Jun 92011

Moral Dilemmas in Agency Work

A few days ago I read a fairly poor post on SEL, which added nothing to the conversation about advertising on one's brand, but did spark me to think about an interesting moral dilemma.

I'd love to get your thoughts on the topic.

The author advanced a number of standard, reasonable arguments for advertising on one's brand: occupying more space on the SERP thereby pushing down competitors, controlling message and landing pages, and the idea that brand ads are generally low cost and therefore not a big deal. We've made similar arguments in the past.

She also advanced some controversial claims about the incremental traffic gained through these ads: we'd argue your mileage will vary dramatically and we've yet to see a comprehensive, carefully done study of the issue.

Finally she said some stuff that was factually inaccurate, but I'm not going to get into that here.

What is much more intriguing is the more subtle message implied throughout the piece and demonstrated quite clearly in the following three quotations:

Removal of brand keywords will seriously impact retailers’ ability to hit forecasted goals from this channel.

Each of us knows that our brand keywords are our bread and butter and allow us to support our testing and expansion plans by floating the traditionally less efficient non-brand keywords and content networks. Removing brand keywords will result in shrinking efficiencies in terms of reported ROAS.

Let’s not forget that the dedicated service and support retailers receive from vertical reps at the engines, account reps from tool providers, and the large teams from agencies. They are all widely influenced by the investments that retailers make in their PPC programs. The reality is that removal of brand keywords could potentially lead to a reduction in dedicated support from vendors.

The message is pretty clear: lumping brand and non-brand search performance together makes the program appear cost effective. If you take out the brand phrases the program will appear to be horrendously inefficient and people higher up in the organization will slash your budget. Therefore you must advertise on your brand so that you can hide the truth about performance from the rest of your company and get a larger budget.

Wowzers! That is some message.

Since inception, RKG has vociferously argued that separating the performance of brand and non-brand (with proper attribution of credit for multiple touches) is absolutely imperative for understanding the true value of the channel to your company. Moreover, since the vast majority of brand search is driven by offline marketing, PR, brand awareness and word-of-mouth referral, lumping that in with competitive non-brand search means paid search performance will be judged based on a huge chunk of sales over which a paid search manager has no control. That is neither fair, nor wise.

Given our mindset at RKG my first take when I fully absorbed that message was: "Holy Mackerel! These folks are essentially suggesting a conspiracy to defraud the advertiser to gain a larger budget/staff for the paid search manager/online marketing team. We'd never participate in that type of behavior! That's disgusting!"

Then, I started thinking: "Hang on now, for whom do we actually work: the people who hired us, or the corporate entity that ultimately pays the bills? If the Director of Online Marketing says: 'We want to justify a larger budget, how can you help us do that?' maybe it's our job to do that. We're being paid out of that department's budget, so maybe what's in the interest of the company as a whole is none of our business? Who are we to tell them they're thinking about this the wrong way?"

To date we've really never been placed in that position explicitly. We do have a few clients who want to look at the "overall" numbers combining brand and non-brand, but we've never been asked point blank to help them make a case for a larger budget.

What do you folks think? If part of an organization asks you to do something that is likely hurtful to the organization as a whole, should you do it?

It strikes me that doing lousy work on demand leaves an agency vulnerable when others in the organization, people new to the organization, or other agencies look under the hood. At the same time, is it really our role to tell clients how to think about their data?


