Google Analytics User ID: Duplicate Users, Low Ecommerce Conversion Rates & High Bounce Rates

This article explains a solution for high bounce rates, low pageviews per session, and low ecommerce conversion rates in your Google Analytics User ID view. Chances are, you’re giving single users multiple User IDs.

Background

Google Analytics’ User ID feature purports to identify and then match up all sessions and onsite interactions for a single user across devices.

Unfortunately the documentation from Google on the User ID is not great and is more suited towards simple implementations. I work with multinational enterprise retailers and often find unexpected behaviour.

For instance, I have had User ID live on one retailer for around one month. Reviewing the data for the User ID view in comparison to the main GA view, I can see that:

  1. Pages per session are far lower for the User ID view
  2. Bounce rate is far higher for the User ID view
  3. Transactions and Conversion Rates are far lower for the User ID view

Investigation

Digging around a little further I can see some odd behaviour. The Landing Pages report is a great place to start.

According to the User ID view, no session starting on the homepage lead to a transaction. Not one! Compare that to the main view and the thousands of transactions I can see there. Something is amiss.

The Landing Pages report also shows me that some of the most popular landing pages are in the account details and order summary pages – those pages users see just after they’ve logged in.

So the obvious place to start looking is the login area. Remember that only hits that have a User ID value (or those earlier in the session, with session unification turned on) are recorded by your User ID view. Something is obviously happening with the User ID around logging in. Are they losing their User ID?

Diagnosis

I then test a number of different scenarios:

  1. A new user, visiting the site, adding some products to basket, creating an account and checking out.
  2. A returning user, not logged in, visiting the site, adding some products to basket, logging in and checking out.
  3. A returning user, logged in, visiting the site, adding some products to basket and checking out.

In each case I have the inspector open and am querying the contents of the data layer (via the command ‘dataLayer’).

What I can see tells me exactly what has happened. Why the bounce rates are high, pages per session are low, and conversion rates and transactions are low for the User ID view.

It all comes down to how the persistent basket – the functionality that keeps products in your bag from visit to visit – works.

On some platforms, as soon as a user adds something to the basket they get assigned an ID. This is for your back end systems – and is likely the value you are using for your User ID.

In scenario 1, the user gets assigned the next available ID within the back end system. This is kept as their ID when they create their account and checkout. That’s fine.

In scenario 3, the user has their ID available in the data layer from landing on the site all the way through to checkout. That’s also fine.

Scenario 2 creates problems though. The user comes to the site and isn’t logged in. Maybe they’re using a new computer or phone, or borrowing a friend’s tablet to browse. They add a product to their bag. This triggers the persistent basket functionality and the back end assigns the next-available ID to that user. This is picked up by GA and used as the User ID. Session unification assigns their earlier pageviews to this specific User ID.

Then that user has to log in to complete their checkout. They already have an account from registering earlier in the year and so login. Upon logging in, the back end pushes their original ID to the data layer – the ID of the account they’d created earlier in the year, a different ID to the next-available ID necessitated by the back end system.

This new ID is picked up and used by GA as the User ID. To Google Analytics, a new, unique user has visited the site, landing on the ‘Order Summary’ page. And the previous user, who landed on the homepage and added products to his bag, has left.

Hence:

  1. High bounce and exit rates
  2. Low ecommerce transactions and conversion rates
  3. Low time on site metrics

But, most importantly, the single individual has been seen as two different users by the GA report that promises to stop this happening.

Solution

The solution is not particularly difficult. Depending on your platform and setup, it is just a simple matter of configuring the User ID (&uid) field or variable to be set only on certain pages – or, more simply, to not be populated on the adding to bag/persistent basket functionality. Depending on the platform, you could also configure it to not create, or push to data layer, an account number upon add to bag. The GA session cookie should still persist throughout this so you should not see an increase in sessions per user or other such metrics.

To Quick View or Not?

Quick View is a feature common to many ecommerce websites. It’s been included in many briefs I’ve received. The assumption appears to be that it must be good, as it is on XYZ competitor’s site. I have, however, seen very varied results for its effectiveness when tested. For instance:

One test we did saw a very positive result for Quick View in generating sales with one retailer.

On another site, I ran a test that saw an increase in conversion rates when removing Quick View.

Both results seem to contradict one another. But they make perfect sense when  you consider the different user journeys of each business, and the role the website plays in each user journey.

