Performance Means Business

SOASTA Blog

Subscribe to SOASTA Blog: eMailAlertsEmail Alerts
Get SOASTA Blog: homepageHomepage mobileMobile rssRSS facebookFacebook twitterTwitter linkedinLinkedIn


Related Topics: SOASTA

Blog Feed Post

New findings: Marketing campaign landing pages are heavier and slower than other retail pages

For online retailers, promotional landing pages represent one of the first steps in enticing visitors to purchase. On any given day, large retailers may run thousands of marketing campaigns, each with revenue expectations attached.

Unfortunately, although the landing pages associated with these promotional campaigns may be a visitor’s first experience with a site, they frequently perform more poorly than other pages. At SOASTA, we’ve found that landing pages for promotional campaigns load, on average, 30-60% more slowly than regular landing pages. Depending on the scale and volume of a business, slow page speed can cost tens of thousands of dollars for each minute of a campaign.

Webinar: How Walmart Manages And Optimizes Online PromotionsThere are a few common issues associated with the poor performance of these pages:

  • page bloat,
  • an overabundance of tracking tags and other third-party scripts,
  • no performance testing of newly created landing pages, and
  • simply the rush to get promotions out the door.

With fierce competition from Amazon, Walmart, and other giant retailers that are increasing their focus on online shoppers, optimizing key pages like promotional landing pages is essential to salvage potentially lost revenue. This post offers a glimpse of both the challenge, and a tested method for optimizing campaign landing pages, and salvaging lost revenue from under-performing campaigns.

Performance study: How do marketing landing pages compare to regular landing pages?

Using Ghostery and WebPagetest, we compared the page size and performance metrics of promotional (paid search) landing pages and non-promotional landing pages for ten of the top twenty-five online retailers (ranked by revenue). Ghostery helped us understand how many third-party tags were found on each page. WebPagetest gave us information about the speed and size of the page itself.

From a data science perspective, this is admittedly a small sample size; however, our goal in this research was simply to gauge the health of a random sampling of pages within a group of leading retailers, and to see what questions emerge from the data.

Five key findings

1. Promotional landing pages are heavier than dedicated landing pages

Customers doing product searches via Google or other search engines often click the first relevant link in the search results. If you’re using Google, the first link is frequently a paid search link. We found that the median page weight for paid search landing pages was 13.1% heavier than for normal landing pages. Similarly, we found that the median number of page resources was 12.6% greater for the paid search landing pages.

Note: The heaviest page we saw during the analysis weighed 5.12MB. (The culprit: Too many unoptimized images.)

Read > What I’ve learned from monitoring four years of web page bloat

2. More third-party tags correlated with slower start render times

The number of third-party tags – including advertising, tracking pixels, site analytics, social engagement, customer interactions, and comments – per page ranged from nine to fifty. While there were some exceptions, we found that for most pages, having more tags correlated with slower start render times.

In a joint research project we conducted with Google last year, start render time was found to be a strong predictor of bounce rates and conversions. Therefore it’s critical that site owners ensure their pages begin to render as early as possible. Limiting/optimizing third parties is a good strategy.

Read > Sampling sucks (or, why you need to collect, keep and use all your user data)

3. Start render times were higher overall for promotional landing pages

60% of the paid search pages registered slower start render times than their dedicated home page cousins. As the graph below illustrates, slower start render times frequently correlate with greater page weight.

In a recent study of 10 billion user experiences, we found that just a 100-millisecond delay in load time can hurt conversion rates by up to 7%. Since paid search is an asset to companies vying for top position in Google search, optimizing these pages is a necessity.

Read > Case study: How Lowe’s cut load times in half and drove online sales past $1B

4. Promotional campaign pages had up to 67% worse Speed Index scores

The Speed Index is a WebPagetest metric that measures how quickly – in milliseconds – a page’s content visually populates. A higher Speed Index score indicates slower visible page rendering time.

We found that eight out of the ten paid search landing pages ranked significantly worse (higher Speed Index score) than their non-promotional landing pages. Below are the sites with the slowest Speed Index scores. (Note that the Speed Index score is indicated by the yellow line in the graph.)