19 Responses to "Moral Dilemmas in Agency Work"
Jim Novo says:
In the analyst role, I've always believed it's my duty to make sure people fully understand the situation before they make a decision. I can suggest the optimal solution, but it's not my job to make the decision, or criticize the decision. If the "client" chooses to hide information from the "business", that's not my call. I don't like that idea, and try to avoid it in every way I can (including turning down new clients who lean that way) but I can't control it and perhaps don't have the full picture. I hear you on the "look under the hood situation", I often suggest to people the best time to implement analytical approaches that defy mainstream groupthink (e.g. using control groups) is when you change jobs! As the mechanic, it's uncomfortable to say as the hood pops, "We suggested the air filter be changed but the previous owner decided otherwise". Hopefully the new owner's answer is, "I think that's a good idea, let's get 'er done".
David B says:
Good post. I lost some sleep when we broke out our branded term performance, but Including it in paid search didn't pass the "makes sense" test.
Sage advice, Jim. That's been our approach, traditionally. Make sure everyone understands the likely consequences of different choices but ultimately they're the boss, we do what they want us to do. I totally agree, we're not in a position to judge what's best for the whole organization. That said, if we were asked to make a pitch to the rest of the organization and directed not to explain the likely consequences of different choices that might be problematic. I agree, we should try to avoid the folks who would ask us to do that sort of thing.
Jim Novo says:
"directed not to explain the likely consequences of different choices that might be problematic". Agreed. If people want to shade the truth it's up to them to take responsibility for such an action. Tough call when you are an agency, for sure. Hey, you can always include the bothersome info in "backup" which nobody will read, right? One of the things that bothers me about the web analytics culture is a tendency to think they exist to prove something was done correctly, instead of analyzing and presenting the outcome - good, bad, indifferent. This is a particularly nasty problem when the analysts report to the person whose work they are analyzing - a really bad idea, in web analytics, or any other business analytics. That's why a lot of business analytics is in Finance, where the analysts essentially report to "the company". The few web analysts I know that work for the CFO think it's the best job situation they have ever known...guess there's something very comforting about having broad permission to be honest...
Thanks Jim, Yes, it is remarkable the degree to which our approach resonates with CFOs. The folks who pay the bills get this stuff immediately :-)
Thanks for your comment, David. For a tremendous brand like Nordstrom it is particularly important. Internal turf battles over budgets just don't make any sense for the business as a whole, but too often the incentive structures essentially force folks to do the wrong thing for the company.
Pat East says:
If folks are using that information to hide the truth, I think that's pretty short-sighted for them and for their companies. I'd hate to employ them. However, while the author insists on including branded terms to lift the overall performance of the account, that doesn't necessarily mean she's doing it to hide the truth. There's a couple of other potential stories: 1) The branded terms give the non-branded terms more "runway" and time - so that the search marketer can figure out how to make those keywords work and generate a good ROI before upper management reduces budgets because things aren't working. 2) The branded terms - just like the non-branded terms should be - have an aggressive conversion optimization program in place that lifts their overall performance. The article implies that including branded terms is a "gimme" - and if you don't do conversion optimization, I would agree that it is - but if you're testing landing pages and messaging, you're improving the overall performance of the account and generating a more positive return for your client.
Pat, to your credit, you are more charitable than I am. Those are reasonable arguments, and maybe that was her intent...I doubt it. I'm also not sure that improving the look and feel of what is either the homepage or something much like it has much to do with optimizing conversion on thousands or tens of thousands of more specific landing pages used for more specific, non-brand keywords. Likely depends on the type of account (retailer vs mortgage loan company) and variety of landing pages that make sense. Thanks so much for the excellent thoughts!
Sid Shah says:
Hi George, From my POV it appears that the crux of your post is about transparency. Needless, to say I am on the same page as you. Forntunately, most of the clients we work with are performance obsessed companies with minimal politics so I havent seen many instances of such moral dilemmas. I have a slightly unrelated question/ask: I keep hearing about SEM cannibalizing SEO clicks but I have seldom seen any good analysis on this topic. My analysis has so far, failed to show any statistically significant effect. Could you share/post any analysis that shows the cannibalization effect ?
Toni says:
Touché. I'm glad you brought this up, George. We share your view in our small agency and handle our clients' budgets as they were our own. Hiding the truth is simply bad business so we always report brand and non-brand performance separately. I'm quite amazed of how few agencies actually do that. That being said, I agree with Mrs Gordon that bidding on own brand terms is justifiable and vital if there's any competition - but it should certainly be reported separately!
