Speaking at TagMan’s most recent client get together TagMeet 2, Silsbury said: “It’s going to be a pig of a year.” And none of the other retailers in the room looked ready to disagree.

So what does Glasses Direct plan to do about it? Given its focus is online, then drive its online channels hard. Efficiency is the overriding mission. But, that doesn’t mean the company isn’t prepared to invest to get as lean as possible. Silsbury has a couple of things in mind:

1. Data unity: Getting all his marketing data in one place so he can learn to optimise his spend against entire user journeys, not channel-by-channel
2. Conversion: Maximising things like site speed to make sure that any customers that do get to his website, buy – and buy big – before they leave

Both these causes are part of his justification for implementing TagMan at Glasses Direct. The system will be used to enable Silsbury and his team, including long-time TagMan user Oli Elliott – formerly of online clothes retailer Boden – to track the entire online path to conversion of customers and attribute credit for sales much more fairly across the channels that made a contribution.

Silsbury is also one of the first to use TagMan v3, the newest version of the software, which includes several world-first features to lighten the load that tracking tags place on websites, and speed up his website. Silsbury said initial tests showed a 25-28% decrease in page load speed.

“We are obsessed with site speed,” he said. “Apart from the obvious impact on user experience and site conversion, it is part of how Google ranks your website.”

Silsbury has committed to TagMan for other reasons too. He explained how “tagging is an unnecessary distraction” that hampers his team’s ability to implement new technologies and campaigns. Tag management enables them to focus on the things that matter.

But, the biggest challenge is attribution. Silsbury is keen to distinguish the use of attribution tools for understanding user journeys better and for optimising campaigns.

He explained: “For me it’s important to distinguish attribution from optimisation. We mainly use TagMan’s attribution data for reporting, but we want to optimise paths-to-conversion, not individual channels so we use ‘assisted views’ [reporting that shows when a channel appeared anywhere in a user’s journey to a sale] to optimise. The Unique User ID assigned in the system enables us to see any user’s complete path to conversion and so see how all channels work together. It’s important to know that investment in one channel may see the conversion take place in a completely different one.”

Silsbury stated that in reverse to the above, there are very real financial efficiencies to focus on too and attribution should be used to deduplicate and ensure that you are paying the right party for its part in a sale. Using email and affiliates as an example, Silsbury’s argument is that if a customer has been sent a Glasses Direct email, then they can’t be considered a customer 100% driven by the banner they clicked on [up to] 30 days earlier. He wants to focus affiliates on customer acquisition and therefore not pay 100% where email is present in the path. Silsbury was quick to add though that “this is not a money saving tactic, more a way of reinvesting the saving in areas like paid placements, which do deliver genuine value from the affiliate space.”

Silsbury closed by stating that the ultimate goal is “getting to a fractional model, where split percentages can be applied to individual channels dependent on their importance in the path.”

“Ultimately it is a question of understanding, as Oli puts it, the ‘golden combinations’ that drive sales. We’re not there yet, but we’re getting there.”

This is the second of a two-part interview. You can read Part 1 here.

How do you find out your site has a page load performance problem?

There are a number of services out there that provide really complete monitoring of your page load times. Those range from fairly simple services that are often free. They ping your site, or a couple of pages on your site, and tell you about page load times. There are fancy, more sophisticated and comprehensive services that will ping your site from a variety of locations, including international locations, and do it on a regular basis. They’ll give you back reports that tell you which pages on your site are loading, what your average page load times are, and in some ways even more important that average page load times, how much variation there is.

It’s just not the case that there’s a single number that represents how fast your pages load. It varies a lot depending on where people are coming in from, the time of day, what your server volumes are, what pieces of the site they’re trying to access.  There’s not just one number to look at here, but there are a number of services that provide that kind of information and intelligence on an ongoing basis. It’s not super-expensive for the most part. It’s something companies who are serious about their web site would be well rewarded for investing in.