  • Site 1: Scored 8126 on paid search vs. 4819 for the dedicated home page. Paid search page start render time was 3.3 seconds slower.
  • Site 4: Scored 4512 on paid search vs. 2628 for the dedicated home page. Paid search start render time was 1.9 seconds slower.
  • Site 8: Scored 4252 on paid search vs. 3223 for the dedicated home page. Paid search start render time was 1 second slower.

Landing page optimization: page weight vs start render time

Read > Three things you can you do with 500 billion beacons worth of user data

5. Even with fewer tags, there’s still a need to optimize

Of the ten sites analyzed, Site 6 stood out with only 9 tags. This translated into start render times of 0.68 seconds for the paid search page, and 0.58 seconds for their home page. However, fewer tags did not translate to a higher Speed Index score, evidenced below. Paid search scored a 6368 while the dedicated home page had a score of 6258.

Read > 10 pro tips for managing the performance of your third-party scripts

Let’s visualize the problem to understand it better

Through SOASTA’s data science visualization capabilities, below we see a typical segmentation of regular site traffic versus marketing campaign traffic. The treemap shows the volume of traffic (size of each square) against its median load time, represented by this color key:

We see that the marketing campaign page (in this case, from an email marketing campaign) represents 6% of overall traffic, but with a median load time of 7 seconds, it takes 2.5 seconds longer to load than the site’s home page. In other words, the campaign landing page was about 55% slower than the home page.

Why is this a problem? These visitors were targeted: the email campaign enticed visitors in, but now they have to wade through slower pages just to get to the deal promised. Numerous studies prove that many customers will bounce after 2 or 3 seconds.

Next, we’ll examine that same campaign web performance by device type. Below, you can see median page load times for visitors on desktop are greater than 6.5 seconds, tablets load at 9+ seconds, and visitors on mobile devices need to wait at least 15 seconds for their promotional page to load.

Now compare the device performance for the above marketing campaign performance with the same devices on the regular site (below). The results: every device performs significantly better. This is common across most of the marketing campaigns we see.

Finally, let’s compare the marketing campaign’s speed across different geographies. Below is the campaign landing page segmented by geolocation. The United States takes up the majority of traffic. The promotional landing page had a median load time of 5.5 seconds.

Compare the above with results from the overall site traffic below. Although there are many more page groups, and more areas to maintain from an IT/Ops point of view, we clearly see that normal site pages simply perform better across all geographies and page groups. The promotional page is as slow as the slowest page group on the regular site (search results) at 5.5 seconds.

How does Amazon monitor and optimize promotional campaigns?

Amazon is a data-driven company. This permeates everything they do, including their marketing campaigns. Amazon runs thousands of marketing campaigns every day and optimizes nearly 75% of them. They use A/B testing – using all of their user-generated data – to rapidly iterate and optimize each campaign.

Bottom line: Amazon recognizes the need to test, monitor, measure, and optimize customer-facing digital promotions.

Read > Case study: How Deckers improved Time to Interactive by 33%

What you can do about your campaigns

Your campaigns are expected to generate revenue, therefore you should treat your promotional pages the same as any other important digital asset.

  • Companies should look at the entire marketing campaign development cycle: the people, processes, and tools.
  • Marketing and IT/Ops teams must work more closely together around builds and releases.
  • Performance testing should be mandatory for each release and should be considered a best practice before launching promotional pages.
  • Since new promotions, especially targeted campaigns (e.g. loyalty rewards programs and buyers) should be measured, finding a real user monitoring toolset for active alerting is key.
  • Finally, your team should adopt tools or products that allow you to work with all of your data (you can’t find complete answers with incomplete data sets) and that provide alerting, predictive analytics, and machine learning capabilities. Adopting this toolset will provide the context necessary to make faster decisions based on your own user intelligence.

With the right framework in place, your marketing promotions will earn more money and help you compete in this fast-paced world of ecommerce.

Try mPulse for 14 days of free real user measurement

Read the original blog entry...

More Stories By SOASTA Blog

The SOASTA platform enables digital business owners to gain unprecedented and continuous performance insights into their real user experience on mobile and web devices in real time and at scale.