Cross-selling and upselling is a popular tactic among online retailers in hopes of increasing average order value, items per sale and improving customer service with relevant suggestions. Amazon shared that cross-sells were responsible for 35% of its sales in 2006! According to the e-tailing group’s 8th Annual Merchant Survey Report (of 190 ecommerce executives), 55% of retailers will include cross-selling and upselling in their merchandising activities this year.

But cross-selling and upselling is one of the most difficult activities to do well and effectively measure, as evidenced in the e-tailing group’s findings:

Cross-sell/Upsell in Shopping Cart, Conversion Rates:

• Less than 1% conversion - 8% of retailers
• 1%-4% conversion - 16% of retailers
• 5%-10% conversion - 9% of retailers
• More than 10% conversion - 3% of retailers
• Don’t know conversion rates - 44% of retailers
• Don’t merchandise in shopping cart - 20% of retailers

Cross-sell/Upsell on Product Pages, Conversion Rates:

• Less than 1% conversion - 5% of retailers
• 1%-2% conversion - 15% of retailers
• 3%-4% conversion - 5% of retailers
• 5%-7% conversion - 6% of retailers
• 8%-10% conversion - 2% of retailers
• 11%-15% conversion - 1% of retailers
• More than 15% conversion - 2% of retailers
• Don’t know conversion rates - 50% of retailers
• Don’t merchandise on product pages - 14% of retailers

The only overwhelming statistic here is that most retailers have no clue how product associations convert. With 92% of retailers citing web analytics as the number one data source for merchandising decisions, it’s disturbing that many retailers are not measuring the outcome of these decisions.

Of course, measuring conversion rates for cross-sell/upsell can be ridiculously complicated, and depends on what kind of cross-sell/upsell solution you’re using. If you’ve built your solution in-house or your commerce platform came with cross-sell/upsell out of the box, you’ll need to figure out how the data will feed into your analytics tool. If you’re using a third party Software-as-a-Service like RichRelevance or Baynote, analytics might be provided for you, but it might not provide the depth and detail that you want.

Measuring the Right Thing?

For example, your merchandising tool might not break out conversion rate by shopping cart vs. product page. It may not be able to show you detail like product category cross-sell/upsell conversion, or tell you “conversion rate for cross-sells in price range $X-Y in relation to product price $A-B is xyz.”

Then there’s the question of “what does conversion rate mean?” Does it mean the product is viewed, added to cart, or a sale is made by a customer who is shown cross-sells on his/her visit? The merchandising tool we use on the Vancouver 2010 Olympic Store tells us that their tool lifts conversion by 140% and average order value by $14.94 (with A/B split testing). That doesn’t tell me if customers are buying more items per sale. I don’t know which suggested products are most successful to refine our merchandising strategy. I don’t know which products and categories have the highest conversion rate.

These problems and questions are common among online retailers, and while tracking these detailed events is possible (with complicated analytics mashups, for example) there’s often not enough IT resources or budget to make it happen.

Though you may not have access to all the data that would be helpful, at minimum, a global conversion rate is a start. I wonder how many retailers who “don’t know” their conversion rate just don’t know where to access the reports from the vendor.

Improving Cross-Sell/Upsell Conversion

If you do know at least your conversion rate for pages with cross-sells vs. pages without, you have a benchmark you can work on improving.

“You’ve read this far in this article. We think you’ll also love…” last year’s Get Elastic post Cross-Selling Tips for Online Retailers for a list of Dos and Don’ts, along with retailer examples.

Elastic Path also did a webinar on effective merchandising with Mike Svatek, formerly of Baynote. Mike offered this advice:

When do cross-sells work?

Cross-sells work well for considered purchases (high involvement rather than impulse - typically higher cost) provided they are lower cost accessories related to the product. They also work for smaller purchases with small accessories like Barbie and an outfit. You want to keep the cross-sells at half the price or less. When they are more than 1/2 the price of the item considered, the attach rate is low.