Are all tag management systems created equal when it comes to increasing site performance?

I don’t think so. Tags in general have become a more significant part of the performance problem. Web analytics tags tend to be some of the heaviest tags on pages. As far as  web analytics goes we have a similar temptation to what I spoke about earlier in terms of user interfaces: it’s natural enough to want to include as much functionality as you can, and sometime that leads to fairly weighty tags. Many tag management systems can actually help reduce the burden tags place on your site.

Many of our enterprise clients don’t have just one tag on their site, they often have half a dozen, 10 or 12 tags. That starts to add up. It’s just one piece of the overall page performance equation, but it’s a very controllable piece.

Most tag management systems give you at least some benefits when it comes to tag loading, but there are very significant differences in the way tag management systems are engineered and the way they take advantage of what’s on the page. Some simply load in a tag governance system but they don’t improve the performance that dramatically.

Then there are ways the tag itself can be engineered to actually improve the performance loads. All those things are significant. The bottom line is the faster your tags can load and still do the job, the more room it gives you to play with the user interface and still have fast-loading pages. It’s definitely a win-win situation where if you can make the measurement component really tight and fast, it gives you more opportunities to do things and experiment and build out your page GUIs without impacting page load times.

Any final thoughts?

One thing that I really wanted to talk about in the webinar is what I often see people not pay enough attention to when they think about page load times. People often over-focus on a few pages on the website. There’s the homepage in particular, and then sometimes the shopping cart functionality. Those are vitally important pages, I’m not going to argue against that. But we have seen sites that were good about optimizing those pages but they let significant performance problems creep into other areas of the site. I’m going to present a case study in the webinar of one client who let that happen to their search functionality. Search is a critical intermediate step between the home page and the shopping cart! Yet it’s not something people really measure on an ongoing basis.

One of the points I want to make is to really explain you need to understand a wide variety of page performance issues across your site. Sometimes there are systems within your site, like search, that will perform dramatically differently than the rest of the site. So it’s really important to look at page load times across the spectrum of pages in your site and try to optimize them – and be aggressive about that.

Please register to attend the Sept. 29 webinar.

We looked back over the past year (well, it’s only been eight months, really, since 2011 began) and realized we’ve undergone quite the growth spurt here at TagMan.

Consider these statistics that reflect where we were in Dec. 2010 versus where we are now:

• Revenue grew by a factor of 3X
• Clients 2.2X growth. And because revenue is growing faster than clients, it means our average revenue per client is increasing, too.
• Space: Both our London and New York offices expanded into larger spaces this summer.
• Personnel: In line with client growth, headcount has more than doubled, from 26 last Dec. to 55 currently.

We’re trying to grow even bigger. We’re hiring in both our New York and London offices. Engineers, developers, product managers, account leaders.  If you’ve got what it takes and are interested in joining a rapidly growing and successful startup, we should talk.

Shanti Shunn, Director of Online Marketing at Harry and David

Shanti Shunn, director of online marketing at Harry and David, discusses why e-commerce sites need tag management, how tag management has evolved, and how to select a tag management vendor.

When and why did you decide you needed tag management?

At Harry and David we’re in all the multichannel markets, we have a lot of third-party tags on our site that relate to tracking and/or CPA payouts. I built my first tag management about 10 years ago, back when it was still called pixel wrapping. You had people who would touch multiple channels and check out, and you’d have multiple third-party tags firing. Based on them touching multiple channels, you’ve reported that order to multiple vendors.  And of course you’re reporting 100% of the order, so you have three different vendors who show the order as one of theirs.

It came down to attribution and at the same time order tallies based on the last channel touch. You don’t always want to optimize to that, but you want to facilitate the visibility.

So if you have multiple CPA vendors you’re reporting a percent of that order out, so you report that same $100 order to three CPA vendors and one’s 5%, one’s 10%, one’s 15%, you’ve respectively paid out three times on that order. Your margin goes out the door because you’re paying everybody, and you’re not paying for their contribution in closing that order, you’re paying them for the full order. Affiliates play a big part of that piece because you can’t send them a partial amount.

