2013: The Year Of Conversion Rate Optimisation

Conversion Rate Optimisation

There seems to be growing interest in the area of online Conversion Rate Optimisation. Here are a handful of tips on how to get the most from your site 

 

1) Keep it simple, stupid

Online users don’t read webpages. They skim them for what they’re after. If you (as the web owner) don’t make this clear enough to them – they will leave! 

Keep text to a minimum, use a simple navigation structure and help users find what they need… which leads us to …. 

 

2) Prominent Call To Action 

One thing is for sure – users come to your site looking for something. Steve Krug said optimise your site so that the visitor doesn’t need to think – so the web journey is effortless, intuitive and flows. 

This is important.

Most web owners presume users come to our site in a balanced state of mental equilibrium… the truth is, we don’t know what mood our users are in when they are visiting our site. Because of this, we have less opportunity than we think to impress users before they wander off – so make the call to action clear and easy to interact with!

 

3) Stay Social

Make use of reviews, feedback ratings and other social proof. People can be cautious when they are the first to commit to something. Ease their hesitance by showing how many people have purchased your item, ‘liked’ it or are talking about it at the moment. Social proof is easy to implement and the benefits are vast. 

 

4) Jargon Buster

I have some news for you – people outside of your company do not speak the same way you and your colleagues do!

Keep jargon on the site simple and at a lower reading age. This caters for a more varied audience and alienates fewer people. Remember when you were in that electronics store and the salesman confused you so much you wanted to head for the exit? – Same rule applies to your website!

 

5) Special Offers and Suggestions

A little while ago, people used to feel good spending money. Nowadays, people tend to feel better about saving money. Show customers what special offers you can give them and how much they have saved. If you have an item that is likely to be required (Eg – selling a torch which may require batteries) bundle the items together and suggest this to the user. 

By doing this, you will sell more and also help the user from making an repeat visit after they realised they didn’t initially get from your store what they needed!

 

The best advice is to keep the web experience simple, clean and clear. Site which succeed most put the power into the users’ hands (think Google). Keep users happy by keeping your site easy to use!

 

Follow me on Twitter Connect with me on Linkedin

The Threat of Real Time Data

I have heard a lot of discussion on real time data as of late. A lot of people are raving about the benefits of it and, whilst these benefits are not without merit, what must be considered is the potential flaws and risks of real time data. For a lot of people, getting in of a morning and running various dashboards for the day before can seem a little ineffective at times because yesterday has happened and we feel like we should be reporting on the ‘here and now’. Well, whilst it’s perfectly fine to feel this way, I would encourage you tho think of the benefits with reporting with this small time delay.

1. Waiting to report the day after allows the analyst to crack on with the actions raised from other analysis, rather than having to spend a lot of time going through manual reporting and then probably do exactly the same after that. If an automated Dashboard can captured all the data you need and it is near the close of play, it can often be worth just waiting for the next day to generate all the data automatically – rather than going though it manually on the same day (taking urgency into account of course!)
2. If an Analyst has their head deep inside the live data, when will they get the opportunity to look at the bigger picture? Tactics are very important – but so is strategy. The same Analyst should be looking at the bigger and smaller picture so they can understand contexts and comparisons more so than having one Analyst for the bigger picture and another for the smaller picture for example.
3. Most business run into the issue that they want Real Time Data but do not have the ability to implement real time action. For the simple reason of resource management, it’s wiser to maximise by gathering relevant data, analysing and then implementing. These decisions should rarely be made real time – as any changes should go through a Quality Assurance process, to ensure they are the right actions.

I’m not against Real Time Data. I’m actually for it – It provides business with more options and I’ve never been in a situation where options have made a scenario worse. The thing that I am against is people asking for something, when they’re not really sure why.

If you are thinking about paying for real time data – ask your self what you are really going to do with it and why you need it before doing so.

Follow me on Twitter Connect with me on Linkedin

UX – Give Users What They Want

It seems, from the many articles I’ve read online lately about conversion etc, that companies are coming round to the idea that User Experience (UX) is really the crux of why online users convert.

Google recently covered this topic in three short humourous videos (Topics covering; Site Search, Online Checkout and Landing Page Optimisation). I thought they did this very well and you’ll see from the videos that common issues are the balance between how much companies go for ‘the upsell’ and how much companies put their users first and their profits second (sometimes companies put their profits first as a conscious decision, even though they’re fully aware it impacts UX, simply because it makes money for the company – Amazon do this with recommended products, yet they often score very highly in customer satisfaction studies).