The former business is borne of catalogues and mail order. It’s ecommerce is successful across the globe, but they still have a dedicated catalogue readership. Many users visit their website to find something they’ve already seen in the catalogue, in some cases going straight to search the site for a SKU. These users have already read the product details in the catalogue. They just want to be sure that the product they’ve found online is the same as the one they viewed earlier in the catalogue. Quick View gives them that reassurance and so they add to basket.

The second site is that of a high-end high street retailer. For this business, the user comes from across the country to their site for fashion advice and leadership, to obtain ideas and inspiration, and to explore the catalogue. Many users aren’t even set on buying when they visit the site. Exposing users to more product information and imagery – i.e. showing them the product detail page as opposed to Quick View – gives the business more opportunity to convince users to buy.

The high-street retailer’s site also contained many, many desirable products. It’s a site where users don’t struggle to find something they like. If anything, they struggle to limit their desire to a single product. Quick View exacerbates this: it lets users browse from desirable product to desirable product. Rather than focusing their attention on a single product, Quick View encourages users to view many products and introduces indecision to users.

There is a famous psychological test that shows how choice can be bad for decision-making. You can read it here.

The effectiveness and appropriateness of Quick View should be considered in the context of the role the website plays within the customer journey. The following five questions are useful to ask when considering whether or not to use Quick View.

  1. Why do people visit your website?
  2. At what point in the customer journey do they visit?
  3. How do they find your website – organic, non-brand search keywords, direct traffic, referral from another website?
  4. Do they have previous awareness of your products?
  5. Do they use the website to buy or browse?

TL;DR

Quick View can be an effective tool when used in the right situation. That situation can be determined – and Quick View’s success predicted – based upon the role and position the website takes in the user journey.

Literary Theory: Death of the Actor

Roland Barthes famously declared the death of the author, a rejection of traditional literary criticism that focused on biographical and psychological evaluations of the author when interpreting the work.

I thought about this when I was watching Good Will Hunting the other night. It isn’t common for critics or viewers to evaluate the film in the context of life of the screenwriter – in fact, in most cases the screenwriters are largely overlooked. So it may be that the author is not required to theoretically die when evaluating cinema.

Actors are, however, a different story entirely. In fact, our knowledge of which actors are significantly well-known enough to demand a recurring or significant part can undermine the suspense of a film significantly.

Consider the scene in Good Will Hunting where Matt Damon’s character starts to assert himself over a bar room bully. In the background is a girl, drinking at the bar. This should be a largely unnoticed detail – Damon’s character has had no interaction with the girl at this point – but the viewer understands that she will be significant shortly. Why? Because she is played by Minnie Driver.

Directors do occasionally play with their viewer. One of the best pieces of casting I have seen in recent years was HBO’s Band of Brothers series, which principally used unknown British actors to play US GIs in an American drama. With no stars in the cast list, viewers had to be ready for any character to be killed off at any point.

And – this may be the only time I mention a Steven Seagal film in a discussion on literary theory – Executive Decision pulled a blinder on viewers by having Steven Seagal’s character killed off in the first 20 minutes.

Generally speaking, though, when a viewer watches a film its apparent who the stars are and that bestows a different context upon the actions their characters do and receive.

Finally – watching Good Will Hunting again I liked how it represents a sort-of-hackneyed version of Maslowe’s Hierarchy of Needs. Matt Damon’s character goes from fulfilling visceral needs (fighting, drinking), to intellectual needs (maths), before chasing his highest, metaphysical needs (love).

The Real Mobile User Journey

I’ve been part of an agency specialising in responsive design since 2011/12. One of the common misconceptions we’ve seen is that getting a site to work on mobile is as simple as

  • Rendering the page on a small screen
  • Shrinking buttons and CTAs
  • Resizing imagery

The more savvy of you may realise that’s far too simple. For instance, what about:

  • Text and emphasis selection? Italics, for instance, can be difficult to read on small screens

But there is more to consider. A mobile isn’t just a smaller screen. Its a device that is inherently different to a desktop and as such will have entirely different use cases in different scenarios. All of which place different demands on a user.

The mobile is a device that removes the barrier between the digital and physical world. It therefore interacts with, and is interacted by, the user’s environment far more than a desktop computer.

Mobile users have far more stimuli competing for their attention. Some of them are not critically important – a billboard on the side of a passing bus, a man walking down the street who looks like a user’s friend – but some are unavoidable. For instance, someone calling the mobile phone the user is browsing on. A WhatsApp notification. The bus they are travelling on braking sharply.