Can you estimate how much money you’re saving through tag management?

I would easily say more than a quarter million dollars in excess payouts. At my previous company we probably saved the company in excess of half a million to more than three-quarters of a million in the first year.

It ends up paying for itself, especially if you have a very robust affiliate program or you’re doing any type of CPA-based ad purchasing or CPA partnerships. You really only want to attribute and optimize based on their role, first in or last in. There’s a whole world of corrections and cancellations and fraud. If none of that existed, it probably wouldn’t be as good a deal. You have to have the ability to give that order a one-to-one relationship with a channel.

Interesting that you’ve been in the field of tag management for 10 years. How have things evolved?

Back in the day when affiliate programs were really starting to emerge and there were gobs of different networks out there you had a massive pool of CPA orders and more people doing CPA partnerships as more of an exploratory way to pre-test an agreement. One thing I noticed was that every time my confirmation page was loading I was using source code and broadcasting pixels to all my third parties. So I asked, “What do I need to do to clean this up?”

So the first incarnation of pixel wrapping was developed to say: if our system develops this source code as the final source code of record with this transaction, then it becomes associated with a pixel. That’s why it became called pixel wrapping. You associated whatever your onsite tracking mechanisms were to the different third-party pixels.

 Now there are new things bundled into tag management such as site speed, privacy, data collection. Do these features play an important role for Harry and David?

Site speed definitely is part of it. When I was doing the RFP for tag management one of the things I brought up was speed was not my main concern. There is a value to that, but it’s more of a side effect. Tag management has a lot more benefits than asynchronous loading. It’s going to save you money, give you a cleaner view of your business, give third party vendors and partners a clearer view of your business so they’ll be optimizing based more on real numbers.

Whereas if you have PPC advertising, there may be a lot of conversions, but those keywords might have been more of an influencer.

Speed is a benefit given the world we’re in, but I would never recommend somebody go to tag management simply to speed their site up another couple of seconds.

This is the first installment of a two-part interview. Come back next week for more!

This will save time and money in the way Confused.com runs its online campaigns since the tags on which they rely can be managed through a single online interface rather than through back-end code changes to the site.

Tom Beverley, digital and customer marketing director at Confused.com, said: “With TagMan on our site we can remove a large portion of the time, money and energy we currently spend on tag management and redirect that focus on marketing the business. TagMan will allow us to make changes to campaigns much more effectively and look to use new technologies and suppliers without the significant barrier of tag implementation.”

Tag management has become a major headache for online advertisers, slowing down pages and prompting major problems with implementing and tracking campaigns.

A recent study by TagMan into tag implementation issues found they have caused a loss of website sales for more than a quarter of marketers and a loss of site traffic for almost a third. More than half have lost campaign performance data, delayed the launch of a new campaign and delayed the use of a new technology. More than a third have decided not to use a new marketing technology thanks to the pain of tag implementation and nearly one in five have decided not to launch a new campaign.

Meanwhile, more than 40% spend more than a day a month implementing tags; more than half of businesses incur direct IT costs to implement tags on their sites and almost a quarter (24%) spend more than $1,000 for every tag implementation.

Confused.com is the first price comparison engine to sign up to the system, which is already in use by the likes of Asda Finance, Boden, Subaru and Virgin Atlantic.

About Confused.com
Confused.com is one of the UK’s biggest and most popular price comparison services, generating over two million quotes per month. Launched in 2002, it was the first insurance price comparison website, specialising in car insurance. Confused.com has now expanded its coverage in this product area and can provide quotes for car insurance for women, classic car insurance, and car insurance for young drivers, amongst others.

Confused.com is not a supplier, insurance company or broker. It provides a free, objective and unbiased comparison service. By using cutting-edge technology, it has developed a series of intelligent web-based solutions that evaluate a number of risk factors to help customers with their decision-making, subsequently finding great deals for them on a wide-range of insurance products, financial services, utilities and more. Confused.com’s service is based on the most up-to-date information provided by UK suppliers and industry regulators.

