MeasureFest has returned for 2014! We’ve had another insightful day of Website Analytics, Conversion Rate Optimisation and Website Usability. Check out my full write-up of the event, with 65 tips on analytics, CRO and usability.
For all talks and links to the speaker’s social profiles, read on!
Need For (Page Load) Speed
Neil Barnes, Head Of Insight And Traffic, Friday Media Group
- Page load time is the first impression your visitors get of your site so it’s important to make a good one straight away.
- 40% of users abandon a website taking more than three seconds to load.
- Research shows a one second improvement in page load time equals a 7% increase in conversions.
- Use the site speed reporting in Google Analytics to review page load times. You can drill down by segment to analyse each factor, such as browser.
- Be aware that load times are calculated from 1% of traffic by default so you can adjust the sample size by changing the analytics tracking code on your pages.
Targeting And Personalisation In MVT
Matt Althauser, General Manager EMEA, Optimizely
- 2% is the average conversion rate but by segmenting users you can find more worthwhile data, such as returning visitors may convert at 10% and mobile visitors may covert at 1%.
- Make sure your landing pages match the exact messaging in your paid Google AdWords adverts.
- Target a small portion of your website to validate tests before rolling out across other parts of your website.
- Minimise the conversion flow or process by removing or hiding steps that are filled in or used a lot.
Split Testing Vs. Personalisation, Is There A Winner?
Duncan Heath, Director Of Search And Strategy, Fresh Egg
- Segment audiences to give a personalised experience, for example those who like cake instead of chocolate should only be tested with cake products.
- Segment users by device and test for each device – what converts for desktop users may not convert well for tablet users. Likewise, iPad users may convert differently than those on android tablets.
- Despite personalisation, always be aware of changing the experience as a whole to not exclude others.
- Over personalisation can become creepy and make a brand lose its identity.
Data Layers 101
Jono Alderson, Head Of Insight, Linkdex
- Tag manager software came about in 2012 to help the issues of tasking development teams to apply tracking data layers to websites. This was previously a timely process that involved having a development team to add in code.
- A data layer is usually just an array, so adding categories and attributes, etc can be a simple process of segmenting data.
- You can capture demographics and previous purchases to gather data to target users more specifically using API data.
What’s The Weather Like? Measuring The Real World In Google Analytics
Simo Ahava, Production Director, Netbooster Finland
- Circumstance is a factor of online behaviour. Whether it’s brand mentions, time zones around the globe, or even the weather.
- To implement weather tracking you’ll need to use google tag manager and a weather API. You should also use session cookies to only target new visits.
- You will also need a visitor and the geolocation to pinpoint where that users is. This data is then parsed and stored in the data layer.
- Be aware that collecting this data can slow down your website with more script calls.
- Consider using weather forecasts to strategise your promotions for the coming week.
Most Common Issues With Universal Analytics
Dara Fitzgerald, Analytics Director, Measurelab
There are five issues with Universal Analytics that you should be aware of:
- Session handling – Universal Analytics will report different numbers as it will trigger new sessions from referral traffic differently. If you use a third party payment provider (so visitors are going offsite to pay) then when the visitor returns, a new session will be created. This could mess up the goal attribution, however you can add referral domains to an exclusions list in Google Analytics to fix this.
- Debugging visibility – UTM tagging can be seen in cookies in classic Google Analytics but in universal the UTM campaign information is not visible.
- User ID and cross device reporting caveats – users should be logged in to best track cross device paths and overlap. A user ID is not revealed in Google Analytics for privacy reasons. If a user logs out, the user session will break.
- Custom dimensions and metrics – you can have more custom dimensions in Universal Analytics compared to classic. Applying these can be different and sometimes more complicated.
- Migration plan – understand how Universal Analytics works for you and if the numbers look correct before switching over the code completely by running both versions in parallel. Classic Google Analytics data won’t be tuned off by Google for at least 2 years so plenty of time to test Universal Analytics.
Bad Breakups And Web Analytics
Phil Haslehurst, Marketing Manager, Decibel Insight
- Phil reporting data on all factors of his previous relationship to identify where things went wrong when her answer was ‘it’s not you it’s me’. Recording time spend talking on various topics, plotting places visited together and clothes worn.
- A declining conversions rate is something that will involve analysis rather than their being a reason given.
- Use heatmaps to visualise where users scroll to or click.
