NEW YORK, NY–(Marketwire – Jan 31, 2012) – LiveIntent, the platform for buying and selling email ad inventory exactly like display, today announced it has integrated its services with advertising technology from Convertro, the leader in providing advertisers enhanced transparency into the effectiveness of their online marketing channels and optimization. This integration provides advertisers with a better understanding of the value of LiveIntent’s inventory within their overall digital media footprint.

Previous constraints in the email space only made it possible to track last click conversions or direct consumer behaviors back to email, but Convertro’s Multi-Attribution service allows advertisers access into every marketing touch that led to a conversion without limits on the number of touch points, cookie windows and other constraints. By partnering with Convertro, LiveIntent can now attribute an action back to an email ad placement, even if the consumer decides to shop around before converting into purchase.

“This integration was driven by our goal to provide advertisers with more knowledge and transparency into the impact of our email advertising offerings,” said Jason Oates, President LiveIntent. “With Convertro’s unique attribution service, we can now prove to advertisers how effective their email advertising campaigns are by identifying how they impact the top (awareness) and bottom (conversion) of the sales funnel. This is important to brand and direct advertisers.”

“By tapping into Convertro’s technology, LiveIntent will now be able to prove to advertisers accurate contribution of email marketing efforts to best optimize their revenue from email ads,” said Jeff Zwelling, CEO and Co-Founder, Convertro. “Convertro strives to provide advertisers with transparency and in turn, give our customers the attribution they deserve.”

This integration is one of many strategic initiatives from LiveIntent, including new product offerings to its portfolio and personnel additions to accommodate the company’s rapid growth.

About LiveIntent

LiveIntent is the only company with a real-time display ad platform for email. LiveIntent works with advertisers, publishers and agencies to provide ad optimization, DSP / trading desk and exchanges capabilities. The company is venture-backed by Shasta Ventures, Battery Ventures, First Round Capital, Lerer Ventures and Grape Arbor VC. LiveIntent was founded in 2009 and is headquartered in NYC. For more information and a demonstration of the company’s technology in your own inbox, visit www.liveintent.com.

“Convertro’s advanced user tracking methods combined with ex post facto analytics capabilities generates the richest possible data set with infinite flexibility. For many marketers, television represents the most significant portion of advertising spend. By leveraging Convertro’s technology and applying advanced statistical modeling, we will be able to accurately predict the contribution of each marketing channel and advise clients on how to best maximize their revenue and profit, said Avedissian. “My goal is to empower clients with marketing intelligence and optimization that to date has been available to only a select few companies that have spent millions of dollars developing in-house solutions.”

Prior to Convertro, Avedissian was Chief Revenue Officer (CRO) of MyLife, where his media mix planning and optimization skills more than doubled revenues in a short period of time while lowering the cost per acquisition. Prior to MyLife.com, he led eHarmony through a period of exponential growth using similar methods.

“Avedissians’ contributions and achievements at MyLife and eHarmony helped scale both companies into the highly profitable, well known brands that they are today by leveraging the power of television combined with quantitative analytics. He has proven success in managing and optimizing direct response media campaigns,” said Jeff Zwelling, Chief Executive Officer of Convertro. “By combining his skills with our leading media attribution technology, we will provide clients with a holistic attribution approach to calculating ROI and optimization for all marketing channels, making this available for the first time as a SaaS model. Convertro gives our clients access to a best-of-breed marketing intelligence and optimization platform with no upfront cost or tax on internal resources.”

About Convertro
Convertro provides advertisers enhanced transparency into the effectiveness of their marketing mix by tracking media channels and their resulting conversions. The Convertro platform’s ex post facto design empowers marketers with infinite flexibility to add or tweak attribution models and apply optimizations, thereby providing them a better understanding of the actions that are actually driving sales and the data to make informed marketing decisions and beyond. Convertro is headquartered in Santa Monica, California with an office in Ra’anana, Israel and is backed by Bessemer Venture Partners.

Convertro is excited to wrap up a strong year that brought lots of excitement to the market for mature analytics tools. In spite of doing practically no outbound marketing, we’ve gained a ton of new customers, with many large clients inbounding to request information based mainly on word of mouth.