Products with natural bundling are also good, like cameras with lens, cleaner, memory cards and warranty.

When do cross-sells fail?

Don’t try to push higher priced items together with lower priced. People who buy a camera may buy a camera lens at the same time, but it’s unlikely someone adds a lens to cart and then all of a sudden wants to buy a camera. Same with sports tires - you wouldn’t try to upsell a Porsche.

Be careful that you don’t just look at correlation in your analytics data - but consider the primary and secondary intent. You may want to manually add constraints to your rules engine so you don’t goof your directional selling.

When do up-sells work?

Upselling (suggesting a similar item instead of the item being viewed) must have a small difference in dollar value or a small nominal percentage difference - 10-20% max. You need to show some incremental value for the increase in price.

When do up-sells fail?

When important attributes are different (red vs. blue dress) or when you show items that don’t have the features the customer is looking for. You can also fail by showing different brands. If a customer owns a Nikon he needs Nikon accessories, not Pentax or Canon.

You also need to consider any contractual agreements you have with suppliers and brands. For instance, you may not be allowed to show certain brands next to each other.

Effective merchandising often requires tweaking your tool with custom rules, rather than a “set it and forget it approach.” Make sure you fully understand your tool’s ability to set constraints, blacklist products and create custom associations. Also understand how to review any available analytics data your solution provider collects.

Last summer we did a collaborative post with fellow Vancouverites VKI Studios called Stop Google Analytics from Stealing Your Valuable Keyword Data. Google Analytics really isn’t “stealing”, rather “concealing” the actual search queries that trigger your paid search ads when you’re using broad match. It’s a “ye have not because ye ask not” situation.

“Ask and ye shall receive,” and by ask I mean set up a couple custom filters that will expose this data to you. I will be so bold to say that if you can not see exact keyword referrals you have no business using the broad match type! (<---And I rarely use exclamation points or blog the same topic twice!!!)

This trick has become the most important keyword research tool I use after a campaigns launch (I use a few methods of keyword research to set up Ad Groups including the Google Keyword Tool). Once the campaign is underway, I use the exact keyword referrals to discover negative keywords, uncover new Ad Group and product opportunities and to understand more about how people search. What's missing is transactional data for each keyword, unfortunately.

I decided to screencast the set up process for a few reasons.

1) To share this tip again with our new readers (we've almost doubled in readership last summer) and remind those who have put off adding the filter to set it up ASAP.
2) To show you how quick and easy this is and provide you with a resource (printable PDF) that will give you the confidence that you can set this filter up yourself!
3) To show you how to find your data in Google Analytics by AdGroup, so you can add apply the appropriate negative keywords at the Ad Group level.

If you bear with me to the end, I share some of the crazy matches we've been getting for the Vancouver 2010 Olympic store’s broad matched keywords. You’ll see why I value this information so much!

Can’t see video? View it here.

Companion Resources

Download “Cheat Sheet” Instructions (PDF)

Cut and Paste Values:

As with almost all multi-part filters, sequence is critical and must be ordered accordingly using the “Assign Filter Order” page for the profile.

First Filter:

Field A -> Extract A: Referral: (\?|&)(q|p|query)=([^&]*)
Field B -> Extract B: Campaign Medium: (cpc|ppc)
Output To -> Constructor: Custom Field 1: $A3

Second Filter:

Field A -> Extract A: Custom Field 1: (.*)
Field B -> Extract B: Campaign Term: (.*)
Output To -> Constructor: Campaign Term: $B1 ($A1)

I’m noticing more and more retailers on Twitter these days, and I expect this trend to continue through 2009. So I thought I would share some tips and tools for ecommerce Twitter marketers.

Sync Your Blog or RSS Feed to Twitter

Twitterfeed
&

Twitterfeed and Hootsuite (formerly Brightkit) allow you to very easily Tweet anything you publish by RSS including contests, new products, product-back in stock, news, deals-of-the-day and new blog posts.

