Last Wednesday (15th Oct) was the 10th UK Web Analytics Wednesday (WAW) run by myself and SCL Analytics, at the Cumberland Hotel in London (fantastic venue courtesy of Coremetrics).

We had another record attendance with over 250 registering for the event and 180+ turning up on the night, making it the biggest non eMetrics WAW event in the world… ever!

So thank you to everyone who made it on the night.

The October 15th 2008 London WAW

Our October event had an A/B and multivariate testing theme to the evening and saw presentations from three perspectives; Vendor, Customer and Consultant.

Our first speaker of the night was Mark Simpson, Managing Director for Maxymiser, who spoke on “Conversion Management – How to Proactively Increase Conversion Rate”. Following Mark was Ewald Hoppen, Web Analyst for Wehkamp.nl who spoke on “Multivariable testing at Wehkamp.nl, the Dutch favourite online retailer”. And the final speaker of the night was Eric T. Peterson, CEO of webanalyticsdemystified.com and founder of Web Analytics Wednesdays who presented his  “Ten Tips for Testing Successfully”.

Some pictures from the night

Below are some pictures I took on the night (I definitely won’t win any awards for photography!): -

Next London WAW - January 21st 2009

Our next WAW networking event is taking place in the new year, at the Calico Bar in London. The agenda is still to be announced, but I am considering having a presentation free event for the next one as the October WAW was quite presentation heavy. For more information and to register to attend, please visit: www.sclanalytics.com/resources/events/waw_january2009.

Thanks again to everyone who is continuing to help make the London WAWs a success and I hope to see you all at our January event!

Web Analytics Wednesdays (WAW) are a free to attend networking event, that are held globally, traditionally on a Wednesday Night.

Background on the London WAW events

We began them back in march 2007 and have been running them bi-monthly ever since. They have gone from strength to strength and at last count, we have had over 400 web analytics professionals attend a London WAW event at some point, with many returning time and time again.

Testimony to the event’s success is the fact that we have had regular attendees from places as far flung as Sheffield, Edinburgh, Europe and even the USA, with Web Analytics Wednesdays very own Eric T. Peterson kindly agreeing to be our guest presenter at our March 2008 WAW event, presenting on ‘the future of Web Analytics’. Eric helped us pull in our biggest crowd to date, with over 200 people registering to attend and approximately 110 people turning up on the night!

The July 16th 2008 London WAW

Wednesday night’s July WAW was another successful event, which saw around 70 people turning up on the night. Thank you to Amanda Kahlow from Web Analytic Consulting LLC for her insightful presentation on ‘Effecting Positive Business Change Through Web Analytics, Case Studies from Enterprise Frontlines’.

They say pictures speak a thousand words… Unfortunately I am not the best photographer, however I did take a few photos from the night, which I have included below: -

What’s next (October 15th 2008 WAW)

Eric T. Peterson has kindly agreed to come back again and present at our October 2008 London WAW event, being held on October 15th at the Cumberland Hotel near Marble Arch (thanks to Coremetrics for providing us with a venue). Mark Simpson from Maxymiser has also been confirmed as a guest presenter at the October WAW.

Finally, I would just like to say a big thank you to everyone who has helped make London Web Analytics Wednesdays such a success and hopefully I will see you at the next one!

Keyword Research

I set up a new Google Adwords PPC campaign within the SearchForce tool and conducted some keyword research with the aid of the keyword generation tool in SearchForce (powered by Keyword Discovery) and also used some other free tools, including a nice free keyword research tool from Wordtracker.

Its important to try and use more than one keyword research tool when conducting keyword research, allowing you to use a triangulation approach to your research. Also remember that these keyword tools are there to provide suggestions, but it is always worth including some of your own knowledge of your products or service when building your keyword lists.

Once I identified some keywords to bid on, I set the keywords predominately as broad match keyword terms and decided to let the new campaign run for a few days to collect some data and allow me to further refine my keyword list.

Match Types: Broad Match Keywords

Several match type options exist within the search engines, including broad match, phrase match and exact match. There is also an embedded match option within Google Adwords and an option to build in a negative match list of keywords, which this post is focusing on (Google Learning Center has a useful article on the different keyword match types).