About TagMan
TagMan is the smart container tag for enterprise e-commerce. By acting as a single, independent, universal tag and interface through which tracking tags and pixels can be deployed to an advertiser’s web site, online marketers can save time and money in the way they track campaigns and capture the complete path to conversion. TagMan was founded in November 2007 and has offices in New York and London. Clients include Virgin Atlantic, Subaru, Boden, Laura Ashley, Thomas Cook and Air New Zealand.

Hanna Kimuyu: Greenlight initially integrated TagMan data (client specific) to understand the relationship between paid and organic search, but have since expanded this into understanding the full user buying cycle. The data findings so far have confirmed our initial thoughts that the user doesn’t have a preference between paid or natural search when searching for products or services. In fact, the cross interaction between the two media, suggests that users are just focusing on their ‘search phrases’, and then picking the most relevant brand to buy with. Furthermore, as advertisers diversify their online strategies more, the conversation becomes bigger than just understanding the relationship between paid and organic search. Brands should be asking themselves ‘what is the relationship between all online channels’ ‘do I have the right media mix’ and ‘is my strategy cost effective’.

TM: How is Greenlight applying that insight to search strategies and tactics for its clients? Can you give any examples?

HK: Even though search marketing is over ten years old, most advertisers are still using the first or last-click model, meaning anything that happens in between is completely ignored. Having both full click and channel path data allows Greenlight to analysis the entire user search journey, from first click to conversion. Resulting in several things:

1. De-duplication of data, no more over paying affiliate commissions.
2. Introducing Attribution modelling, fairer way of awarding affiliate commissions and sales/revenue to contributing channels.
3. The importance of each channel, specifically the role each channel plays within the buying cycle.
4. Transparency of the full click/channel path, what are users doing, what search phrases are they using and what channels are they interacting with to find the most relevant product or service? Are there are patterns in how users search? Can we use this data to shape our online strategies?

Stage one is very much about discovery – understanding the data/making sense of it. It should be noted that this isn’t a quick process, and getting all the relevant people around the table to discuss and analysis the findings is critical. At this point it makes sense to apply a ‘version one attribution model’, which again will need to be discussed and evaluated before making any further changes. Although this may sound like a time and resource heavy exercise, it is beneficial. Our TagMan clients are seeing cost benefits in year one, and now have a transparent view of how their users interact between the channels before converting.

TM: Paid search listings are often accused of gaining disproportionate credit for sales thanks to the predominance of last-click models, and the channel’s intense measurability. What’s your view of that debate and is the TagMan data Greenlight sees providing any illumination?

HK: For a lot of advertisers paid search represents a larger proportion of their online advertising budget, suggesting almost a reliance on the channel. If this is the true, paid search represents a larger proportion of the sales because in most cases users will finish their search journey on a ‘branded click’ i.e. brand search term. Especially, if the advertiser is bidding on branded terms, then the last click model will of course favour paid search over the other channels. In this instance, this is very much down to the strategy in place and the fact the advertiser has chosen a last click model – hence why ‘paid search is often accused of gaining a disproportionate credit for sales’.

The TagMan data has been pivotal to highlighting the ‘real’ role brand and paid search (as a channel) plays in the entire user search journey. The trick is not to work to a last-click model continuously, but to start off with a last-click model as a benchmark before moving into attribution, so you can really appreciate what each channel brings to the overall picture. Having gone through the same exercise for a travel client, initially we discovered that, on the last-click model, paid search represented 29% of all sales (of which 22% were branded sales).

Having analyzed the full channel and click path (search only) data we then applied a fair attribution model, taking the decision to downgrade paid search branded sales. The outcome saw paid search still being the second biggest contributor, delivering 25% of all sales. Proving that paid search was working just as hard as the other channels, if not harder, to drive sales.