In online marketing, some of the most common metrics employed are a performance measurement looking back over a thirty day cookie window and crediting the last click. Although marketers are moving away from last click measurement and attribution techniques, the thirty day cookie window continues to be the most predominant timeframe used in analytics tools. Here we consider whether this is a valid model and whether it skews the metrics used for decisionmaking.

Let us take Convertro data across twenty-plus clients to examine this. The conversion data across these clients runs into millions of conversions, however, for purposes of simplification, and also to see if data skews differently based on purchase amount (which might influence consideration/sales funnel length)we sliced the data into strata based on the conversion value.

Although it appears that conversions peak a day after initial visit, followed by a tapering off, it’s important to note that the first thirty days account for only about 72% of conversions; 28% happen thirty or more days after the initial visit. This occurs across all customer pricepoints[[stratas]], although, as expected, it is more pronounced for businesses with long sales cycles. The clear conclusion here is that by limiting the cookie window to thirty days, companies can wind up misattributing 28% of sales, resulting in suboptimal allocation of your marketing dollars.When we examine this same data for number of sources preceding conversion, we see something similar. There are very few purchases which occur after a single visit to the site. More often, there are several visits before a purchase, even if the purchase happened within the first thirty days of initial visit.

 

A slightly different way of looking at the same data is to combine the number of events (left axis) and number of sources (right axis) with number of days to conversion (x axis). The long tail of conversions emerges clearly in the graph below.

 

 

So, then why has analytics relied on a thirty day cookie window for this long?  Let me compare this with an interesting story (from a redditpost) about how certain beliefs and practices become mainstream and how they are never challenged:

At Convertro, we constantly challenge the status quo of how things are done by introducing a no-limits approach to data collection and analysis, empowering marketers with data on every single marketing source no matter how long ago it took place. We make this data readily available via the dashboard and multiple advanced reports, giving marketers the ability to analyze quickly and take action.  For more information on our products and solutions, please visit us at www.convertro.com.

Success stories abound about how Groupon (and its ilk: LivingSocial, Google Offers, Facebook deals, etc.) is helping local businesses acquire new customers. Running a Groupon deal delivers a huge inflow of new customers and by the looks of it, it seems like a great way to grow business. The deals that Groupon offers to customers are hard to resist, and businesses often see hundreds or thousands of sales in a matter of days. Groupon offers, on average, are at a 50% discount on merchandise prices, so there is no question that they are great for customers. But, what about the businesses? How do you know if Groupon is right for your business? And how much should you pay Groupon if you can negotiate with them?

With Groupon, you typically have to offer merchandise/services at a 50% discount to be featured as a Groupon deal. Groupon then takes a 50% cut of the discounted fee charged. That leaves the business, for purchases made with a Groupon, with a maximum of 25% of the revenue they typically get (excluding, of course, any spend in excess of the certificate value). Finally, a Groupon potentially results in spikes in traffic to other paid marketing sources (mainly PPC/Affiliates/CSE) when potential customers use search as a navigational aide after receiving a Groupon. As such, the real cost of acquisition using Groupon for a business is often more than what it appears to be at first glance. So, for example, a $50 Groupon good for $100 in merchandise, assuming margins of 20% costs the company $55 in acquisition cost. This is after we account for the fact that the customer might not have converted without the Groupon and using loss on COGS of $80 as the acquisition cost.

Is the high traffic and awareness that a Groupon generates for a business enough to warrant the high customer acquisition cost?

Let’s delve deeper using generalized Convertro client data. The general business case for running a Groupon campaign is to acquire new visitors and new customers; we see in the following chart depicting new vs. return visitors to the site (after being exposed to the brand by being offered a Groupon) that visitors tend to be predominantly new. On average, for the time period we considered, 93% of total visitors coming from Groupon links to our clients’ sites had not visited previously.