Broad match is the default setting with in Google Adwords and Yahoo! for your keywords (or keyword phrases as they may be more accurately defined). Using broad match mean that your ads will show whenever that keyword is typed, as well as when other similar terms are used in the user’s search query.

For example: If your keyword was Used Cars and was set as a broad match keyword, then your ad could also appear for the following: -

Buy Used Cars

Used Cars Dealer

Used Cars Locator

Used and New Cars

Mechanic for Used Cars

I’m never ever going to buy a Used Car

This means that your ads will reach a wider audience, but bear in mind that in doing so, you may be reaching a partly off target aaudience. Using the example above, if you sell used cars but do not offer any mechanic services, then you could be paying money for irrelevant traffic, such as users who searched for Mechanic for Used Cars, or even when someone searches I’m never ever going to buy a Used Car.

This is where some further investigation needs to be conducted within your Web Analytics data, to help you build a relevant negative keyword list.

Investigating the Actual/Referrer Keyword

Looking into the web data further, it is well worth investigating the actual or referrer keyword phrases used, which is often included in the Search Referrer (see the WAA standard Web Analytics Definitions document for more information) and is recorded within the log files and can be accessed through your Web Analytics package. Investigating the actual/referrer keyword phrase used is a great way of identifying off target or irrelevant keywords, which your ads are appearing for, which are often the result of using broad matching.

In the fictional example below (see screen shot), a TV retailer broad match bidding on the search term plasma tv is appearing for the search term history of plasma tvs. Now chances are, if they are purely an e-commerce site looking to make money from selling tvs online, then they are likely to want to make the word history a negative keyword.

Sample Paid Keyword Vs Actual Keyword Report (click to enlarge)

This will mean that in the future, anyone searching for the history of plasma tvs, will not be shown the tv retailer’s paid search ad. The tv retailer is likely to want to only pay for visitors who are in the market for buying a tv, and are not likely to want to pay for someone searching for the history of plasma tvs.

Free is often another popular negative keyword, if you are offering a paid for product or service, as free often is a popular addition to many search engine queries.

Building my Negative Keywords List

I am in the early stages of optimising a Google Adwords campaign I have set up for SearchForce, our Pay Per Click (ppc) bid management tool.

I have several broad match keyword phrases that include the term ppc as a keyword and this has meant that on several occasions, our ads have appeared for the keyword pda. This was identified by investigating the actual/referrer keyword report within the Web Analytics data and therefore made a good addition to my negative keyword list, which I am in the early stages of compiling (see screen shot below).

Sample Negative Keywords List (click to enlarge)

Appearing for pda when ppc is the search term that you are bidding on, also highlights the Search Engines’ slightly liberal approach to broad matching a user’s search terms with the paid for keyword phrases that are being bid on. It also demonstrates the importance of investigating the actual/referrer keywords being used.

Identifying Exact Match Keyword Phrases

To avoid paying for terms that are not relevant to your ads, keep an eye on your referrer/actual keyword data to help you identify your negative match keywords, as well as helping you to identify good candidates for exact match keyword phrases.

A recent Greenlight Search PPC webinar that I viewed, discussed the benefits of identifying exact match keyword terms. If you can identify search terms to make exact match terms then you can, according to Greenlight, save up to 80% on the cost of that phrase when compared with the cost of that phrase as a broad match term. This is because you end up paying extra for any additional keywords added to the broad match keyword phrase.

Taking the example in the earlier screen shot, the plasma tv products search term could be a good candidate for an exact match phrase. According to Greenlight’s webinar, It can be much cheaper if you make this an exact match term.

Not sure how true this is, as I have read other articles that say you pay more for exact match terms. But in any case, it can’t hurt to build up a list of exact match keyword terms as it helps you to narrow down your audience, potentially saving you money in the long term, by ensuring your ads are more precisely targeted.  Also, providing you are properly tracking and attributing conversions, exact matching allows you to customise your bidding, letting you bid higher for those exact match phrases that convert at a higher rate.

Clearly though if you have a keyword list longer than “War and Peace”, then you would only want to do this on your higher volume or higher converting keyword phrases.