So to what degree do companies really want to put the customer’s first? Many companies will make this their mission statement – yet will still default annoying extras etc, for the sole reason that it earns the company more money.

I personally believe that you can get away with both. Amazon have made the payment process of their site one of the best online (if not the best). I am not brand loyal to Amazon, yet I find myself shopping more regularly there than anywhere else online – simply due to the UX. Despite this, Amazon still have features such as ‘Recommended products’ and ‘Other people that buy X normally buy Y’. Don’t get me wrong, these features can be annoying – but the rest of their journey is so smooth it really doesn’t impact the overall experience.

Online companies normally sell out to advertising. Google didn’t. Google’s ethos early on was, ‘Give the internet user what they want, simplicity and clarity’. Take a look at the example below;

Google keeps their layout clear and precise

Google literally asks the user nothing more than – ‘what is it you want?’. There are not flashing adverts or banners. There are no major design issues (the user doesn’t even need to move the cursor to the search box – as Google have defaulted this to make it even easier). I like this. In doing this, Google have more users willing to search, making the AdWords advertisements more relevant then if it were to display these adverts pre-search. Google have a nice balance.

On the flip side, some companies make changes and simply don’t realise that such changes impact the user. Take WordPress for example – I used to be able to simply write a blog post in a large window and see most of my text as I write. Now WordPress have implemented a design that wastes space and means the central feature to WordPress (blogging) is harder to do (I’m aware I can still blog in a large window, but why introduce this newer/poorer feature?);

Wasted space and small blogging space adds up to poor usability.

Small things like this add towards my evaluation on whether to continue using a site or search for something which offers a better experience.

I guess largely it is down to the company to decide whether they truly wish to put the user first (and aim for traffic quantity and higher revenue through this tactic) or go for revenue (and receive higher order value, yet lower traffic and create poorer UX). One thing to note however, is that UX improves revenue – whereas revenue does not improve UX.

Follow me on Twitter
Connect with me on Linkedin

Why Pie Charts Suck

Pie Charts Suck

I don’t like pie charts. I used to though. I am a big fan of reporting using context and not just data and I do not like it when someone reports a number and expects the listener to understand it with no context…..

This exactly the trap that your average pie chart falls into. When I read data, I like to go through everything – not everyone likes doing this. Often is the case that Managers and Directors just want to get an idea of what ‘it is’ that’s important and why. The pie chart offers no such support in terms of why, but merely states; This is what it looks like, at only one chosen point in time, with no comparison to any other time period or metric. For me, this isn’t good enough. I have often looked at data, for it to wildly fluctuate another day – this is how trends are formed and the pie chart crumbles when it comes to this.

One thing that makes me smile is when people tell me they like pie charts because it makes data look simple. Thanks for the opinion. It’s doesn’t.

Take the following example of a random data set – each segment of the pie chart looks about the same size, yet on a different type of graph, the results are clearly different;

I’m going to tell you something that not a lot of people will like. Size matters. Don’t listen to what people tell you, it does.

To explain what I’m going on about, often when you enlarge a chart, the results are even more clear. Take the following pie chart (BBC: European Election Results) as an example;

I don’t know what you think … but my thoughts are that I am none the wiser for the enlargement. Now check out the same data, enlarged too, in a different format;

In the second graph, it is a lot easier to not only see the data more clearly but the viewer can also see the close fight between 2nd, 3rd and 4th place – whereas this is all guesswork on the pie chart. Due to this issue, it’s also harder to make comparative judgement on a pie chart.

So when people tell me that the data is clearer on a pie chart, maybe they actually mean that they like the pretty colours – because, for the reasons stated, analysis of data is a lot easier when the analyst can view the data clearly, compare clearly and make a trend too.

I will be thinking of a way to report using pie chart and displaying context, until then, we shall remain firmly separated.

Follow me on Twitter
Connect with me on Linkedin

Why Dashboards Are Not For Reporting

It struck me the other day, whilst looking at various software presenting data visualisation, that Dashboards maybe aren’t used as effectively as they could be at present. Often the analyst opens the Dashboard and has to go through loads of data, interpret it and then understand the significance in context – only to then report it and have anyone else need to do exactly the same thing (if they really want to understand the data).

