The SEM branding problem (that current technology can’t solve)

Search Engine Market (SEM) has a branding problem, actually two, and big ones I may add.

The first one is that companies don’t spend the significant amount of money in SEM for branding purposes. Even though everyone can agree on the fundamental principle that users go through asearch funnel” when making a purchase (i.e. broad generic searches at first to learn more about the product category, followed by specific ones to conclude the purchase), very few digital marketers would be willing to spend money inupper funnelclicks ; this because it is harder to quantify their contribution to the final transaction, but really easy to see their (high) contribution to cost.

The second one is that no one in this space is willing to pay for what they cannot measure directly. Since the dawn of SEM, every company, technologist, and professionals in the space, based their pitch on the fact that they could finally offer to the ultimate form of trackable marketing: “the TV era is over, now you can understand and track your real contribution of your marketing dollars to the cent“. And marketers started believing it, to the point that no one would give SEM any credit for its brand value: everyone know is there, but since you can’t measure it, no one is willing to pay for it.

Photo credit of MKFeeney
Photo credit of MKFeeney

And while companies are desperate to find ways to spend additional money to move the needle marginally in one of the key constituents of branding (awareness, recall, consideration, intent), SEM is getting none of that budget. Not only that no one can accurately measure the correlation between awareness lift and sales lift, but in many cases the branding impact is given on faith. Don’t you believe me? Let me know if the following sentence sounds familiar: “This is the year of mobile, you want to reach teens, but they are leaving Facebook in droves, teens spend an enormous amount of time on mobile devices, you need to buy mobile ads.” And you do, regardless of the fact that tracking mobile ads is still very hard, surveying teens is often not allowed, and registering a lift at 90% confidence level requires volumes that very few networks offer. Since the value is there in theory, it becomes acceptable to invest. It’s the most classic example of deductive fallacy.

Why have marketers no problem accepting a high degree of uncertainty with ‘brand marketing’, but pretend absolute measurability when buying SEM?

It’s like two brain hemispheres:

  • The left one, where SEM lives together with scientists and mathematicians, can’t conceive a world that cannot be measured, analyzed, and studied in all its components;
  • The right one,  where Branding lives with artists and free spirits, things are done because they ‘feel right’, and questioning this logic means not belonging there in the first place.

Put the two together and you will have the perfect picture of the schizophrenic brain of today’s marketer.

But I get it: upper funnel keywords are expensive, and you can quickly rack up a huge bill in non-purchasing clicks; in the last-click attribution world described above, you can’t get budgets for long if you can’t bring back some measurable result.And bringing results is anything but easy: cookies deletion, multiple devices, offline sales – just to name a few – make almost impossible to get any read on the impact of generic keywords.

Image credit James Royal-Lawson
Image credit James Royal-Lawson

Click-based analysis hardly ever offers a satisfactory impact on sales to justify the increase in spend, and that is also the reason why even very sophisticated companies often default to last-click attribution. It also ignores the branding value of a search impression entirely.

But the truth is that the current solutions for measuring the Brand Value of Search (BVOS) is anything but good, and until the situation improves, SEM will not see an increase in branding dollars. Here is a snapshot of the most common methodologies used nowadays for measuring BVOS:

  • In lab studies offer a very distorted view of the navigation patterns of a real-life user: “imagine this”, “type this keyword”, “look at the results”, “answer these questions”. If you can get past the self-fulfilling prophecy of users remembering what they just saw in a forced environment, you have to deal with the fact that they cannot navigate past the search results page, making it almost impossible of measuring a change in consideration and intent.
  • Panel studies offer you a more realistic view of the world, but panels where you can modify the navigation experience of the users (e.g. in SERP) are very expensive to maintain and hard to reach volumes large enough to have good statistical significance and overcome sample bias. When companies promise not to alter the navigation experience of their users, they can manage to reach larger panel size, but even the biggest panel in the US cannot achieve an incidence rate that would allow you to monitor anything but the broadest terms. For this reason, companies are not willing to pay the cost of advertising on a national scale to hit a tiny fraction of users in the panel.
  • An excellent solution would be surveying people that have been exposed (or not) to a SEM ad, through display-based systems (i.e. banner ad), but not being able to re-target people that have searched for specific terms, and control for actual ad exposures, doesn’t allow to create a control / exposed test environment.

In conclusion, these are the reasons for the lack of significant branding dollars on Search: a reputation of being always efficient and ever-measurable, none of the ‘cool factor’ of Mobile and Social, the lack of visibility of takeovers, and without the loose metrics of display and video. 

Every one of us has a dominant hemisphere, and for the advertising industry always has been, and still is, the right one.

Author: Paolo

Economist by education, marketer by profession, coffee roaster by hobby.