Do not rely solely on keyword conversions

However, bear in mind the conversion rate of each keyword should not be your sole factor for making a decision on which keywords to cull and which to keep. This is because some keywords can contribute to a conversion event indirectly.

e.g. A tv retailer broad match bidding on sony tvs has their ad appear and clicked when a person uses the search term information on sony tvs on Monday, then their ad appears again on Friday when the same person searches for buy sony tvs, clicks the ad and makes a purchase.

In this example, the conversion event is likely to be attributed to the last search term by default, but with some tweaking in your Web Analytics package, you could attribute some value/weighting to each of the search terms used by a visitor in a given time period (defined attribution window). This way, you are less likely to cull keywords that are actually have an impact on the final conversion event (the actual purchase!).

In Summary

It is important to drill into the actual or referrer keyword data to help you identify candidate keywords for your negative match keyword list. You should try and make it a habit to regularly investigate the referrer/actual keyword reports for your paid search campaigns, helping to keep your negative keyword lists up-to-date.

Also, try and gain a good understanding of the different match type options available to you and then use them to your advantage, helping you to more precisely target your paid search ads to the appropriate audiences.

Final thought of the day… Investigating your referrer/actual keyword phrases will give you some very useful insights.

What are your views? Have I missed anything? Do you agree or disagree with me? Please let me know by adding a comment…

]]> http://www.marketing-demystified.com/2008/07/web-analytics/ppc-campaigns-building-a-negative-keywords-list/feed http://www.marketing-demystified.com/2008/07/web-analytics/ppc-campaigns-building-a-negative-keywords-list http://feeds.feedburner.com/~r/MarketingDemystifiedBlog/~3/310330964/hits-a-stupid-metric-for-measuring-success http://www.marketing-demystified.com/2008/06/web-analytics/hits-a-stupid-metric-for-measuring-success#comments Thu, 12 Jun 2008 11:16:16 +0000 Chris Clapham http://www.marketing-demystified.com/?p=15 Since the dawn of the web and the birth of web analytics, some webmasters have reported back to their respective businesses, the number ‘hits’ their websites were getting, as a measurement of success.

I still see some businesses reporting ‘hits’ as a measure of success today, especially SMEs, where they have had little exposure to Web Analytics or little education about it. This was highlighted to me very recently and inspiration for this blog post came from a headline that caught my eye in a local business magazine, that found its way to my desk a few weeks back.

The headline read: -

200,000-plus hits on our website!

Every time I see ‘hits’ reported as a measure of success, it makes me both smile and cringe at the same time. I think over time, some people have misinterpreted this metric ‘hits’ to mean the number of visitors a website has had over a particular period of time. This is a very inaccurate interpretation and can give a false sense that a website is performing well, when someone is shown a graph plotting ‘hits’, which over time goes up to the right.

So why do I think ‘hits’ is a stupid metric for measuring success?

A ‘hit’ is simply a request to a web server for web page content. Every element that makes up that web page is served by the web server and as a result an individual ‘hit’ is recorded against each element requested. This includes the web page itself, any images, style sheets, JavaScript code, bits of Flash and anything else that forms part of that web page.

Therefore, every time a web page is requested by a browser, a series of requests are recorded in the server log files, which are commonly referred to as ‘hits’ (illustrated below). As a result, a single visit to a website could contain multiple page views and each page view has multiple ‘hits’.

Example of some web server log files (click to enlarge)

This means that if you make any changes to your website, adding an additional style sheet for example, then measuring ‘hits’ will make it impossible to do any comparisons of old vs. new. As using an additional style sheet in your new website design means each web page requested also makes an extra request for the additional style sheet and therefore records an additional ‘hit’, making any meaningful comparisons impossible.

It also means that if you are tasked with increasing ‘hits’ to a website you simply need to add more images to the web page! Anyone who needs some help with that, let me know as I charge a very reasonable rate!

In all seriousness though, these are the exact reasons why ‘hits’ is a dumb and pointless metric for measuring success and in this context, should be avoided like the plague!

A move from page views to page events

Visitors and page views tend to be the standard metrics used for measuring a website’s growth performance over time. However, with the increasing use of Rich Internet Applications (RIA) based on Flash or AJAX for example, measuring traditional page views has effectively gone out the window.