TM: How, specifically, are organic listings revealing themselves as drivers of customers through the purchase funnel?

HK: First and foremost we’re confidently tracking all organic listings. I have to point this out specifically because in most cases the traffic and sales resulting from organic listings are normally assumed from the ‘whatever is left’ bucket. To explain, most advertisers will tag their paid for online activity, that being paid search, display etc, therefore leaving a bucket of remaining traffic and sales. With TagMan, Greenlight has been able to identify Direct to Site (DTS) traffic, i.e. domain, bookmarking and organic traffic specifically. Allowing us to confidently understand what value organic traffic and optimisation brings to an overall search strategy.

There’s no one approach or outcome though, the value is completely is down to the intended strategy. Using the travel client mentioned above, their intention was to use organic search to pick up the long tail, e.g. ‘hotels in New York with a swimming pool, close to Central Park’. This search term might sound unreal, but when a user is close to converting or has a specific product in mind, the long tail proves quite useful.

Traditionally (and it’s still the case) paid search picks up on most of this traffic. Long tail search, although low in volume, is cheap and effective, normally resulting in a strong conversion rate. However, if you have the flexibility to create specific landing pages for these type of terms (to which the travel client mentioned did) then it also makes sense to drive your organic listing there as well. Especially if targeting some of the more competitive terms does not cost in. With this strategy in place the travel brand was able to see the value in building such landing pages, and the role organic search played to its overall strategy.

TM: What new insight have you been able to derive about the way in which different keyword groups drive people from research to conversion?

HK:
1. The less surprising lesson but most valuable is the fact that a buying cycle does in fact exist. Users will combine generics, e.g. hotels in London, with branded, e.g. Guoman hotels in London; with more product/service led searches e.g. conference hotels in London. There’s no bias towards paid or organic (natural) search. If the listing or advertisement is appealing and relevant, a user will click – it’s simple! It may sound strange to highlight this point, but advertisers still don’t appreciate the value in combining a range of research, consideration and branded terms when bidding via paid search or optimisation for organic search. Users have become more sophisticated when searching for a product or service, therefore if a brand wants to be considered, then they have to be present throughout the users search journey.

2. The biggest surprise has seen the average buying cycles almost doubling for both retail and travel. Typically an advertiser will set a cookie length of 30 days maximum to capture a user’s interaction before conversion. From recent analysis into both the retail and travel sectors, it has become apparent that user’s are taking longer to convert. Part of this can be explained in the click/channel path. Firstly, the number of channel interactions have grown; as well as the number of searches conducted. Secondly, the user is looking for a ‘bargain’ – even if they decide they want to buy from a particular brand, they will still search for discounts. Lastly, users are savvier at searching, although generics still rule in terms of the number of searches made per month (volume) – users are combining a range of keywords before converting.

TM: Can we offer any more insight into the debate around the way in which paid and natural search strategy can be better integrated to increase conversions and reduce cannibalization?

HK: Yes – position analysis as part of the attribution. One of the biggest debates in search is – what is the benefit of having both paid and organic search listings for the same search term? Although we’ve been able to prove cannibalisation when advertising on the same keyword for both paid and organic listings via an attribution model; this data doesn’t take into consideration the position of each listing. There are plenty of research studies that prove there is an uplift in click through rate if both paid and organic listings sit alongside each other. That said, what is the perfect position combination?

TM: As online retailers gain clearer sight of their customers’ entire paths to conversion, how do you see their approach to paid and natural search evolving?

HK: Revenue-driven strategies; as a result of having a clearer understanding of how and what to integrate, how do we develop a strategy that can almost predict a revenue outcome? Allowing advertisers to decide where and when to spend. Furthermore, search as a channel isn’t just about paid or organic search. Social media advertising channels such as Facebook’s Placement Programme, or Google’s Display Network (AdSense and Double Click collaboration) are also components of search. How is this changing the data and our view on how search is contributing to the wider picture?