(No that’s not a typo. It says Anti Monkey Butt Powder).

For example, Musician’s Friend could tweet its Stupid Deal of the Day by Twitter every day automatically by signing up once with Twitterfeed or Hootsuite. Unfortunately, twitter.com/musiciansfriend has already been brand-jacked by someone, so they’ll have to get creative with its Twitter Name. (As of today, it could scoop up twitter.com/stupiddeal or /stupiddod (deal of day) — hurry Musician’s Friend!)

This also brings up another tip - register a Twitter account for your brands today even if you’re not using Twitter to avoid such a problem down the road.

There’s no limit to how many RSS feeds you can tweet. You can even tweet others’ feeds if you want (for example, if you sell a certain city’s sports team merchandise, you could sync with updates from the team’s home page or game recaps from the news.

There 2 major advantages of Hootsuite (winner of this year’s Shorty Award for best App) over Twitterfeed:

• You can manage multiple Twitter accounts through one interface with Hootsuite and schedule tweets in advance. So you can have dedicated deal-of-day account, new products and customer service accounts and manage them without logging in and logging out. Plus, you can stagger your deals of day in advance - set it and forget it.
• If you’re tweeting out links (your calls-to-action) you want to measure if people care. Hootsuite will show you stats on how many people click on links you tweet:

Hootsuite shortens your links using its own ow.ly service, so you will only get stats for ow.ly links, but they will track if your tweet gets re-tweeted by other Twitter users so you can see the full impact of your seeded (or Tweeded) links, even if they’re clicked through social media aggregators like Friend Feed.

Twitter Analytics

Default Google Analytics Report

If you’re not using Hootsuite, you can still track how often your Twitter links get clicked in Google Analytics (or the analytics tool of your choice). But, it’s complicated. Here’s why:

1. Since most URLs get shortened to ow.ly, is.gd, bit.ly or one of the many shortening services, all traffic that came from Twitter might not show up under the Twitter referral source. And, sometimes your Twitter links end up in FriendFeed, LinkedIn, Facebook or MyBlogLog aggregators:

2. You can see which Twitter profiles referred the most traffic to you by clicking “Twitter” in your Referral Source list to see detail. It should look something like this:

The problem is “Home” referrals. They include any user who clicks on a link from their Twitter feed rather than another user’s profile. Not perfect tracking, but interesting nonetheless.

3. Google Analytics doesn’t show you which individual tweets are most popular. (I chatted with Hootsuite and there may be a way down the road to import Hootsuite stats into your analytics program, but not yet). There IS a workaround if you must track everything in your analytics package (sales conversion and ROI, for example) which is to tag your links with campaign parameters before shortening and sending them out.

Tracking With Campaign Parameters

Create a URL like:

http://www.yourURL.com/?utm_campaign=dealofday& utm_source=twitter&utm_medium=social-media

And paste it into a URL shortening tool like this (Hootsuite will create an ow.ly link for you):

The short link will redirect to a copy of the landing page, registering as an entry page that will appear in your analytics reports (in Google this appears under Site Traffic / Referral Source).

Hat tip to the Epik One Analytics Blog on this one.

Unfortunately this will only track links you generate, not organic tweets about you by other Twitter users. And if you use automatic RSS tweeting, this won’t work because the link will shorten without parameters.

Reputation Management

Tweetbeep

I hope you all are using at least the free Google Alerts service to notify you when your name, brands, important products and other trademarks are being mentioned online, if not a more sophisticated tool like Trackur for your general reputation management.

But Google Alerts isn’t good at reporting what the Twitterverse is saying about you. I used to enjoy a tool called Tweetbeep which would send alerts to my inbox for whatever keywords I wanted to track. Unfortunately it’s been down for a long time now. Fingers crossed that it rises like the phoenix or something emerges to take its place.

Alternatively you can subscribe via RSS to individual searches using Twitter’s search engine.

What I preferred about Tweetbeep was it would report all your tracked keywords in one daily beep, rather than signing up to each search’s feed.