If you have ever visited a Flash based website, then you may have noticed that the URL stays static as you browse the website. Requests may still being made to the web server for content, but no new URLs are loaded. As a result, page views may not automatically appear within the log files.

You therefore need to think about tracking visitor interactions instead, often referred to as event based tracking. An interaction could be anything that you define, for example it could be a request for a ‘page’ or content, or you could take this a step further and track visitor interactions with a form. Which fields do they fill in? Which field do they take longer to fill in? which field do visitors bail on?

A great example of a Flash application is the BP Carbon Calculator. By tracking interactions, BP could see how visitors are using the application. Providing BP with information, for example, on; Which check boxes are being toggled? Which options are being selected from a drop down? Or how many people click the next button?

Tracking web page interactions can provide you with some really useful insights into how visitors are using your website, helping you streamline your website and is certainly worth further investigation.

Measuring online versus offline

Taking this notion of event based tracking further, how does online vs. offline differ?

Well the great thing about online, when compared with offline is that you can track virtually anything and everything!

Take the example of an organisation sending out a seasonal product brochure in the post… They might be able to track the number of calls they receive from the brochure or attribute sales to that brochure with the aid of promotional codes, but what they can in terms of interactions with that brochure is quite limited.

Where as online, with the help of web analytics and event based tracking, a business could track which individual web pages were viewed or individual interactions within a web page. e.g. which field within a web form caused them to bail? Or even how that visitor interacted with a flash video - did they watch it all the way through to the end?

With event based tracking, the possibilities for tracking and measuring are almost endless.

Conversion Rate - a better measurement of success

There are almost a limitless number of metrics, which a business can use to measure how successful a website is over a period of time. Unique visitors will give you a good indication as to how your website is growing over time, but anyone can drive more visitors to a website.

The key is obviously driving more relevant visitors, who ultimately do something that you want them to do… They convert!

Woohoo!!

A conversion could be a sign up for a newsletter, a purchase online or perhaps any visitor who looks at your ‘contact us’ page. The point is that a conversion (referred to as “Goals” in Google Analytics) can be anything you define as a useful and desired action on your website and is completely customisable to each business.

Therefore, number of conversions over time is a useful metric to measure your website’s success. Better yet, looking at your conversions or conversion rate by channel or marketing campaign. This way you can then see which campaigns are converting more visitors, allowing you to make decisions on which campaigns are working best for you and which ones need more of your attention.

You could take this a step further still, and see how each element of your campaign is converting, providing their are multiple variables. For example, take a Google PPC search campaign… you could investigate the conversion rates for each of your Google paid keywords and decide which ones are worth bidding on (illustrated below).

Sample Google paid keywords report (click to enlarge)

Another example might be measuring the effectiveness of banner creative or landing pages. Its always a good idea to test different creative and landing pages to see which type converts at a higher rate, because over time, one creative or landing page will always out perform another. That way, you can then divert campaign traffic to the highest converting landing page, or show visitors the most effective banner creative.

Summary

In summary, the metrics available to you for measuring success are virtually endless and should be unique to your individual business. However, number of unique visitors or number of conversions are usually a safe starting point for measuring success and they make much more sense than measuring the number of ‘hits’!

Final thought of the day… ‘Hits’ is a pointless metric for measuring success, so please avoid it!

What to learn more about Web Analytics metrics? - The Web Analytics Association have published a nice document detailing some standard Web Analytics Definitions for metrics… well worth a read!

What are your views? Have I missed anything? Do you agree or disagree with me? Do you know anyone in your business reporting ‘Hits’? Please let me know by adding a comment…

You can’t optimise what you don’t measure

If you don’t know whether a campaign or channel is successful in comparison to another, then how do you know where to spend your finite marketing budgets to maximise Return On Investment (ROI) and meet those ever important Marketing objectives?

I know half my Marketing budget is wasted, I just don’t know which half

A common ideology held by those who are not tracking and analysing everything they do. Without some form of Web Analytics, knowing which parts of your Marketing budget is being wasted is practically impossible to accurately answer, when driving visitors to your organisation’s website.

