Website analytics provide a wealth of information, and even the free Google Analytics is well supported by myriad graphs and charts dissecting website traffic in just about every way imaginable. Leveraging this data to improve electronic direct marketing (eDM) efforts and refining site structure, however, seems to be a challenge and an ignored discipline.
There are aspects of mobile website usage that, when combined with good analytics, can provide a very valuable insight into the behaviour of your customers and should be used to refine a range of customer engagement strategies. Many of these insights can be derived only from mobile usage.
I was invited recently to join a US-based team that was given the task of analysing their site statistics and using this data to refine many aspects of their marketing tactics encompassing email, SMS, letterbox catalogues, in-store promotions and even site design adjustments. In the past, the primary use of analytics was to see how ‘hits’ or sessions were increasing and when to look at conversion rates to sales (the site was an online store generating around 2.1 times its largest physical store in a network of 780 stores).
The approach to reviewing the site traffic was simplistic and focused on conversion rates and the raw dollars being generated. Very little was done to link back eDM executions and their content to the impact on sales versus traffic.
If a particular promotion generated a 20% traffic spike, everyone got excited. But nothing was being done to see if ‘normal’ conversion rates increased or decreased and how deep visitation went into the site. No effort was being applied to refining the site structure to adjust the journey for a consumer, nor was there any effort to discriminate between mobile and desktop.
Some very interesting insights emerged from the team analysing six months of data, which was provided along with a detailed schedule of all eDMs, product specials, above the line activities and so on. The following paragraphs are some of the more intriguing facets of that analysis work.
With mobile, there is the ability to collect and log location-based data. This will prompt the user for permission to ‘locate’ their device. Interestingly, over a six-month period consumers agreeing to be located went from 32% up to 67%. There are so many applications on mobiles that request this data that consumers are becoming less fearful of accepting.
This location-based data, especially if logged with every progressive page transition, provides some very valuable data when combined with things like search terms used on the site.
Analysing the search data delivered some fascinating information. The search terms were simplistically broken up into three categories: category type words like ‘jeans’ or ‘televisions’, model-specific phrases like ‘Belkin N600’ and then the rest. On the desktop site, model-specific phrases were less than 17% of all searches performed, but on mobile they accounted for a staggering 67% of the phrases.
We then looked at the locationbased data and started mapping where consumers were using specific make/model search terms. I am sure there could have been an easier way, but for us this was a very arduous task! Due to the manual nature of what we were doing, we only had a small sample space, but we found that in nearly 90% of cases when a model-specific search phrase was used the consumer was close to or inside a competitor’s premises. The consumer was price checking!
It’s important to appreciate the number of sessions analysed was relatively small compared to overall traffic. But on mobile, where the consumer was clearly in a residential or office area, the behaviour was more aligned with desktop, and search terms became more generic and less targeted.
Think about what you present to a consumer searching your site for a specific make or model of product, on a mobile you are able to locate. It’s a consumer that knows what they want and are simply after the best buy, or are comparing models and makes for prices. Regardless, they are more than just a ‘site visitor’ browsing through your offering.
Imagine if at this stage in the site design you implemented some ‘close’ sale strategies. You could de-clutter web pages to focus on their requirements. With your top three or four competitors’ locations loaded into the back-end systems, you would immediately be able to discern if someone on your site is inside a competitor’s store and then tailor the search result to close the sale. Or, if they’re located near your own retail outlet, you could drive them into your store.
If there is good competitive understanding on prices, then ensure whatever price you deliver is the ‘best’ price and make it a time-limited offer if less than usual; for example: ‘Our regular price is $X, but for the next 30 minutes you can have it for $Y’.
Drive the customer into your store with special bundles that make the price check more complicated, but your offering more attractive, such as: ‘If you come into our store just three minutes away from where you are we can offer the TV with gold-level HDMI cable, personal video recorder and free delivery for just $X’.
The interception of search terms with location-based data delivers a new level of insight about the digital consumer. It gives you back the ability to use sales strategies to close a deal that can be customised by behaviour almost in the same way as in-store staff would approach a consumer. Remove the generic one-size-fits-all approach to their online experience.
A similar strategy was implemented for click-through off eDM pieces where location-based data was known. The consumer location was analysed to discern the type of page and content being presented and was customised based on proximity to a physical store, time of day, whether they were in transit or a fixed position and the type of promotion being delivered.
Time of day was very important and saw some big variations in search terms, navigation behaviour, page views and transaction rates on mobile. Multi-product purchases on mobile, for example, were significantly higher after hours than during normal shopping hours.
The next interesting discovery was when we started looking at basket composition by desktop and mobile, along with conversion rates and shopping cart abandonment. Per transaction, average session page views were almost 50% of that for desktop. In other words, on mobile, consumers browsed far less before concluding a purchase. So, on mobile we turned off banner adverts and simplified navigation and saw a 1.1% increase in conversion. Unfortunately, I haven’t seen the data over a long period, so I am unsure if this is a related spike or not. But hopefully what this example is doing is getting you thinking about how to take the volumes of data and actually use it to refine and improve sales.
The other interesting thing was the structure of the shopping carts between devices. Single-product purchases on mobile were seven times higher than on desktop. This is where some statistics can hide the nuggets of valuable information. The early site reports showed ‘average items per transaction’, but we wanted to know how many consumers purchased a single item.
The other piece of data we had was cart abandonment rates, which were noticeably higher on mobile than on desktop. I am sure there are many ways to explain this, given mobile online usage is far more ‘interrupt’ driven than someone sitting at a PC. When your phone rings, you receive a text message or arrive at your destination, then the online session is interrupted.
We implemented two changes on mobile. This is always dangerous, as it can be hard to correlate what impact each change had. First, we changed mobile to be single-click checkout by default. After checkout (we removed the shopping cart), consumers were offered other products to add to the order. This reduced abandonment rates by 9% and had no material impact on multi-product purchases.
The second change was to ask for an email address at the time of displaying any product detail page. This was intended to enable a follow-up email for anyone not concluding a sale or abandoning at checkout. It also provided the ability to link the consumer to previous purchases, gauge their ‘loyalty’ status and engage them more personally.
Originally, we actually asked for email at the start of a session and found less than 3% entered. When prompted at the product-level page, on the basis of ‘we can email you all this product information’, it jumped to a 35% collection rate. Of those emailed with a ‘hot offer’ after abandoning a checkout process, a further 4.3% were driven in-store showing the emailed coupon or went back online to conclude the transaction.
The analysis, recommendations, site changes and rule guidelines extended to a 200-page report and changed entirely the way the in-house team now thinks about analytics and website design. The significant efforts that physical retail invests in store design, merchandise layout, structures, approaches, visuals, consumer flow, register positioning and pricing has traditionally never been so deeply applied to online. But after six months of many changes and a significant ‘rules engine’ implementation, the overall sales volumes and revenues have increased far more than natural growth and have generated a measurable increase of in-store foot traffic.
Use the wealth of data that can be collected and leverage all aspects of the online experience across different devices and times of day to ‘personalise’ every online experience. Leverage location-based data and start to plot access trends by location and device and transaction ‘styles’. Online is not a one-size-fits-all experience. Online websites are being positioned as ‘middle ground’ to try to suit all consumers with a compromise. It doesn’t have to be that way. Marketing to the mobile audience can be very granular and very tailorable with the impact on sales potentially quite significant.
View at the original source
View at the original source
No comments:
Post a Comment