TagMan: Your presentation at the first ever TagMeet looks at how you and your colleagues at Boden are using TagMan to optimize paths to conversion rather than channels. Can you explain a bit about what that means?

Oliver Elliott: Optimising channels seems like a bit of a contradiction to me, or at least, a difficult thing to do in isolation of other events. You know the vast majority of customers use multiple channels to reach your brand so to optimise conversion goals you need to be optimising successful channel interactions in combination.

TM: What is your view of the last-click model of attribution, whereby only the channel that delivered the last click in any user’s journey to buying online gets all the credit?

OE: It’s one of many ways to attribute activity across multiple channels and can be helpful as long as there is awareness of what sits outside of that view.

TM: Why in your view, does so much of the digital industry work to the last-click model?

OE: It’s legacy and simplicity. I would love to work more with first click biased models but you never know with certainty when and where that first marketing interaction is. With online you are defining this artificially with your attribution window. Add offline and things get even more complicated (interesting) – was the first touch-point a prospect catalogue they received, a PR piece they read, word of mouth? You can argue further and further back with first click. With last click, you have the conversion, you attribute 100% to the closest event and there’s no room for argument.

TM: What is the impact of optimizing channels, rather than paths to conversion, in terms of an e-commerce business’ marketing strategy and execution?

OE: Using only a single attribution view you will have instances where co-operative channel interactions in successful paths to conversions are hidden. With a channel centric strategy you will naturally optimise away from these interactions as you are not attributing success to them. This places the whole path at risk and you’ll see a drop in contributing channels without being aware of the reason.

TM: With the ability to look at entire paths to conversion, what insights areBoden able to draw about its customers’ behaviours? And how is Boden shifting its activity to suit?

OE: We’re still trying to join the dots at the moment. The objective, as it’s always been, is to co-ordinate activity into integrated marketing campaigns. Currently we’re looking at high volume high impact initiator activities such as prospect catalogue mailings and attempting to analyse what downstream marketing interactions are required to increase the propensity for conversion.

TM: Based on your own experience, how would you recommend other clients go about moving beyond last-click models?

OE: The first step has to be about getting awareness of the picture outside of last click. Obtain visibility of assisted conversions for each channel – this is the conversion picture without any attribution applied, complete with overlaps. What is the difference between last click and assisted click? How much channel contribution are you not seeing because it sits outside of last click?

TM: What has been the reaction of your agencies and other partners to your use of TagMan and the shift away from pure last-click attribution?

OE: Fairly positive as the shift we’re making for CPA activities just now is to optimise with assisted conversions thus giving greater merit to the channel. Bizarrely, this takes quite a bit of getting used to… Of course, targets are toughened up accordingly and we keep a careful eye on attributed and overlapping pictures.

TM: How have you been able to justify your investment in TagMan in terms of ROI?

OE: The sell is differs depending upon the ‘customer’. The easiest financial justification is around inevitable dedupes that will be applied (analysed by marketing and applied by marketing). The benefit to the business, of course, extends far beyond this – with IT the sell is the saving to their time and increase in site speed. With our CRM and data team I can talk about the bigger objective of touch-point analysis.

TM: In your view, what is the short and medium-term future for online marketing attribution and what impact might that have on the development of the digital industry as a whole?

OE: The short term will be a move away from business models that favour last click to those that benefit conversion across the full customer path. Not that I see this as a huge concern at the moment but a slight shift can only help publishers to naturally innovate rather than being forced into a particular business model. Moving to the mid term, I’m hoping to see full integration of offline and all digital into the path to conversion. This should give us an almost surgical like ability to supply the correct marketing interaction at the correct time to maximise response goals with any campaign.

Next week, Hannah Kimuyu, PPC director at search agency Greenlight, on how it uses TagMan data to inform its understanding of paid and natural search results in users’ paths to conversion.

Find out more and follow the conversation at http://www.twitter.com/#tagmeet