Thanks Sid, Working on a post on this topic right now. I think the bottom line is the degree of cannibalism (or reinforcement) depends tremendously on the vertical, the business model, and the name of the business itself. Manufacturers who essentially compete with their distributors for prominence will probably probably find the ad traffic to be largely incremental, online pure-play retailers with brand names that might be search terms might find the same. Very unique business names with tremendous brand equity (eg Geico) probably have much more cannibalization because there are likely no other competing ads to show, and there's not much question as to what the user seeks. I've yet to see a really careful study looking at a variety of business models to see what's what. We're interested in playing with the data, but are having a hard time talking clients into putting it to a thorough test (the handful who don't want us to advertise on their marks don't want to spend the money on a test; the ones who do, don't want to risk lost sales). But we're working on 'em. George
George, What do you guys see the role of brand keywords playing in regards to first, last and multi-click attribution? Does non-brand PPC look undesirable because it starts the purchase process but the brand eventually closes it? For those who give credit to the brand in this example then non-brand PPC will indeed look miserable. If you factor the non-brand PPC into the equation and give it some measure of attribution (anything greater than 0%) then your non-brand PPC may look better. It won't improve your overall results (the pie is still the same) but it gives you a clearer picture of the overall health of your program. A fairer way to get senior management to increase your budget is to see what keywords are generating a lot of first-click traffic that ultimately close on other PPC terms. If you're under-bidding on terms that are more prone to give an assist (top of the funnel terms?) then maybe you're selling your entire PPC program short. Thoughts?
Mary Bowling says:
Very insightful post, George. I think it all comes down to who are you working for, ie who is paying you. I can't tell you how many times I've been told to optimize a site to include what I call the CEO list (terms the CEO insists on ranking for) when other terms would clearly bring the website more targeted and well-converting traffic. All you can do is give your best advice, make a solid case for that advice and then do what "the boss" wants you to do.
The bigger moral question is why brand advertisers should have to pay for their brand terms in the first place? "We noticed a major search engine would not return a large airline’s homepage when the airline’s name was given as a query. It so happened that the airline had placed an expensive ad, linked to the query that was its name. A better search engine would not have required this ad, and possibly resulted in the loss of the revenue from the airline to the search engine." How do you think Google would respond to that last paragraph? Because it turns out to be a direct quote from Larry and Sergey's original Page Rank algorithm paper (The Anatomy of a Large-Scale Hypertextual Web Search Engine - http://infolab.stanford.edu/~backrub/google.html). A longer version of the quote is here: http://blogation.net/2008/06/10/we-expect-that-advertising-funded-search-engines-will-be-inherently-biased-towards-the-advertisers-and-away-from-the-needs-of-the-consumers/. We SEMs buy brand terms because we have no choice - it's the right thing to do by the client. But really no one - not us, nor the search engines - should be profiting from brand keywords!
Mary, that's a great example. As Jim pointed out, we all have a responsibility to be candid in our advice, but if that advice is rejected then we're obligated to fulfill our contracts and do what the client asks. Thanks for your comments.
Glenn, it's a great question. The stats modeling methodology we've built for our Attribution service recognizes the distinction between navigational brand searches and competitive non-brand search in dividing up credit. Since the earliest days of our business, long before our attribution work, we've given preference (and credit) to non-brand searches in instances where both brand and non-brand ads were touched -- regardless of the order. The thinking being that if I found the Wine Enthusiasts site while searching for Riedel Wine Glasses, then left, then came back through a search for Wine Enthusiast, pretty clearly the non-brand ad deserved the credit. On the flip-side, if someone comes to your site on a brand search, leaves, and then searches for wine glasses, even though it's pretty clear they knew you before they clicked on the non-brand ad, the fact that they searched for the product, not you, arguably indicates that they're throwing their business up for grabs and are willing to shop elsewhere. We think the non-brand ad deserves the credit in either case, and since 2004 we've given credit to the non-brand ads in these cases. Now, even for well-known brands customers with both brand and non-brand cookies on their browsers represent ~ 4 - 8% of orders. The highest we ever saw was 12% for a company with >50% market share in their industry. So, while worthwhile to get this right, it generally doesn't greatly change the picture.
David, thanks for your comment. As I mentioned on an earlier post, it does remind me of the crazy days when advertisers paid AOL to put "AOL Keyword MyBrandName" on MyBrandName's own marketing material! I agree with you in principle: companies have already paid for the brand-equity they enjoy, and paying again for the privilege of cashing in on that brand loyalty grates. That said, the engines are not obligated to send traffic to anyone. They're paying for the servers and bandwidth to produce search results, so I don't think we can tell them how to allocate space on their webpages. I have less patience with agencies who dupe their clients into paying them on a rev share model (counting sales on brand), or covering up the distinction between brand and non-brand performance. Those folks should be jailed.
Toni, thanks for your comments. It sounds like you folks do business the right way!


Check out what others are saying...
[...] Moral Dilemmas in Agency Work, Rimm Kaufman [...]

Leave A Comment