No Excuses - Tracking Can Offer Great Insight

It’s surprising how many marketeers I meet who still don’t have a complete grasp of this concept of tracking everything they do or who cannot, with any level of certainty, answer which Marketing activities or channels are giving them the best ROI. I guess being lucky enough to work for a Web Analytics company has meant that this concept has been ingrained into me since day one.

Now clearly, tagging up campaigns is never 100% foolproof, as some individuals will remove tracking codes and head straight for the destination URL, in a stealth like fashion. However, there are still no excuses for not tagging your marketing campaigns, as it is so easy to do and often yields some fantastic data and a great insight into what’s working and what isn’t or how visitors from different campaigns or channels interact differently with your website.

For example:

• Do visitors from a Google Adwords campaign spend more time on your website than a visitor coming from a banner campaign placed on an advertisers website?
• Are Google PPC campaigns more effective at targeting a certain demographic than say Yahoo or MSN?
• Do visitors from an email campaign look at different areas of content to someone coming from an organic listing?
• Which email creative works better or does email link placement matter?
• Over time does one channel convert a higher number of visitors than another channel?
• Which Paid (PPC) keywords have the most conversions and what are their conversion rates? (illustrated below)

Sample report showing top paid keywords by conversions
(click image to enlarge)

If you effectively tag up your marketing campaigns, then the level of analysis open to you once a visitor lands on your website is enormous. But remember… as with everything data related - the more data you have, the greater the level of statistical significance and as a result the higher chance you have of making a sensible and informed decision.

Offline Channels

Now arguably, offline channels driving visitors online can be much harder to track, but not impossible. An offline channel could still be tracked by publishing a vanity URL (e.g. www.example.com/vanity) and then using an automatic 302 redirect and appending a query string to the end of the destination URL (e.g. www.example.com/destination?campaign=XYZ), allowing your Web Analytics package to attribute a visit to that particular marketing campaign (illustrated below).

Diagram illustrating a query string being appended to a destination URL, using a vanity link and a 302 redirect (click image to enlarge)

Another common offline channel you may use as part of your marketing mix might be TV advertising. Although TV advertising is likely to form part of your brand building strategy, as apposed to a direct response strategy, you should still think about tracking visitors to your website as a result of the TV commercial. There is nothing stopping you from using vanity URLs to do this or maybe even registering the .tv extension for your domain e.g. www.example.tv and promoting this solely via your TV advertising. This way, you know that anyone who arrives at www.example.tv is as a result of having exposure to your TV advertising campaigns.

Tracking offline campaigns can never capture every single visitor who came to your website off of the back of being exposed to an offline campaign, but it will give you some useful data and give you some indication as to the success of an offline campaign.

Tracking offline campaigns will also allow for some useful trending and comparisons, if exposed to a sufficient number of people.

For example:

• Did print creative A perform better than print creative B?
• Did showing a TV advert on the weekend result in a bigger spike in traffic to the .tv domain than showing the TV ad in the week?
• How well did your full page print advert perform over the last three months? (illustrated below)

Sample report showing the success of a full page print advert
(click image to enlarge)

This sort of trend data, gleamed from tracking offline campaigns will allow you to make some informed and data based decisions on your offline campaigns, providing you have a reasonable amount of data to base those decisions on. The chances are, with a national TV advertisement for example, over time you are likely to be able to use your Web Analytics data to help shape your future TV advertising campaigns.

Statistical Significance

As mentioned earlier in this post, when using Web Analytics data to help base decisions about your Marketing campaigns, ensure that your data has a degree of statistical significance. Clearly basing a decision to change or axe a Marketing campaign based on how 10,000 visitors behaved on your website, is much better than basing that decision on only 100 visitors.

You may be thinking… but my website doesn’t get 10,000 visitors a month?

A level of statistical significance is going to be unique to an individual business. Basing a decision on a few hundred visitors might be enough for a low traffic website, however if you have huge traffic volumes then you are going to want to base that decision on thousands of visitors. Work with what you have and trend data over weeks and months, this will help in your Marketing decision making.

You may also be thinking… will looking at just Web Analytics data give me enough information to make an informed decision?