But, because you’re taking such a big hit on revenue on the first purchase, even under the most optimistic scenarios, the true value of Groupon lies in its ability to drive repeat purchases and/or order totals in excess of the Groupon value. Upon drilling into the data, we see that only about 33% of buyers that were exposed to Groupon before their purchase went on to make a repeat purchase, and most of the repeat buyers had only 2 purchases (disclaimer: for this exercise we only looked at a three month period from when the Groupon went live).

Normalizing this number to 100 buyers and using varying margins and a 33% rate of repeat purchases, we see that when margins are high, Groupon works very well, but not so much when they are low.

Although this is interesting in itself, this type of general marketing analysis should be done before running any marketing campaign. However, there is a deeper analysis that can and should be run in addition which is far more insightful, and tells you the true cost of Groupon.

Note:  The numbers cited so far are using Convertro’s multi-attribution solution. When you only take last click attribution into account, your analytics solution will report significantly lower sales attributable to Groupon although the number of Groupon coupons redeemed will be higher resulting in mismatch and misrepresentation of actual impact.

As we alluded to earlier, the real cost of Groupon is significantly higher since a Groupon run results in significant spike across all sources for visitors who use these for navigational purposes and results in the following additional costs:

• PPC cost when potential customers use search engines as a navigational aide – which is an immediate cost irrespective of whether it results in a conversion or not.
• CSE fees and commissions from customers who arrive at the site from from CSEs and/or affiliate sites after search
• Unless the retailer is careful/prevents it, some of the customers can combine the Groupon with other offers on coupon websites and/or purchase the Groupon multiple times causing additional low- or no- margin sales for the merchant.

We can use Convertro’s multi-attribution data to effectively factor in all of these into channel analysis and evaluation.

The following chart illustrates the source position – specific position in the sequence of sources that the visitor saw before converting. As you can see Groupon is the only source in 15% of the cases.

In all other cases, there were marketing sources that assisted with the conversion, which leads us to the next question: with what other sources does Groupon most frequently overlap? From the following chart, we see that search engines, google affiliate network and email, all of which result in further costs/discounts, combined account for overlaps in over 60% of purchases.

Let’s consider the case of search engines with which Groupon has about 40% overlap. In the example we used earlier, of 100 visitors, only three converted, but about 40 visitors to the site used search as a navigational aide. Not all of these visitors resulted in PPC costs – some of them used organic search – but nonetheless they can result in another $1 of cost per acquisition (not all of this cost is wasteful as it has brand engagement and awareness lift that can result in subsequent conversion). In comparison to the deep discounts on Groupon, it might appear to be negligible, but it increases the cost of acquisition and combined with the other costs from affiliates/CSE’s, it could be a significant amount to warrant a thorough analysis. Especially, in the case of PPC, it is easy to run through monthly or quarterly budgets within days of running a Groupon, which means that valuable marketing opportunities are lost subsequently even on very targeted PPC matches. On the other hand in the case of Affiliates/CSE, double payments can result in further erosion/loss, that is difficult to compensate just by a minor increase in sales from increase in awareness.

We can start getting an accurate picture of the true cost of acquisition using Groupon only after you add in all of these costs and account for all future purchases from Groupon-browsers at the moment. Only at that point can you analyze if Groupon works for your business. If running a Groupon is causing additional expense/traffic spike from other channels, you have a few options:

• Renegotiate payment options with Groupon to see if they can reduce their commission
• Better manage Affiliate/CSE payouts using selective pixel firing to make sure that conversion pixels do not fire on Groupon sales
• Put budget caps on PPC
• Put frequency caps/pause display campaigns
• Incorporate terms on the Groupon so they cannot be combined with other offers
• Make sure that your visitor tracking is robust so you can prevent users from using the Groupon multiple times

Convertro gives you not only the ability to get better insight into your campaigns but also gives you a recommended course of action, and once you decide how you want to go about reducing your costs, enables you to enforce these decisions by providing you with Tag Container and Selective Pixel Firing features. The above analysis was done using generalized Convertro data, and makes use of only a tiny bit of the multi-attribution, user conversion tracking and other features that Convertro provides. For more information, visit Convertro.

If your current analytics solution doesn’t provide you with these capabilities, maybe it’s time to reconsider your options.