The inhernet physicality of the mobile device represents a unique user journey that needs to be considered when designing for mobile. A mobile user may be more easily and completely distracted from your website. So, what happens when they’ve responded to the Facebook chathead that’s appeared over their screen and they return to your site? Will they remember where they were in your checkout, what they were filling out and why? Will they feel inclined to continue filling out a 10-field form just to access your sale preview site? When the raindrops from the bad weather are messing up their touchscreen? Will they want you to send them your leadgen whitepaper as a PDF that renders badly in the browser – or would they rather you sent it as a link to their email for later reference?

Mobile usability should consider the user journey that users have on your site, within other apps in the mobile ecosystem, and also the journey that they are experiencing within the physical world at the same time. Good mobile design has much to consider – but that’s why it is a difficult discipline with few specialists.

What Scottish Independence Will Mean for Online Retailers

I’ve been thinking about what Scottish independence would mean for online retailers and digital marketers such as myself.

Personally I’m 100% in favour of keeping the union (for various reasons which I won’t get into here). But if Scotland votes for devolution then there will be consequences for online marketers and businesses:

  • Any geo-targeting will need updating. This will include on any tools such as Adwords, Webmaster Tools, Qubit etc.
  • Any ad copy & site promotion referencing the UK may need changing. e.g. An Adwords ad promoting ‘Free UK Delivery’ may need updating, or for an alternative ad to be launched in Scotland.
  • Related to the above, delivery costs and the value proposition may need to be looked at.
  • Scottish-specific businesses may want a different TLD; a .co.uk is no longer appropriate to shoppers based outside the UK.
  • Checkouts & address books will need updating to include Scotland as a different location than UK. Managing this for existing addresses may prove difficult.
  • A new currency will likely need adding to checkouts.
  • VAT and other costs will need to be considered by retailers; an independent Scotland will have powers to set these of their own accord. In fact, this might even be possible with DEVO MAX (maximum devolution yet not full independence).

Obviously the time frames for devolution, if its voted for, will give us all time to adapt. And whatever changes we have to make to the above will largely depend on what the third-party stakeholders such as delivery companies, payment gateways and advertising platforms decide.

But its worth our planning ahead now for what may be a more significant change than some of us had expected.

Pescetarian Escapades

So my latest shopping bag is solely pescetarian. No meat or poultry. Two reasons why:

1. I’m a complete hypocrite. When I watch a video of a slaughterhouse I find it very sad, horrific even. I wanted to buy a lobster but wasn’t comfortable with killing it. If I’m not OK to kill a creature myself, why am I OK with someone else doing it for me?

2. There’s a whole load of other foods, herbs and spices that I’d like to experiment with. Too often I make steak, burgers, chicken – the easy choices that I can cook by heart. Creatively I’ve often found that restricting options opens up new opportunities – I’m thinking the same here.

Even in spite of the first point, this isn’t a moral point for me. Its a fortnight experiment and I’m still going to probably buy meat if I’m out at a meal, wedding or other event. It’s just a personal decision – as much about trying to reduce my hypocrisy as my carbon footprint.

If you’re fine with eating creatures that someone else kills, or even killing those animals yourself, then fine. My personal feeling is that my own hypocrisy is uncomfortable – to me. So like I say, personal rather than moral. And an experiment. Let’s see how it goes!

The Problem With Magaluf Girl

I have just read The Sun’s expose of ‘Magaluf Girl. Magaluf Girl is a teenage girl who performed oral sex on 20-odd men in a resort nightclub in order to win a prize.

Shocking? Yes. But not for the reasons you might expect.

The girl was ‘mamading’, a competition organised by the Carnage Magaluf promoters. In the competition a participant – usually a girl – has to perform a set amount of challenges in a given time. These challenges are often sexual in nature. Magaluf Girl was promised a holiday if she won.

The girl’s subsequent attempts to win the competition saw her walk around the club, and drop to her knees every few seconds to take a different man’s penis in her mouth. The crowd cheered. The DJ baited her on. Someone filmed it on a mobile phone. The Sun published it. The girl completed the challenge. The ‘holiday’ was actually the name of a bottle of cava she won.

Whatever this girl chooses to do with her body is, in my mind, her business. I can’t understand why you’d ever want to give 24 blow jobs in a couple of minutes but hey, one of the costs of equality and respect is that you get to do things I wouldn’t want to.

