engine keywords. If there is a noticable percentage of visitors who arrived at the site after having searched for “analyst job” or “web analyst job” on search engines, then it is clear that the hopes and desires of visitors coming in from the search engines is not being met by the site – as well as a PPC (pay per click) overspend on keywords that are resulting in a high number of failed searches and exits from the site.
Some final thoughts – this really is all about marketing, effective marketing (Kotler anyone?).
The key is to reflect on the site what visitors are looking for, in a holistic and thorough way. Not just add-on a quick note on one of the pages that this site does not currently offer web analyst jobs (although in the short-term, a quick fix is better than nothing), but have a think about the big picture, what the site is trying to achieve and use this information about what visitors want and aren’t getting to improve and add to the service offering – in this case, widen the remit of the recruitment agency itself and add web analyst jobs to the site.
I really do (I promise) welcome your agreement, disagreement and opinions, so please do share some of your thoughts by commenting on this post.
Marketing performance management (MPM) has begun to be bandied around quite a lot as an acronym for web analytics and it (finally) has begun to be taken seriously by numerous marketing departments. Marketing performance management sounds very grand and is how we would like to imagine it should be “wow, now I can analyse and act on my marketing data to improve my site/sales/customer engagement etc.” – a holy grail for switched on marketers. But web analytics with the best analysis in the world isn’t a “get out of jail free” card. Here are three things to look out for from either an analyst or company perspective.
We’re not there yet…
Scenario 1 – our data sucks, so let’s get over it – collaboration
For example, we are using the most high powered analytics solution on the market to measure all online visitor activity. However, we’re not getting our true conversion rate due to offline sales generated by the web that aren’t being tracked – visitors that search for the products online and then complete the sale over the phone (this frequently occurs for complex products such as mortgages). And, although call centre staff have been trained to ask where the customer completing the sale came from and then input 3 for website (into their telephone tracking system that links up to the analytics solutions) how many actually do (input 3 and/or have these systems in place)?
Before I start, if we recall that Walley is blind and Dave is deaf (they have to collaborate) and together they stop the criminals getting away with it (stop the lies) and clear their names (with persuasion). Read on to find out why on earth I am mentioning the storyline of this film and hence the blog title “see no evil, hear no evil”.
What about visitors that delete their cookies so we think they are first-time visitors but they are actually returning? (Comscore released a report recently saying that nearly 30% of visitors delete their cookies but this may also be overstated because they have a competing system of measuring visitors they are trying to sell). There are other unquantifiable elements that I haven’t mentioned/thought of here.
Unfortunately for the analyst, it is impossible to ascertain with 100% certainty what is the perfect data/true conversion rate from the site alone. So we move on and try to come up with assessments that allow for other hard-to/non-measurable elements to be included in the “big picture” – be that through better data integration, assumptions or probability theory.
eg P(A) = The number of ways event A can occur
The total number of possible outcomes
To get the full “web analytics” picture, we need collaboration between different sides of the business and the context in which activity is occuring.
Scenario 2 – conflict of interest means good advice goes out of the window – lies
The analyst works for a respected online advertising/search agency. As a result of analysis, it is discovered that PPC/banner campaigns are generating a much lower conversion rate than referrals and direct and that PPC banner campaigns have a much higher bounce rate (visitors who spend less than 5 seconds on your site).
Three wise monkeys – see no evil, hear no evil, speak no evil
Does “we will use this search data to optimise your spend” ring any bells? Search marketing agencies are making commission from the search engine on their PPC spend and the creative spend from any banners created. As a result of which, the analyst is inevitably compromised and will tell the truth from the context of “let’s optimise” but may not mention what might be best for the client – eg your organic traffic is being destroyed by our PPC spend on brand keywords, let’s cut PPC spend as it is unnecessary. Or, PPC/banner traffic has a much higher bounce rate than other traffic sources, let’s cut our spend.
With compromised advice, the true value of analysis and of the analyst has been removed. So, in my opinion it is much better to have an unbiased analyst who is on your payroll for that reason alone.
Traffic Source Conversion rate % Bounce rate % Total visits
Direct 6.1 19 27,000
Organic 3.4 23 14,599
PPC 2.8 54 12,200
Banners 0.2 52 3,129
Referrals 8 21 7,802
Scenario 3 – insights need persuasion to make them a reality
Often great actionable insights generated from site analysis aren’t enough to get buy-in from our boss/client. Until we can convince others of our insights’ value, they are meaningless. There are plenty of people good at crunching numbers but I believe the real value is in the ability to persuasively present that data and persuasively push through the changes needed to improve one’s marketing performance. Without persuasion, one may have insights about what could or should be done but it is only when action is taken, that web analytics is truly transformed into what it would like to be “marketing performance management”.
The conclusion – how to get web analytics working better for you
In conclusion, we collaborate with different parts of the company/data that aren’t talking/integrating to one another (just like Walley who is blind and Dave who is deaf work together), ignore the tall stories from our search/marketing agency (we stop the criminals getting away with it) and realise our goal by using persuasion (in Walley and Dave’s case, clearing their names).
The end result is a better platform using web analytics from which to manage our marketing performance.
I welcome your suggestions, feedback or even complete disagreement so please let me know your thoughts. Apologies in advance if you found this blog post a little long and I do hope that someone out there actually reads the whole thing.
- Campaign effectiveness (7)
- Consumer segmentation (3)
- Customer experience (5)
- Engagement metrics (5)
- Events (7)
- Google Analytics (5)
- Ideas (12)
- Influence (3)
- Management theory (2)
- Marketing (18)
- Microsoft Gatineau (1)
- Podcasting (1)
- Site search (2)
- Social Media (7)
- Social media analytics (2)
- Social media strategy (3)
- Statistics (2)
- Virtual Worlds (1)
- Web Analytics (39)
Wordpress theme by Wordpress Themes
Web Analytics Princess by Marianina Chaplin