- Visualising data helps identify issues and gives insight into what needs to be actioned.
Is It A Bird? Is It A Plane? Is It A CTA?
Joe Doveton, Head Of Client Services, GlobalMaxer
- Photos of people help with call to actions (CTAs) but make them relevant. It adds human interaction.
- Text should be designed and written to match all steps of the user journey.
- Consider the use of imagery to distract and direct the users’ eye.
- Page scrolling can be encouraged with arrows, teasers and use of language.
- For global websites localise accordingly. For example, colours have different meanings in different cultures. Test CTAs for each location as they can differ significantly.
How To Completely Screw Up Your A/B Testing
Craig Sullivan, Optimiser of Absolutely Everything, Optimal Visit
- You need to look in the right places to identify opportunities and generate more valuable outcomes.
- Hypothesise aren’t all valuable, instead come up with insights rather than guesses.
- Prove your insights and predictions using data. You must integrate analytics and A/B testing software.
- A/B should be left long enough. Know the length of your business cycles. Decide in advance the test period.
- Check your A/B tests in different browsers and mobile devices to prevent issues
Conversion Rate Optimisation For Non-Ecommerce Sites
John Mills, Customer And Consumer Insight Manager, Jobsite
- Understand your business by becoming a customer. Review goal funnels to confirm the customer journey.
- Set all your KPIs, these can include multiple goals that are related to a priority goal.
- Report results in the language of your seniors. Demonstrate the value of a non-Ecommerce site by measuring against how your business makes money and spends money.
A Product Approach To CRO
Craig Bradford, Senior Consultant, Distilled
- In Craig’s experience, 80% of the success of CRO comes from the discovery phase.
- The discovery framework can be broken down into the following: learn, look, ask, and try.
- Learn by doing error analysis and identifying un-needed elements, confusing options or warnings.
- Identify what people are looking for.
- Ask people using surveys, interviews. Ask novices and experts to gain new insights from different types of customers. Use services, such as ethn.io for this. Your existing customers won’t always give you an honest answer.
Practical Data Integration: Tips And Pitfalls
Andrew Hood, Managing Director, Lynchpin
- Integration is needed to make up for the flaws that individual digital tools have.
- Context is important. A tag management system will import all sorts of data from your webpage
- Merge data such as transaction IDs and CRM data to get the returns by marketing channel
- Automate integrated data from each digital tool and produce reports that matter to you
- Traditional webstats and raw log files still provide useful data
- The pitfalls of using digital data tools is that you need obey privacy rules and have personally identifiable information in your data. Plugging in all the different data takes time so don’t integrate for the sake of it. Think about what you need.
Our SME Digital Marketing For, ULA: Spend – Measure = ROI / Brand (Explained)
Wayne Moussalli, Managing Director, D2 Integrated Digital
- For clients with small budgets the majority work is done on attracting visits and capturing data (leads). However, businesses measure agencies on sales. The steps between lead and sale are something that is difficult to measure.
- Look at the customer journey and identify changes that can be made to capturing leads that are more likely to convert.
- Call tracking helps attribute performance. Report on how many of these calls resulted in sales.
Social As A Grown-Up Consumer Insight Tool
Rebecca Carson, Head Of Insights, Brandwatch
- The history of social media Analytics: 1) volume and shares, 2) sentiment, 3) sentiment driver, 4) thematic category analysis 5) author segmentation
- Demystifying social data is important because you’d can expect to find exciting, messy, detailed and intimidating information
- Understanding your consumers is why you may want to analyse social data. it needs to be helpful as the time investment could be wasted
- Use what you already know: your target audience and your historical sales
Network Mapping And Measuring Relevant Influence
Beth Granter, Digital Consultant, Brilliant Noise
- Basic influencers identification: twitter bios, twitter conversions. Analyse these by klout score. Tools you might want to use for basic influencers search: followerwonk, twiangulate
- Be aware that follower value may not mean that person always talks about that topic, this is where advanced influencer identification comes into place
- Advanced influencer identification looks at who those people follow and the theme of their followers. Using NodeXL for Microsoft excel you can visualise follower networks
- Using the results from the advanced identification and basic identification you can create a percentage figure of how much they influence the particular topic, weeding out irrelevant influencers
If you’re looking to enhance your digital marketing efforts, please visit our Services page or get in touch, and learn what Koozai can do for your business.