I don’t particularly blame the men involved either. I can’t understand why they wanted to put their penises in the mouth of a girl doing something like this any more than why the girl wanted to do this, but I can see how things may have got out of hand in the context of a huge binge drinking session, loud music, testosterone, peer pressure and – importantly – consent.

So what, then, do I find shocking about this? Simply, that it shows the extent to which we have devalued and degraded female sexuality. The two biggest culprits being Carnage Magaluf and The Sun.

I can’t ever imagine sitting around an office thinking up a new event for a client and saying

‘Yes. Blow jobs. We’ll get these girls drunk and then have them perform blow jobs for prizes. But the prizes won’t be what we promise them. They’ll be joke prizes. That will be good fun. Top banter.’

You have to be pretty fucked up to think of that as a good idea.

But hey, it’s all fun and games in a world in which the information a woman can use to make decisions about her sexual behaviour need not be right. And it’s absolutely fine to have that as part of your business model.

Just like its fine for The Sun to deride a girl for doing what she wanted to do with her own body, all within the law. And it’s fine to show her face to millions of people without her permission.

This is the same newspaper that wouldn’t print pictures of the Duchess of York topless on a beach. But they will show a girl perfoming oral sex. And they’ll deride a woman for being openly sexual but show topless women on only the third page of their daily paper.

Run that by me again? Oh right: it’s because women aren’t allowed to be in control of their own sexuality. That old chestnut.

Doing something that neither I nor the editors of The Sun would want to do doesn’t make something morally wrong. They aren’t the arbiters of female sexual morality that they might like to think they are.

Both men and women should have the right to do what they want with their bodies – provided it is legal and with consent.

For me, the real shocker of this story is the birth of slut shaming at an industrial level. Ultimately, I feel sorry for Magaluf Girl. She’s been exploited for money by both the promoters and The Sun.

Don’t Call It A Comeback

Some time ago I wrote about my long-term view on recovering from chronic injuries. Today I took part in my first running race for over five years.

It was the Pudsey 10km Challenge. Now, normally I’d scoff at a 10K. If you’re running 7, 8, 10 miles in training then what’s the challenge of running 6.2?

Pudsey’s answer to that question is hills. Big hills. In particular, 3 horribly nasty ones – one of which lasts around 1.5km in itself. Climb the hills, hurt the calves. Plunge the descents, knacker the quads. 800ft of ascent, up mud and down rocks, in only 6 miles.

As a designated flatlander this was all new, unexpected and bafflingly difficult. In 24 degrees heat and clear skies. By the second mile I was more in the red than Northern Rock.

Happily, I dug deep and got round the course inside my target time, coming in 252nd out of 408 entrants. So, by no means some glorious Rocky-esque success but in my own remodified modest view, and with an eye on the long-term, one I am very happy with.

You finish where you deserve to. There were 251 people in that race alone who were better than me. But I got round. And now I have a base upon which to build.

It was great to be out there again, after pulling out of, and missing, too many events in the last few years. To beat my target and finish ahead of lots of club runners was great. This was never an event for me to be at the sharp end of. This was purely just practice and the thrill of an event. Mission accomplished.

On The Rob

My apparently glacial middle-class existence was shattered today; leading me plunging through barriers I never knew existed and thrusting me into a decadent life of crime.

As mixed as those metaphors may be, they weren’t as mixed as the fruit salad I saw the woman in Sainsbury’s steal. Brazenly browsing the aisle, past the cream slices but not yet into the fizzy pop, she plucked an item that I could only define by its subsequent absence on the shelf, and placed it inside her coat.

Jokes aside, it was weird. Weirder still was that I didn’t do anything about it.

Firstly, I think she was stealing. I’m not sure. I didn’t see her take the item off the shelf and put it in her pocket. Officer. But you know when you get that vibe that something just happened that you weren’t supposed to see? Yeah. I got that.

But there was enough doubt in my head though to not want to tell anyone, lest I got it wrong. What if SWAT descended through the ceiling, abseiling ropes on the attack, only to find she’d just been putting her phone back in her pocket?

Secondly, there was a child with her. I guess it was her child (unless she stole that too, and that would be on an altogether different level of crime indeed). And if she was stealing, and it was her child, then maybe I perceive that differently. Stealing a treat for your child is different, on the moral scale, to stealing diamonds and flogging them to drug dealers, no?

I live in the city and see so much wealth, and so much poverty. You realize the fine dividing line between these apparently disparate groups. You understand that sometimes people need a little help.