By doing so, we’ve been able to facilitate a shift from faith-based marketing to a true ROI-driven approach based on high quality data detailing the exact path to conversion.The quality and granularity of data we collect at Convertro affords us an opportunity to derive numerous other corollaries which we haven’t had a chance to share. This post focuses specifically on the fact that some customers tend to buy over and over again – and, more importantly from a marketer’s perspective, we can often distinguish between those marketing sources that drive higher repeat orders from those that tend to drive a one-and-done sales event.

If you could use historical data to accurately predict which marketing sources are more likely to attract higher value customers, then wouldn’t you want to allocate more revenue to those higher-yielding sources? In fact, we often see that some marketing sources are better at driving new customers who then go on to become repeat customers as illustrated by the following graph.

Rather than giving 100% credit for an event to the last known paid source, Convertro is able to attribute some of the revenue from subsequent marketing events back to the original acquisition event that first brought the customer to your website.  In contrast, the last known paid source approach will give credit for the subsequent orders not to original marketing source but invariably to either direct or navigational search terms, such as the company’s brand.  The impact of this flawed attribution is that the marketing sources that are successful at bringing you new customers – customers that come back and become repeat buyers – don’t get credit for subsequent orders but only the first orders.  And for many e-commerce categories, the first order isn’t always profitable.

For, example, in the case where there are three sale events of $100 each, the first source (Google in this example) that introduces the brand to a customer and serves to add a new customer who then goes on to become a repeat customer gets an additional $83 attributed to it. Correspondingly, the revenue attributed to subsequent sources which act as navigational/directional sources tend to get partial credit.

Although, some marketers use email address as a common key to tie multiple orders together and are able to credit some portion of the second order back to the first order’s marketing source, this is a less than ideal solution because it is manual and fails to easily provide real-time attribution of repeat orders back to the original marketing sources.

At Convertro, we offer a standard report called the New vs. Repeat Buyers report.  Marketers rely on it as a means to gain visibility into which marketing sources are driving repeat orders as opposed to those that are driving single orders.

In addition, when we view a click trail of an actual user in Convertro’s dashboards, we’re able to see each and every sale that the user engaged in, both validating the underlying data of the attribution model and creating an audit trail that can be used both for validation of the attribution model as well as for modifying attribution models on the fly.businesses that intend to grow their sales without increasing their marketing budgets.

The ability to analyze and derive such insights from data is critical for businesses that intend to grow their sales without increasing their marketing budgets. A Harvard business review study comes up with the following findings about repeat customers.

• Repeat customers spend 33% more than new customers
• Referrals among repeat customers are 107% higher than one-and-done customers
• New customers in effect have six times higher selling costs than repeat customers

These numbers are far too compelling for marketers to ignore. Data is only as good as the ability to identify users uniquely over long periods of time. Studies show that at least one-third of all cookies are deleted every 30 days. Because this key piece of tracking data is lost, the analytics software “forgets” the user it has already seen – resulting in lack of insight into the actual lifetime value of a marketing source leading to mis-allocation of budgets. Make sure that your analytics solution has robust tracking with a minimum of the following features.

• Transparent and easy to understand attribution model.
• Gives you high quality and granular data allowing you to analyze and derive powerful insights such as above.
• Robust tracking that can accurately identify all sources without limiting/cutting out number of sources tracked arbitrarily.

How much do you know about your affiliates?  Do you know who they are and what they are up to?

Convertro has dozens of clients using all the major affiliate networks, giving us extensive experience with affiliate taxonomy and the relative reputations of affiliates.

Most online publishers join affiliate programs and manage them in aggregate.  Publishers evaluate their affiliate programs’ overall conversion rates but are often not concerned with how individual affiliates perform. From their perspective, they mitigate risk by paying out on a CPA or revenue share basis; however, this opens up the door for affiliates to use suspect methods that take advantage of publishers that have well known brands!

Having looked closely at traffic data from specific affiliates within many different affiliate networks, we’ve uncovered numerous suspect methods of driving traffic. One egregious example: Do you think Zappos.com would approve of an affiliate that is renting ISP 404 error pages such as Cox.com to redirect users who were clearly looking for Zappos.com but mistyped the URL or brand?

