thefunnel

On-line advertising is a tactic in the promotional mix tool box, because it can be measured. It seems easy enough to look up the variables and costs from the publisher’s rate cards, but things can easily be misinterpreted, unless marketers understand some of the more in-depth details of the various on-line metrics and what they’re trying to achieve. Typical metrics include Impressions, Clicks, CPI, CPC and CTR.

What does an impression really mean? Technically speaking, one impression means that the ad was served up once through a browser. That’s all fine, but what if the ad was below the fold, or the user never looked at the ad? A similar challenge also exists for print advertising and here many marketers use something called OTS (opportunity to see), to estimate the number of real impressions. In its simplest form what OTS means is that a reader needs to be exposed a certain number of times to an ad, before it’s counted as one impression. For B-2-B print advertising this means that you have to run the ad 3 – 5 times to reach the reader. For on-line advertising you can estimate the impressions in more detail and with greater accuracy, but only if you know the number of unique visitors and the visitor return frequency for the duration of your campaign. In addition, all impressions are not equal. A web site have many different types of visitors, many of which aren’t in your target market, so unless you can determine exactly when your ad in running (through for example context sensitivity) or the persona your ad is served up for, you need to be more conservative with your calculation of the total number of impressions.  From a cost perspective this means that to reach the intended number, the campaign will cost more, but you’re always better off knowing the true cost, than some artificial number.

A click is one step further in the funnel, and therefore easier to put a value on than an impression. The only problem here is that the CPC for an on-line ad tend to be way more expensive than for email marketing or for a PPC campaign. So why do marketers run on-line ads, although the ROI comparison doesn’t support it? The answer here is that on-line advertising, email marketing, or PPC marketing can’t and shouldn’t directly be compared, because they work in different stages of the funnel. Marketers need to know for their particular audience what promotional tactics work best in what part of the funnel?

Even when being fully aware of all these details it can be difficult to sometimes justify or to explain the return for an on-line ad campaign. Therefore, the campaign always needs to be looked at holistically, to understand the interaction between advertising and other promotional tactics, in particular PR is, and what other benefits it provides. It is also a good practice to run pre and post impact studies, and to do message and concept testing to minimize the risk of wasting any money.

From a global campaign perspective, you also need to measure each country you’re targeting on its own and not compare countries to each other, because the cost of reaching a particular population varies greatly, as does the general quality of for example a click through.

Finally, a few words about awareness, because it’s very common that a marketing plan calls out for increased awareness through the use of advertising. To start with awareness on its own is never a good measure, because it is seldom measurable in a way that you can contribute a change in awareness directly to a specific tactic. In addition it doesn’t tell you anything about perception or preference, or what action a customer might take.

Here’s one example how to think about measuring the success of a B-2-B web site. To start with you should first organize your metrics into different high level groups. The typical high level groups would be web acquisition, web engagement, and web conversion. The next step would be to define sub metrics for each group.  For your acquisition metrics should count and track the number of visits, unique visitors, returning visitors for different sources of referrals, such as organic search, paid search, marketing campaigns, and direct traffic. For engagement you should measure, visit depth (or page views), visit time (duration), visit returns, visits per visitor, bounce rate, exit rate, number of comments and content consumption (or form completion). In addition you must also be able to cross reference acquisition sources with the various engagements. For paid search, you might also want to drill even deeper down separating golden key words from, generic ones and the long tail words. This will enable you to scratch the surface in understanding ROI for your search investment. Finally, the web conversions also need to be divided into sub-groups. My suggestion is a four tier approach. Responses are the lowest value conversion events, where the visitor leaves a trace, but there’s no follow-up activity. Generic success events are things you want to count and put a value on. These could be certain page views (for example, example a new product), new names registered to the database, or sign-up for newsletters. The third tier is the selected few, most valuable success events. These could be contact requests, partner referrals, service requests or more valuable page views such as pricing views. And fourth are your key performance indicators such as a request for a quote or sales leads. Similar to web engagements, you should be able to trace the web conversions all the way back to the acquisition source. In addition since we’re talking about a conversion event, that normally requires a web form of some kind, you should also track completion rates or abandonment rates.

And finally you should be able to slice and dice this data based on the business segmentation (your web taxonomy). In addition you should also create a dash-board where you track the trending of all the different conversion percentages rather than the absolute numbers. If you then are also able to correlate this data to your web satisfaction data, measuring things such as overall satisfaction, would recommend, able to find, and a positive search / browsing experience, combined with how much money and effort you’re putting in, you’re much closer to understanding your real performance and business contribution of your web site.

Web analytics is hard and difficult! How hard and difficult can it be? It’s only data. Only data? – that’s exactly what the problem is. Unfortunately many companies get stuck, trying to fit their existing marketing metrics thinking into the area of web analytics, when to be successful, it requires a whole new approach, and thinking, and understanding in the limitations of the data based on how it’s captured and configured, in addition to an investment both in technology and in people.  But before even looking at what to invest in, companies first need to decide what they want to measure and why. Too often things are just measured for the sake of measurements, without knowing what it means for the business.  Even if the data in some cases is taken to the next step, and analyzed, it often stops short in providing decision makers with actionable managerial information. So the first step is to work backwards from the high level business objectives and overall metrics, to understand where web analytics can provide some insight. If, it for some reasons doesn’t, that’s also ok, but then companies needs to not waste resources pretending. Secondly various business metrics needs to be holistically aligned. This means segmenting the data and isolating key audiences, so that all measurement sources segment the data in a similar way, aligned with the business / customer segmentation. For example, lets say the web satisfaction data is segmented into large and small businesses, but visitors tracked on the site can’t be filtered with the same granularity, and maybe this is not at all the way the business segments their customers, would basically mean that the measurement system is more or less useless (for business intelligence purposes). Even if the segmentation is aligned too often various business data live in silos and can’t be easily correlated. Thirdly, the business needs to invest in technology and people, because unfortunately good information does not come free. Technology is easy to acquire, but extracting the value from it, typically needs some unique resources. And I’m not talking about the mainstream marketing person. What is required is someone with a combination of analytical skills, business analysis skills, marketing skills and domain expertise. A quite common mistake in many companies is to not invest enough in analytics resources versus technology. And last but not least, any limitations, either because of system limitations or the way the system is configured also needs to be considered, because web analytics is often instead of being perfect or the aboslute truth, more about the trending.