Thirdly, the security guard. It was only when I got home that I thought maybe I’ve screwed him over by not telling him. Maybe his bosses will ask him why he didn’t apprehend the woman? Why are thefts in the store up?

Hmmmmm.

Whilst this is undoubtedly a prime example of middle-class liberal white guilt (probably sponsored by The Guardian and a bespoke farm marmalade, extra chunky), two things about the whole thing fascinate me.

One, that the brain can size up a situation and extrapolate such depth from what was a glance of under one second in duration. I don’t know if this is a good thing or not; you could certainly go crazy by over thinking things.

Two, that all people sit within a certain implied structure of social behaviour. This behaviour set is pretty much imposed on us. No-one really wants to stick their head above the parapet and stand out.

A similar thing happened to me a few weeks back. Walking through town I notice a man walking towards me. An odd angle. Something seemed off. My girlfriend pulls me quietly, but markedly, away. She tells me that the guy was about to pick pocket me.

Looking at him as he walks past me, I see he’s wearing a walky-talky. He’s presumably pretty pro as a thief, and working in tandem with some form of spotter. Yet I don’t say anything. Why? I mean, of all the situations, surely that is the one to shout: “HEY EVERYONE, THIS GUY IS A THIEF. WATCH YOUR BAGS.” No?

But you feel that implied behavioural structure. Flip that situation around and if someone started shouting that, I’d probably just think they were a crazy shouting man.

People are weird creatures, myself included.

Virtual Pageviews for Modals via Google Tag Manager

If you want to track views of lightboxes or modals as virtual pageviews in Google Analytics through Google Tag Manager, it can be quite a confusing process. Here is how I do it.

Why This Guide?

Whilst Google Tag Manager is undoubtedly a good tool, it sits at an intersection between developers, marketers and analytics guys. I’m in the latter two camps; Google Tag Manager speaks in the language of the former camp. Furthermore, most of the guides I saw focused on setting up events rather than virtual pageviews, or on pages that had single clickable elements on them.

This guide is for pages with multiple clickable elements (e.g. a product page that has ‘Add to Bag’ and ‘Reserve in Store’ buttons) of which only one is to be tracked. We want to track these as virtual pageviews, not events, because we can use virtual pageviews in a conversion funnel.

So How Do I Do It?

We’ll need to set up two different tags that work together. It is best to test this set up in UAT rather than on the live site, as it aids the debugging process (discussed later).

Tag 1. The click listener.

This is the straightforward bit. Set up a new tag of the tag type ‘click listener’. Set up the firing rule for ‘all pages’ (this will probably be the condition {{url}} matches RegEx .* ).

What this means: You’ve just set up a tag that listens for clicks on all pages.

Tag 2. The virtual pageview.

This is a bit more involved but still straightforward if you follow these steps.

  1. Set up a new tag of the classic (or Universal) Google Analytics type. Use whichever type relates to your GA set up (classic or Universal).
  2. Set this up with the Track Type ‘Page View’, and under ‘More Settings’ specify the virtual page path you want to be sent to Google Analytics.
  3. You need to now set up a rule that will see the virtual pageview sent whenever the conditions of the rule are met.

To do this you set up a single rule consisting of two conditions. Those conditions are:

  1. {{event}} contains gtm.click
  2. {{element classes}} contains (whatever the element class of your button is). You can find this out by highlighting the button the page, right-clicking and choosing ‘Inspect Element’. Elements are just a posh word for ‘things’ – don’t get too worried about the terminology.

What this means: the virtual pageview will be sent to Google Analytics whenever your click listener tag fires on an element that has the specified element class.

Debugging

Now preview and debug the site. Keep Google Analytics’ real-time reports open and track your progress – you want to check that GA is recording the virtual pageview at the right moment. If you’re doing this in UAT it will be easier to see the impact your specific behaviour is having because it won’t be confused with other site users.

You can also use the Google Tag Manager debug pane to check tags are being fired at the correct times. If you aren’t familiar with code it may take you some time to identify the specific element class of your button and you might find your virtual pageview tag is firing too readily, or not at all. Remember that the class has to be as specific as possible here – that of the button itself, not of the page, div or wrapper surrounding it.

Remember that with any amend you make to the Tag Manager set up you’ll need to re-preview and debug it.

Conclusion

Play around with it and you’ll get there. Let me know if you have any other questions, and if you’ve found this helpful, take a look at my other articles about ecommerce and digital marketing.

Thanks!