Mobile Traffic Patterns in Paid Search
With smartphone adoption and mobile internet usage continuing to grow at a rapid pace, it is increasingly imperative for advertisers to understand the nuances of this market segment and adapt their strategies to ensure their ad dollars are providing value on all fronts. We’ve previously discussed the performance gap between desktop and mobile paid search ads and noted that issues both technological (tracking, unoptimized mobile sites) and of user intent (local focus, research only) are leading to low online conversion rates from mobile users.
In trying to get a read on the balance of those issues and as a consideration of the incrementality of mobile traffic, we’ve seen others, including Google, contrast time of day and day of week usage patterns by device. Not content to rely upon the data of others, we decided to take a look ourselves.
Before going any further, we should point out that due to differences in user behavior, we are making a distinction between mobile and tablet devices, with the former essentially meaning smartphones and the latter being something that looks a lot like an iPad. Google is making changes in that direction, but when they and others refer to “mobile,” they are often lumping the two together.
Time of Day Usage
Below are hourly ad click levels by device for a basket of our paid search clients. Because desktop still dominates all traffic, these figures are normalized such that the average hourly level for each is aligned.
We get a different impression of mobile usage patterns here than what we’ve seen presented by others. In our data, mobile ad clicks trend pretty similarly to desktop throughout the day. If we subscribe to the argument that mobile trending differently than desktop is an indication of its incrementality, this would be a strike against that notion. Rather than mobile traffic ebbing and flowing inversely to desktop as users find themselves with or without access to a traditional computer, both trends appear to owe far more to our circadian sleep rhythms and work schedules.
Tablet traffic, on the other hand, does show an obvious divergence from the patterns seen in mobile and desktop. It looks like iPad users are hesitant to risk toting their prized possessions around town as tablet ad traffic hits a plateau during the work day, but then begins to spike as those of us on the East Coast start returning home from work. Consumers don’t appear hesitant to use their tablets in bed though, as tablet traffic between 9 and 11pm ET runs more than double the tablet average for the rest of the day.
Viewing both mobile and tablet traffic in relative terms to desktop accentuates the hourly trending differences and may ultimately be more actionable for advertisers. Here we are looking at ad click share by hour:
Mobile share clearly spikes late at night and into the early hours of the morning. For these clients, mobile comprises a low of around 5% of all paid click traffic at 10am and highs of around 9% between 1am and 5am. Though not obvious in the graph above, tablet traffic share undergoes an even greater swing in percentage terms with a low of around 1.5% in the early afternoon and a high of nearly 4% around midnight.
Again, from what we are able to measure directly, mobile traffic simply does not perform nearly as well as desktop or tablet traffic. Advertisers using dayparting in their bidding should be segmenting out mobile in their campaigns and analyses to ensure that any ROI patterns seen throughout the day are not simply a result of relative changes in the amount of mobile traffic.
Day of Week Usage
Looking at day of week traffic trends, we see a picture that’s more consistent with the general wisdom that mobile traffic trends very differently from desktop. To start, we again look at normalized click levels with average daily clicks aligned for each device type:
Both mobile and tablet clicks show a relative spike on the weekend, while desktop clicks peak on the Monday return to work before consistently falling through Saturday. These figures are a little noisy as we are only looking at a 5 week period, but the weekend effects were consistent across clients including the surprisingly deep nadir in tablet clicks on Fridays.
Considering that issue, we are looking at ad clicks and all devices show below average traffic on Fridays, suggesting a lower propensity for shopping in general. The hourly traffic data also suggest that tablet usage has the strongest link to times when we would expect a shopper to be at home (i.e. not Friday).
Looking at click share by day, our mobile and tablet figures are smoothed out by being put in context to desktop:
Tablet traffic share is pretty consistent throughout the work week before spiking over Saturday and Sunday when users are most likely to be home. Mobile traffic share also spikes over the weekend, but only after making gains late in the week when users are more likely to be out, but not at work.
Once again the most actionable takeaway from this type of analysis is probably in ad bidding. Any algorithms or rules attempting to apply week-parting should be cognizant of changes in mobile share by day as it has the potential to skew overall ROI.
Attempting to infer the precise incrementality of mobile traffic based on usage patterns alone is a bit ambitious, but the data suggest that a substantial percentage of mobile usage occurs during times when users likely have access to multiple devices. In those cases, mobile traffic can be considered cannibalistic of desktop to a large extent, but that assessment may ultimately be more important to web development priorities and attribution analysis than advertising tactics.
Your target audience exists across mobile, tablets and desktop in some combination at all times and if you’re not in each space, you’ll be losing out on some portion of your potential traffic. Mobile vs. desktop is very different from paid search vs. organic, where a weak showing on one front can be compensated for by the other for each individual search. We wouldn’t expect a large percentage of users to switch devices just to see better search results.
Whether or not a mobile ad click today on a particular keyword would have been a desktop click three months ago or whether it is entirely new traffic, it is still a mobile click now with all of the challenges we face in getting a conversion from a user on a tiny screen. Making sure your website and advertising efforts are positioned well to appeal to users no matter how or where they choose to browse the web is critical. Google claims that 79% of large advertisers don’t currently have a mobile site. That just won’t cut it for long and it’s a real opportunity for everyone else to gain some ground.