Click here to see the actual live search result

Here you can see the same thing happening with Overstock.com:

Click here to see the actual live search result

The following statement from the Commission Junction Terms of Service for Zappos.com & Overstock.com clearly prohibits trademark search arbitrage:

Zappos.com Search Policy: We accept PPC Search publishers, but please do NOT conduct search bidding on our trademarked terms (like Zappos, Zappos Shoes, Zappos.com) or variants thereof (Zapppos, zapos, etc).

In addition, publishers are NOT allowed to have Zappos (or any variation of Zappos) anywhere in their main display URL. Please review Zappos’ Terms and Conditions and program terms for an in-depth list of protected and non-complete SEM keywords.

Violators risk losing all unpaid commissions and termination of their Zappos Development relationship. This is strictly enforced 24 hrs a day

Overstock.com Search Policy: Publisher shall not bid on any word, keyword or term in Pay-Per-Click Search Engines (PPCSEs) that only contains Company’s registered or unregistered trademarks or any word, keyword or term that is likely to cause confusion regarding Publisher’s affiliation with Company (“Overstock Keywords”). Publisher shall not bid on any word, keyword or term that only contains ANY VARIATION of the Company’s registered or unregistered trademarks. Examples of Overstock Keywords include, but are not limited to the following: overstock.com; overstock; overstocks, overstocking, over stock, www.overstock, www.overstock.com, and overstocked.

The publisher above is clearly in violation of these rules. They include their affiliate link in the banner as well as in the link redirecting the customer to Zappos (“Did you mean Zappos?”).  When the user clicks on the link or banner, the affiliate cookies the user and then gets paid on the conversion that ultimately occurs.  This method of hijacking the brand is very effective in driving affiliate revenues with sites that have high brand recognition.  (Of course, if you don’t have a prominent brand with heavy search volume, this type of fraud method is less effective.)

This is easy to miss.  One reason that managers don’t “get this deep” is that they think these kinds of affiliates are true contributors, since in reports they show high conversion rates and large numbers of successful orders.  In reality, if you’re seeing affiliates like this, with these kinds of conversion rates, it’s likely because your customers are trying to navigate to your site to buy directly from you, but are getting intercepted by these “brand squatting” pages. The fact is that these brand squatters are just getting in the way, and the only reason they’re getting so much credit is that the links they provide to your site are the shortest path your customer can take to your site. Many of these visits end up converting, and this makes the affiliate in question look productive.*  But in reality, you would get these sales anyway – the customers’ intent in these cases is clear. With the affiliate in the way, however, you end up with 10% less gross revenue and anywhere between 20-80% less margin!

If you considered the conversion rate of the above affiliate in a vacuum, the affiliate would appear to be an excellent partner; but, as we’ve just shown, the reality is that conversion rate doesn’t tell the whole truth.  By implementing Convertro’s advanced tracking, you can eliminate this risk of brand arbitrage by affiliates.

Take a moment to ask yourself the following questions about your affiliates.

Do you know:

–Which affiliates are bidding on your branded terms?

–Which affiliates are accretive and which are cannibalistic?

–Which affiliates are violating your terms of service?

–Which affiliates are hijacking 404 error pages for your brand?

–Which affiliates are using Adware?

–Which affiliates are using Toolbars?

If you have an affiliate program and you can’t answer “yes” to all the questions above, you need Convertro’s affiliate monitoring solution.

*The fact is that conversion rate just isn’t the right way to measure affiliates — rather, when it comes to affiliates, you should be looking at incremental contribution.

It’s October again, and that means it’s time for retailers country-wide to fight it out to acquire the millions of potential customers looking for the scariest, sexiest, and creepiest Halloween costumes around. But while the prospect of grown men wearing Sesame Street costumes may be terrifying, we’re here today to present some equally (if not more) frightening facts about mistakes you might be inadvertently making in managing your online marketing campaign. Read on, if you dare.

