We’re attending the MeasureFest conference in London today and will be sharing the best bits from all 14 talks and 4 groups of sessions about CRO, web analytics and usability. Below you can find our insights from the talks so far and check back at 2:30 for slides from Anna Lewis who will be speaking at the event about the power of segmentation.
We’re trying something a little different for this conference and will be updating this blog post at the end of each group of sessions. That way you won’t have to wait until the end of the day to get everything and you’ll still be able to see what you’ve missed any time. Please let us know what you think in the comments.
A Framework for CRO by Paddy Moogan, Distilled (@paddymoogan)
1. To become a true 1% expert in anything it takes 10,000 hours, which is 5 years full time. To get 95% of what you need to know, you only need 20 hours.
2. Discovery – You need to fully understand what it is that you need to test by understanding the business and ensure you are gathering the right data to review the various aspects. E.g. Business goals, competitor USPs, sales process including offline, customer concerns and needs, customer demographics, website usability.
3. Experiments – Use the data collected in the discovery stage as a means for designing the experiment. Make sure you have a clear hypothesis covering what you are testing and what you expect to happen.
4. Ensure you keep some traffic going to the existing page as a control version. Understand the risks.
5. Review – Be patient and collect a sufficient amount of data. Compare the results against your hypothesis. Check for any anomalies and consider their causes.
Bonus Tool: Optimizly
Cultural Conversions and International eCommerce by Joe Doveton, GlobalMaxer (@GlobalMaxer)
1. Two Approaches: localised design or a global template. CRO should be carried out in small stages, isolate aspects for testing, don’t try and tackle all aspects at once.
2. Language and Communication – We are seeing a crossover between the language being used. Language research needs to be carried out using tools related to the search engines used within the region of the world. Localise, don’t translate.
3. Colours: Colours are perceived different in different countries, however, this is just a rule of thumb – research still should be carried out.
4. Technology: Consider what technology is accessible to each location. There is no point localising your content if that region of the world does not have the technology to view it.
5. Payment preferences: Consider what payment methods your target audience are comfortable with. E.g. Germany are averse to credit payment.
Mobile CRO by Stephen Pavlovich, Conversion Factory (@conversionfac)
1.With the sharp increase in mobile and tablet sales, mobile CRO is essential. Just the adoption of a mobile site alone can influence a HUGE increase in conversion rate – imagine the potential of a site which have leveraged mobile CRO.
2. Start small – split test one landing page at the start of a funnel by optimising it for mobile.
3. Use Qualaroo to gain feedback to understand user Intent on mobile devices. This will highlight why people are not converting and ensure your mobile site is optimised for the most likely situations. Example question: “was there anything which nearly stopped you from converting today?”
4. Use heat mapping to determine the key aspects of your desktop site. Then use this data to prioritise what data you show on your mobile site.
5. Sketch the design of your mobile site on a business card. Use a fat marker so that you aren’t tempted to write too small.
MeasureBest by Philip Sheldrake, Euler Partners (@Sheldrake)
1. Social is more than selling. Marketing is about creating value for customers not just about selling.
2. When talking to clients about social, don’t focus on metrics (followers, likes etc.) focus on outcomes. High metrics do not translate to good public relations.
3. Just because things are easy to measure or are already being measured does not necessarily make them worth measuring. Consider what that measurement counts for.
4. When presented with a set of data – ask so what? What does that mean for the business in terms of business outcomes.
5. There is a difference between ROI and value when it comes to social. Value is far more key when carrying out measurement on social.
The Analytics Renaissance by Ben Harris, Decibel Insight (@Decibelinsight)
1. It is important to consider what information in your Analytics account is going to be most useful. 49% of people update their website on a weekly basis but only 14% of people track this.
2. Traffic + UX + Content = The success funnel.
3. Learn about your visitors – Make sure you segment your traffic down and base your judgement predominately on the valuable visitors on your website (those which convert).
4. Experience as your users experience: Make sure you test various browsers, view visitor playback to determine visitor navigation patterns and carry out basis UX tests.
5. Determine which content is most popular: Heatmaps are the best tool for this, however, this can be misleading if you have a wider clickable area as clicks will be more spread. You can then arrange your content in such a way that your content is in an optimisation location.
Excel, You Love To Hate It by Russell McAthy, Stream:20 (@therustybear)
1. Spreadsheets are only useful in context.
2. To easily view a deconstructed URL use the ‘text to column’ function using ‘/’ to identify where the URL should be split.
3. Use pivot tables – they’re easy to use and make data manipulative.
4. To transpose data from vertical to horizontal, do a find and replace and then replace the = with a #. Transpose the data using paste special. Find and replace # with = and your data has been transposed.
5. In total Russell offered 32 easy to use tips. For more advanced tips and tricks including formulas view his slides here.
Attribution with Google Analytics by Dara Fitzgerrald, FreshEgg (@darafitzgerald)
1. Assisted conversions and last click conversions are not mutually exclusive. This can lead to reporting the same data twice.
2. The attribution modelling tool can be used to determine how you spread the attribution; equal credit, time decay, position based.