I then proceeded to think, wouldnt it be better if I didnt have to do the analysis myself – but have the dashboard do it for me? This was around the same time that I became aware of a report feature within Google Analytics called Intelligence Events.  The whole point to Intelligence Events is that you don’t type in various metrics and KPIs which you wish to keep an eye on – Google Analytics does this for you. It tells you what you should care about. Intelligence Events works on logic that looks at change v significance – which just so happens to be similar things that I look for during my analysis.

Dashboards are often created with some traffic light colour coding; red for bad, green for good etc, but why have ‘-1%’ in red and also have ‘-100%’ in red too?! ‘-100%’ often requires a lot more attention than ‘-1%’. Having said this, some metrics wildly fluctuate in percentage too because of low volumes. Intelligence Events understands this and reports to the users the metrics, rated on significance, that have changed the most – either on a daily, weekly or monthly basis.

When a user opens a dashboard, it should be a simple and gloriously straightforward process. We should change the old method of dashboard reporting from;

1. Seeing lots of data all in one report.
2. Spending time analysing the data and working out what has changed, along with how important this is.
3. Reporting this change with suggested actions.

to;

1. Seeing the data clearly and logically displayed in-front of them.
2. Seeing straight away what metrics have changed the most and how significant this is.
3. Act accordingly.

Using the second setup, the analyst shouldn’t need to report the data – the dashboard would do it for them – whereas, as the moment, dashboards seem to be all about a place to house a lot of data, put into a slightly more readable format.

I genuinely believe that a decent analyst shouldn’t make things any more complicated. I believe that they understand complex issues and data, yet also understand how to relay this to the reader so that they understand. This method makes it even easier to see what really matters straight away – and allow us to tackle it.

Dashboards are not for reporting, in my opinion. Dashboards are for action.

Follow me on Twitter: https://twitter.com/WebsiteAnalyst
Connect with me on Linkedin: http://www.linkedin.com/in/websiteanalyst

How To Act Upon Test Results

So – picture the following – you have a budget allocated to a big project and a schedule for a number of tests to conduct, hopefully to validate putting your project into action. You get the results back – and they only agree with you 30%… What do you do…? Do you act, or scrap the project until more meaningful results are returned?

The answer is really a lot more reasonable than it may initially seem – just spend 30% of your budget. After all, the results didn’t come back 100% against your hypothesis – so you should act in some way. By doing this, you mitigate the risks and still allow yourself to dive into the opportunity – allowing further analysis of whether it was beneficial to do so.

Another example could be NASA, planning to send a rocket into space for exploration. If they ran a test that came back 99% against their hypothesis, then they shouldn’t cancel the whole project – they should just buy a telescope.

As I said before, this logic limits risk and still allows exploration of new ideas etc. So ultimately, test results shouldn’t decide whether or not you engage in a project – more just how much you engage in it.

 

Follow me on Twitter - https://twitter.com/WebsiteAnalyst

Why A/B Tests are ineffective ~ A Tip For Improvement

I am fairly new to the world of Web Analytics, having been in my role for little more than 6 months. One thing that struck me as odd pretty much from the offset however is the faulty logic behind A/B testing.

Many testers presume that, by running an A/B experiment on their site etc, they are proving something. They believe that ‘A outperformed B, therefore it will be worth implementing A from now on!’.

I’m not a fan of this – firstly, many great thought leaders have pointed out that correlation does not prove causation (http://en.wikipedia.org/wiki/Correlation_does_not_imply_causation), but secondly and arguably more importantly – What evidence, besides A performing better than B does one have to implement a new method? There are many temporary factors which may influence the test and influence the tester to come to the wrong conclusion.

To reduce this issue, I came to the understanding that it would be more effective if the tester operated a AA/BB test. So instead of a 50/50 A/B test, the experimenter would run a 25/25/25/25 test.

By running this type of experiment, the tester is reducing the influence that temporary factors play on the A/B results – by allowing us to see whether A is consistent with A and B consistent with B. If, after running the experiment, Both A tests perform better than both B tests – the experimenter will have far more significance with the test results as a result.

If both A tests do not beat both B tests, this would imply that the results are not as significant and a retest, or even a AAA/BBB test may be necessary. This will produce more confidence and trust in the data. Isn’t confidence and trust in data something all analysts aim to achieve and retain?

Connect with me on LinkedIn: http://www.linkedin.com/in/websiteanalyst

Follow me on Twitter: https://twitter.com/WebsiteAnalyst