Odds are that you, or a marketer at a company you work with, uses Google Analytics to help manage online marketing campaigns. That’s because it’s a free tool that appears to track marketing, and so has become the default site metrics measurement tool for many online marketers. The truth is, relying solely on Google Analytics “visits” as a metric of marketing performance can be misleading at best, and a dangerous business practice at worst. If you take a close look at how Google counts visitors across Google Analytics and AdWords, you’ll realize that oddities in some of their counting mechanisms, while subtle, have the potential to drastically alter where you, as a consumer of this data, think it makes sense to spend your advertising dollars.

At Convertro, our goal is to provide marketers exact visibility into their customers’ marketing interactions prior to purchase. When they start to dig into our data, prospects invariably ask why our visits count doesn’t match what they see in Google Analytics (of course, they expect GA to be the gold standard). The development side of our company had been relatively skeptical of Google Analytics’ numbers, but the client services side wasn’t convinced, so we did some research to find out who was right and who was dead wrong.

The most glaring piece of data we discovered involves the fact that Google Analytics considers any number of visits to your page by the same user within a thirty minute window to be a single visit, rather than multiple instances of a user landing on your page.

Visits represent the number of individual sessions initiated by all the visitors to your site. If a user is inactive on your site for 30 minutes or more, any future activity will be attributed to a new session. Users that leave your site and return within 30 minutes will be counted as part of the original session. (http://www.google.com/support/analytics/bin/answer.py?hl=en&answer=57164)

Why is this an important fact to note? Because if you under count the number of times users land on your page, you’re also under counting the number of ways they get to your page, and thus you’re missing information on what drove those users to your site in the first place. We speculate that GA is combining multiple visits to reduce the number of inserts to their database, not because they are trying to skew the numbers to their benefit, but the result is the same.

Consider a user looking to buy a sexy Halloween costume for this year’s festivities (pretend, for the moment, that you are a purveyor of such goods). She searches on Google for “halloween costumes” and clicks on your enticing PPC ad, bringing her to your home page, CrazyCostumeCreations.com. After looking around a bit and not seeing anything all that sexy, she decides to try another web search, this time using Bing. A PPC ad on Bing with a creative actually mentioning your large selection of sexy costumes delivers her back to your page, where she finds the perfect Big Bird tube dress and checks out about half an hour after starting her search. One more customer satisfied!

In the Google Analytics world, her purchase was driven by one referral — because they both happened within the same half hour, her two visits to your site are considered just one. But in the real world it’s clear that both Google and Bing PPC (and, importantly, the two different creatives) are responsible for her purchase. The PPC click from Bing, however, won’t show up in your Google Analytics numbers, making your spend on Bing seem like a poor investment. In contrast, the Google PPC click ends up seeming more effective than it truly was. Scary, isn’t it?

Now let’s pretend our user is a Google loyalist and instead her whole search process takes place on Google. After her initial “halloween costumes” search, ad click, and dissatisfaction, she decides to give Google another shot and searches for “sexy halloween costumes.” This time she sees your more targeted creative on Google, clicks, and the rest is history. How are your GA numbers leading you astray this time? First of all, GA thinks this is one visit when it’s really two, so right off the bat you’ll be over-estimating your conversion rate. You’ll also think all the revenue from the Big Bird dress is the result of one visit instead of two, so you’ll over-estimate your revenue per visit. And, to add to the terror, this purchase actually cost you twice as much as the visits count in GA would lead you to believe.

So why not integrate your Google Analytics and AdWords accounts, and then use the “clicks” metric in GA that’s available if you do this? This metric is supposed to reflect the clicks that come to your site regardless of when in a user’s session they occurred. But now we must deal with the issue of over-counting clicks, because invalid and spammy clicks are reported as clicks in GA but aren’t counted in AdWords. Now, why should this strike fear into your heart?