3. Data-driven attribution – available to premium clients only! Multi channel funnels only shows you how many people passed through a particular path and converted. It doesn’t take into account the amount of people going through that path that didn’t convert. The data-driven attribution takes this into account and provides you with a ‘likelihood to convert’.
4. 80% of conversions happen within a path of 4 steps or less according to Google.
5. Issues with attribution: doesn’t take into account offline conversions, doesn’t account for cross device paths, attribution focuses solely on acquisition stages – it doesn’t take into account the retention methods (to get around this set up a goal to identify repeat customers).
Getting Google Analytics Going from Get Go by Nikki Rae, Future Insight Analytics (@AnalyticsGirl)
1. Sometimes organic keywords can appear that you don’t rank for. This can be paid search traffic which isn’t tagged for GA from Yahoo, Bing or other search engines.
2. Sometimes direct traffic isn’t direct traffic, it can be traffic that Google doesn’t know where it has come from. E.g campaigns tracked incorrectly, servers strip the manual tracking, social media which is read through an application such as Tweetdeck.
3. Sometimes a referrer is not a referrer.
a). Your site can be a referrer to itself if your cross domain tracking is incorrect.
b). Search engines can appear as referrers when it doesn’t recognise it as a search engine, this can be sorted by a piece of code.
c). When Google appears as a referrer, it could be because it comes from another Google product as opposed to a search.
4. Use a simple search and replace filter to ensure you only have one homepage registered within your GA account.
5. Good tools are Screaming Frog, GA site search.
The Power of Segmentation in Web Analytics by Anna Lewis, Koozai (@Koozai_Anna)
1. Why segmentation? Users come to websites with different intents. This allows you to market to users in different ways dependant on how we want them to behave on our site and also to see the differences on performance.
2. 52% of digital marketers agree the ability to segment data is key to a success online marketing strategy.
3. Google Analytics solutions gallery allows you to import other people’s dashboards and segments.
4. Google now allows you to segment by visit or by user across multiple visits.
5. It is important to remember that an advanced segment is based upon a sample of data. So be careful about making critical decisions based on a small amount of data.
To view Anna’s slides, click here.
Real Time Business Insights by Ryan Gallagher, IOVOX (@ryangllghr)
1. Big data refers to a collection if data so large that it becomes difficult to process. It requires looking at this huge stack of data and pulling out what is most important to you.
2. You must look at the data within context.
3. Using a proxy phone call allows you to collect more data – this allows you to value the leads more accurately.
4. Mapping trends – you can view what factors effect visits. E.g. On sunny days mobile visits increase whereas on a rainy day, desktop visits increase. This allows you to determine which stock you should advertise at which time of the year.
5. Consider how you can used the data gathered to influence sales.
Customer Value Optimisation by Andrew Hood, Lynchpin (@lynchpin)
1. We should move away from trying to improve our sites and focus on increasing customer value. This should be used alongside CRO.
2. In a lot of businesses acquisition and retention are often set up as antagonistic functions, however, these should be focused on simultaneously.
3. There is no such thing as the average customer, therefore it is important to use predictive Analytics to target valuable customers. E.g. Pulling in information such as gender.
4. Identify the “propensity to stay” actions to see what works best to keep customers as customers.
5. Your customer value analysis needs to be determined by in depth statistical modelling.
What is Possible with Call Tracking by Ali White, CallTracks (@AlistairWhite)
1. 83% of shoppers need support when completing a transaction online. 71% of this support is needed within the first 5 minutes. 61% of this support is carried out by telephone. 70% of users call businesses directly from the search results.
2. You can use different tracking numbers for different types of media to help identify which media is the most effective.
3. Dynamic tracking numbers allow for individual visitor level tracking. This allows you to determine which path they took through your site before calling.
4. If a user phones you up, it is a good idea to exclude them from your remarketing. This is where CRM integration is essential.
5. Understanding all the different touch points online and offline, will allow you to invest in the right areas of marketing.
Dealing with Not Provided by Dr. David Sewell, FreshEgg (@seoeditors)
1. Not provided traffic isn’t necessarily a bad thing as it is causing various marketing channels to speak to one another and it is forcing us to look at the wider picture and use other methods of targeting.
2. Common approaches are; % breakdown of brand terms, % breakdown of visits with keywords, landing pages and finally, advanced filters showing landing pages and rank.
3. Alternatives; look at location to estimate location based keywords, look at device to estimate keyword length, look at landing page to estimate keyword category.
4. The not provided tool allows you to estimate the keywords of your not provided traffic.
5. The not provided tool takes these pieces of evidence and passes it through the Google Prediction API and it will present you with a estimated keyword. Do this with keywords you currently have data for, if it is correct enough at this stage then you can use it to estimate your not provided data.
Leveraging Your Competitors’ CRO Efforts by Yousaf Sekander (@ysekand)
1. Review what tests your competitors are running and consider how you can use the results they achieve to save you the budget and time of carrying out the tests yourself.
2. If a competitor is carrying out split testing, when they stop running one version you can determine what version won and then implement the successful elements.
3. Cromonitor.com – allows you to automatically monitor the split testing which is taking place on a website to avoid you having to monitor the the website’s manually.
That’s a wrap! We’d love to know what you thought of the conference or our notes so please let us know.