Coupon Marketing

TwtQpon & Cheap Tweet
&

TwtQpon is a free coupon generator specifically for Twitter. You can create a coupon with a description (max 140 characters, of course), coupon code, URL (landing page), photo and optional expiration date. Here’s an example:

You can also promote your deals through CheapTweet which is like Digg for tweeted online deals.

Check out the guidelines for how to get your deals included.

Competitor Comparison

Tweetvolume

Curious how popular your brand is compared to your closest competition? You can quickly create a bar chart at Tweetvolume:

Or, how popular are various brands or keywords of products you carry on your own site:

Keep in mind that retailers that tweet themselves will naturally be higher, as will retailers with affiliates who use Twitter as a channel to broadcast deals and coupon codes.

Bonus: No Turn on Red has built a Twitter-aggregator for some of the largest retail presences on Twitter that updates itself every 30 minutes so we can get a *bird’s* eye view of how they are communicating (deals, links, customer service, random thoughts etc) for inspiration.

Just announced today, Google Analytics is rolling out new features to its free tool in Public Beta, meaning you can expect to gain access to these new goodies sometime in the next couple weeks.

Click here if you can’t view this video.

As Avinash Kaushik says: Analyzing data in aggregate is a crime against humanity. …
If you want to find actionable insights you need to segment your web analytics data. You need to separate out the various Sources, Behavior and Outcomes.

When I first watched the YouTube video tutorial I thought “man, this looks complicated.”

Avinash assures us it’s not, and reassured us his 7 year old had no problem creating 2 segments in minutes (talk about precocious!)

In addition to the default segments like New, Returning, Paid and Non-Paid Visitors; and Search, Direct and Referral Traffic, you can now create custom segments. A great place to start is segmenting branded search vs. non-branded search terms for organic and paid traffic. Why?

Research by Compete and Google suggests that 70% of purchases made from non-branded keyword referrals occur in later online sessions. Of that 70%, 16% of the purchases happen only after the searcher searches again with different keywords. Non-branded searchers are more likely to be Howsers (know in general what they want but not which specific product, may need more research and education) than Hunters (know exactly what they want and are close to conversion). And if you’re working on strategies to optimize your conversion for customers who are beginning or mid-way through the decision process - you’ll want to know if your conversion rates for non-branded keywords is improving after you implement your testing/strategy.

You can also create segments for non-branded category search. For example, Garmin could segment non-branded motorcylce GPS keywords by creating a segment that must include “motorcycle” but exclude “Garmin,” “zumo” “Quest” “street pilot” and “streetpilot.”

Image credit: Avinash Kaushik

Someone who uses the branded term is likely ready to buy a Garmin product. A good objective for Garmin might be to increase non-branded conversion through strong Garmin value propositions, persuasive copywriting, advanced customer support tools or other multi-media selling aids. The saved segment can produce reports after the testing period, and compare to metrics before the optimization tactics were employed.

Then you can check out which pages these visitors view and what they buy.

This is just one example of what’s possible. Please bookmark and read Avinash’s thorough post for more segmentation ideas.

According to Group Manager of Google Analytics Brett Crosby, segmentation can help you perform better through an economic downturn and “go with what works, find new ways to drive revenue, find segments working for them and invest there to increase their conversions.”

Hat tip for this quote, Marketing Pilgrim.

Expect more segmentation tips and screenshots once I get access to the tools in our Analytics account…if you haven’t subscribed to Get Elastic - sign up today, won’t you? (It’s free, or you can subscribe by email at the top of the page).

Motion Charts is a new visual tool from Google Analytics that launched today and will be rolled out across Google Analytics accounts over the next couple of weeks. Until you get access, here’s a preview:

May help in your decision which keywords to keep and to kill in pay-per-click without pouring over pages of numbers. Or show you which terms to focus on for SEO.

Excellent.

But before you pull the trigger on a PPC keyword, you might want to give this a read.