Say that our friendly user lands on your site after her first search, hits back to do her second Google search but then suddenly remembers that she did see a link near the bottom of your page that she could have sworn started with the letters S-E-X. She hits forward to check it out again and makes her purchase. In this case GA will count one visit (remember the half hour rule) and AdWords will count only one paid click thanks to some magic deduping logic (read about GCLID’s [here]). But GA will count two clicks because Google doesn’t apply this filtering logic in Analytics. Unfortunately for you, this over counting of what seem like paid clicks will artificially deflate your conversion rate and your revenue per visit (you think you have more clicks than you do). Not to mention, when you see those extra clicks in GA coming from Google PPC, you’ll run screaming to your accountant and demand to know how your cost per visit got so high.

The simple fact is that Google Analytics isn’t designed as a platform for managing marketing. It’s a tool meant to enable analysis of aggregate data about your website, and to give insight into common referrals and the paths users traverse on your site. Google AdWords, on the other hand, is an excellent marketing platform for managing marketing — as long as that marketing only uses Google products and properties. If data-driven marketers want to help their clients make a killing this Halloween season, they need to do their job with the best resources available. That means understanding exactly what Google Analytics is capable of and using it appropriately, while leveraging other tools that are designed specifically for accurately tracking marketing. So this October, whether the prospect of sexy cartoon character costumes sends you running for the hills or not, don’t let your GA numbers scare you into misallocating marketing spend!

written with Allyson Taylor – Convertro Sales Engineer

Odds are that you, or a marketer at a company you work with, uses Google Analytics to help manage online marketing campaigns. That’s because it’s a free tool that appears to track marketing, and so has become the default site metrics measurement tool for many online marketers. The truth is, relying solely on Google Analytics “visits” as a metric of marketing performance can be misleading at best, and a dangerous business practice at worst. If you take a close look at how Google counts visitors across Google Analytics and AdWords, you’ll realize that oddities in some of their counting mechanisms, while subtle, have the potential to drastically alter where you, as a consumer of this data, think it makes sense to spend your advertising dollars.

At Convertro, our goal is to provide marketers exact visibility into their customers’ marketing interactions prior to purchase. When they start to dig into our data, prospects invariably ask why our visits count doesn’t match what they see in Google Analytics (of course, they expect GA to be the gold standard). The development side of our company had been relatively skeptical of Google Analytics’ numbers, but the client services side wasn’t convinced, so we did some research to find out who was right and who was dead wrong.

The most glaring piece of data we discovered involves the fact that Google Analytics considers any number of visits to your page by the same user within a thirty minute window to be a single visit, rather than multiple instances of a user landing on your page.

Visits represent the number of individual sessions initiated by all the visitors to your site. If a user is inactive on your site for 30 minutes or more, any future activity will be attributed to a new session. Users that leave your site and return within 30 minutes will be counted as part of the original session. (http://www.google.com/support/analytics/bin/answer.py?hl=en&answer=57164)

Why is this an important fact to note? Because if you under count the number of times users land on your page, you’re also under counting the number of ways they get to your page, and thus you’re missing information on what drove those users to your site in the first place.
We speculate that GA is combining multiple visits to reduce the number of inserts to their database, not because they are trying to skew the numbers to their benefit, but the result is the same.

Consider a user looking to buy a sexy Halloween costume for this year’s festivities (pretend, for the moment, that you are a purveyor of such goods). She searches on Google for “halloween costumes” and clicks on your enticing PPC ad, bringing her to your home page, CrazyCostumeCreations.com. After looking around a bit and not seeing anything all that sexy, she decides to try another web search, this time using Bing. A PPC ad on Bing with a creative actually mentioning your large selection of sexy costumes delivers her back to your page, where she finds the perfect Big Bird tube dress and checks out about half an hour after starting her search. One more customer satisfied!

In the Google Analytics world, her purchase was driven by one referral — because they both happened within the same half hour, her two visits to your site are considered just one. But in the real world it’s clear that both Google and Bing PPC (and, importantly, the two different creatives) are responsible for her purchase. The PPC click from Bing, however, won’t show up in your Google Analytics numbers, making your spend on Bing seem like a poor investment. In contrast, the Google PPC click ends up seeming more effective than it truly was. Scary, isn’t it?