It is certainly much more reliable than making a decision by sticking a finger in the wind or throwing a dart at a dart board, but obviously the more different types of data you have, the better the decision can be. Think about integrating other data sources into your Web Analytics where possible or maybe surveying your website visitors, as they will tell you when something is good or bad, enriching your quantitative data with some valuable qualitative - iPerceptions 4Q offers a great little free on exit tool to survey your website visitors.

Summary

Tracking your Marketing campaigns should be at the forefront of every marketeer’s mind when planning and creating your campaigns and should never be left as an after thought. As once the campaign has been launched and traffic is coming to your site, if you haven’t tagged that traffic then you can’t accurately attribute it to a particular campaign.

Arguably there are exceptions to this, in that if you are only running one banner creative on a particular advertiser’s website at a certain time and there are no other links to you on that site, then you could use referrer data to track this. But why risk inaccurately tracking and attributing the success of that banner, when it is so easy to just add a tracking code, allowing your Web Analytics package to uniquely identify visitors from that marketing campaign.

Also, think about levels of statistical significance when analysing your Web Analytics data. Make sure you have sufficient data for your business ,when making decisions on optimising the ROI of your Marketing spend. Clearly the more data you have, the better.

Final thought of the day… If You Do It, Track It!

What are your views? Have I missed anything? Do you agree or disagree with me? Please let me know by adding a comment…

So why did I decide to start this blog?…

I recently organised the London May 2008 Web Analytics Wednesday event (WAWs are designed to bring together anyone interested in Web Analytics for an evening of drinks and Networking) in Conjunction with the eMetrics Summit. Due to not having enough time to run a Web Analytics vendor panel during the day, they had changed the evening’s schedule and swapped out Jim Sterne’s (Founder of the Web Analytics Association) presentation to accommodate the eMetrics 2008 vendor panel discussion.

Eric T. Peterson (Founder of WAWs) speaking at the March 2008 London WAW event

I found out 30 minutes before the start that they wanted me to introduce the panel to the 150 strong audience and also wanted someone from my company (SCL Analytics) to take part on it as we were organisers and sponsors of the WAW event being held that evening. Arriving with a couple of colleagues relatively new to SCL Analytics, I offered to represent SCL Analytics on the panel. The panel consisted of some of the leading minds in Web Analytics (including representatives from Omniture, Google, OX2, iPerceptions, Coremetrics, ForeSee Results and Stratigent LLC), so to say it was a daunting prospect is a slight understatement.

If I am honest, I ended up struggling a little to actively partake in the discussion, but I did come away feeling quite positive, as everything these industry leaders said (some of who had been doing this for ten years plus) made a lot of sense and reflected a lot of my own thoughts on Web Analytics and the state of the industry today.

Emetrics Summit/WAW London May 2008 Vendor Panel Discussion (I’m on the far left)

I came away with the feeling that I have a fairly good grasp of online and Web Analytics and this was one of the primary motivators for this blog. To further develop my understanding and offer a place for me to get my ideas and thoughts down on paper - hopefully next time a vendor panel discussion comes my way, I will have more confidence in my ideas to actively participate and answer some of the audiences questions.

I’d been thinking about setting up a blog for some while now and some secondary inspirations came from a desire to understand how to set up a website from scratch. I have learnt a lot about websites and writing HTML while at SCL, but not really had that much exposure to the back-end and building a website from the ground up.

I thought that registering a domain, organising web hosting, setting up a blog, setting up tracking for it, editing CSS, using FTP packages and writing regular content would give me a greater understanding websites and also provide me with a playground to test out some online marketing techniques, such as Search Engine Optimisation (SEO), to try and drive some relevant traffic to the blog.

Inspiration also came from Avinash Kaushik who writes a very interesting and engaging blog on Web Analytics and in particular a post he made recently on the Benefits Of Blogging - A Practitioner’s Perspective.

I also thought it would be nice to hopefully makes some new friends along the way and build up some dialogue with regular readers (if I can manage to hold your attention for long enough). So hopefully you enjoy reading this blog and find parts of it interesting and useful and I will do my best to keep it up-up-date with some regular posts!