Now let’s pretend our user is a Google loyalist and instead her whole search process takes place on Google. After her initial “halloween costumes” search, ad click, and dissatisfaction, she decides to give Google another shot and searches for “sexy halloween costumes.” This time she sees your more targeted creative on Google, clicks, and the rest is history. How are your GA numbers leading you astray this time? First of all, GA thinks this is one visit when it’s really two, so right off the bat you’ll be over-estimating your conversion rate. You’ll also think all the revenue from the Big Bird dress is the result of one visit instead of two, so you’ll over-estimate your revenue per visit. And, to add to the terror, this purchase actually cost you twice as much as the visits count in GA would lead you to believe.

So why not integrate your Google Analytics and AdWords accounts, and then use the “clicks” metric in GA that’s available if you do this? This metric is supposed to reflect the clicks that come to your site regardless of when in a user’s session they occurred. But now we must deal with the issue of over-counting clicks, because invalid and spammy clicks are reported as clicks in GA but aren’t counted in AdWords. Now, why should this strike fear into your heart?

Say that our friendly user lands on your site after her first search, hits back to do her second Google search but then suddenly remembers that she did see a link near the bottom of your page that she could have sworn started with the letters S-E-X. She hits forward to check it out again and makes her purchase. In this case GA will count one visit (remember the half hour rule) and AdWords will count only one paid click thanks to some magic deduping logic (read about GCLID’s [here]). But GA will count two clicks because Google doesn’t apply this filtering logic in Analytics. Unfortunately for you, this over counting of what seem like paid clicks will artificially deflate your conversion rate and your revenue per visit (you think you have more clicks than you do). Not to mention, when you see those extra clicks in GA coming from Google PPC, you’ll run screaming to your accountant and demand to know how your cost per visit got so high.

The simple fact is that Google Analytics isn’t designed as a platform for managing marketing. It’s a tool meant to enable analysis of aggregate data about your website, and to give insight into common referrals and the paths users traverse on your site. Google AdWords, on the other hand, is an excellent marketing platform for managing marketing — as long as that marketing only uses Google products and properties. If data-driven marketers want to help their clients make a killing this Halloween season, they need to do their job with the best resources available. That means understanding exactly what Google Analytics is capable of and using it appropriately, while leveraging other tools that are designed specifically for accurately tracking marketing. So this October, whether the prospect of sexy cartoon character costumes sends you running for the hills or not, don’t let your GA numbers scare you into misallocating marketing spend!

written with Allyson Taylor – Convertro Sales Engineer

========

Instead of letting our mutual client know that I thought this company wasn’t productive, I emailed them to make sure that they were 1) only doing a test to determine the ROI of this channel and 2) that they had the means of determining the ROI of this marketing partner. Unfortunately, I reached them too late because I learned that they had already entered into a yearlong deal. This surprised me on multiple levels and motivated me to write this column.

From my perspective, here’s how to scale up a marketing program with one person and very little risk.

1) Determine your ROI. This should be close to your gross margin but not always (e.g., if you expect to gain repeat business from a single visitor). Make sure that everyone in your organization is in agreement.

2) Establish that you are going to buy all the marketing that meets your ROI goal and discard those marketing programs that don’t. Programs that are close to your ROI goals should be given a short leash upon which either the vendor or you improve until they reach the ROI goal.

3) Establish a test budget. I’ve often used $5K per week or month depending on the expectations of the category. Most marketing programs can establish what’s going on with a $5K spend.

4) Measure the performance of each program. You will need to use a multi-attribution model or else you will not be able to determine the actual net impact of the test program.

5) Put the programs that meet your ROI goals on auto-pilot. Buy as much inventory as you can. Make sure that it’s clear in your contracts that if the ROI goes down, you can terminate with minimal financial damage.

6) For the programs that don’t meet your ROI goal, either terminate the program or provide the salesperson the data that you’ve collected so that he or she can reduce the price to a point that would make economic sense.

7) Set up email alerts to notify you if the program suddenly goes below the ROI threshold (or suddenly spikes). React to these emails accordingly.

8) Move on to the next test.

Most of my time is spent tweaking those programs that are close to my ROI goals so that they will work. Sometimes it involves reducing the reach of the program to exclude visitors by geography, time, etc.