SEM Thoughts : From an Abacus to AI and Beyond

November 3rd, 2008 by bparshall

I’m going to dedicate this posting to the billions of dollars being wasted on SEM marketing, and those who unwittingly contribute to it. Even in ideal market conditions you’d think this would be a concern to any major advertiser, but given recent events now in particular is the time to re-evaluate the effectiveness of your SEM advertising dollar.  Of course I’ll be discussing WebTrends Ad Director specifically, but other similar technologies exist, such as Efficient Frontier.  And while I could dig into some biased opinions on why Ad Director is the technology leader in this space, it’s more important to me (for now) that readers understand the principle of automated SEM optimization, as a stark contrast to bid management. As is a tedious habit of mine, I’ll start with some history…….

Let’s go back to the abacus (just for a sentence!). Although very simple, just a wooden frame with beads that slide along wires, this device gave humanity an addictive first glimpse into a reality of machine-aided calculations. Move forward a few centuries—past the invention of the hand-held calculator and parallel computing, a computer’s victory over a world champion chess player and successful forays into artificial intelligence—to the present, where technology is freeing humans from the drudgery of repetitive tasks. With relative confidence we can now relinquish control of everyday things like balancing our check books and navigating through city streets, for example, and can shift our focus to activities requiring more non-linear intuition and decision-making.

For the business of search engine marketing, however, this balance between humans and machines has yet to be found. The industry remains dominated by a constant cycle of testing, reporting and analysis, each step of which relies on humans to do the majority of the heavy lifting. Yet the industry continues to grow. According to a recent study by JPMorgan, search engine marketing will reach nearly $30 billion this year, most of which will come from advertisers paying upwards of $50,000 per month and managing tens of thousands or millions of unique keywords. This growth in spending and competition places more pressure on the search marketer to be as efficient as possible. At the same time, search engines continue to add new segmentation and targeting options, resulting in marketers being faced with almost unlimited choices in the number and complexity of variables for their paid search campaigns. As many companies have millions of requisite keywords, these new demands can quickly overwhelm even a roomful of SEM specialists.

Why then, in our current technologically advanced world, do we search marketers still fail to fully enlist the staggering computational power of machines? This failure stems from at least one of the following three factors: an unawareness that the technology is out there, a fear that this technology will cause people to lose their jobs, or the perception that, by letting computers take on some of the processing, their human operators will lose control.

First – the technology does exist. For years, solutions have automated the bid optimization process. Here at WebTrends, our SEM optimization product Ad Director has been significantly out-performing manual input techniques and so-called automated bid-management tools.  [“Automated bid management” -- surely one of the great oxymorons of the information age.]  Ad Director uses self-learning to automatically test the performance of millions of combinations of unique keywords, landing pages and ad creative, and make tens of thousands of bid changes daily.  For you fellow math geeks, the product uses experimental design techniques to evaluate the independent variables (e.g. bid rate, channel, landing page, etc.) that contribute (or not) to the desired outcome (e.g. conversion rate, ROAS).  Perhaps in a future post I’ll go into the details.

Second – automated solutions do not exclude humans from the search marketing process, they merely shift the human focus from repetitive tasks to higher-level processes such as strategic planning and accounting for external variables unique to the business— skills that computers don’t possess.

Finally – it is precisely this need for humans to retain control that led us to introduce the ability to override the automated system to achieve goals outside of normal parameters. For example, a company may wish to set minimum bids on certain terms, such as brand terms, regardless of the observed return.

As we look toward the future of our industry, where the bar for search marketing is only going to get higher, the current SEM methods will no longer be viable. In fact, recent research performed by WebTrends has suggested that more than $4.5 billion is wasted annually as a direct result of how the current manual and bid-management campaigns are ran.

To be blunt, if you spend more than $50,000 per month in paid search and use a bid management tool (either on your own or through an agency) you might as well be shoveling cash into a furnace. Forward thinking managers must turn to automated solutions, re-task personnel to strategic functions, and as a result will out-perform their competitors and realize substantial efficiency gains during hard economic times.

Let me know what you think.

Tags: , , , ,

5 Responses to “SEM Thoughts : From an Abacus to AI and Beyond”

  1. Jeff Cram Says:

    Good points Barry. I was also glad to see you qualify the size of account at the end:

    “…if you spend more than $50,000 per month in paid search and use a bid management tool (either on your own or through an agency) you might as well be shoveling cash into a furnace….”

    The technologies are indeed there, but the starting point (price) for many organizations is still far too steep. I’m anxious for this technology to reach the mid sized firms that have the same or greater) strategic needs as the ones that have the large spend.

    Jeff Cram
    ISITE Design
    http://www.isitedesign.com

  2. Rathna Says:

    Excellent read. Thank you for sharing this excellent article.

  3. Michele Warther Says:

    For some additional insights on how Ad Director works, you can also visit the “Waste-o-Meter” site. See how much money has been wasted since Jan 2008. Visit the site here >>
    https://www.webtrends.com/ms/waste-o-meter/default.asp

  4. cory huff Says:

    If your company has the money, why not just bid to be up there 24/7? You can call the time branding spend and justify it that way.

    Also, what about small businesses that perhaps cannot afford software like mention, or don’t have the proper training to make it effective? In other words, how does this affect a company that doesn’t have $50k to spend on SEM each month?

  5. Barry Parshall Says:

    Hi Cory,

    As I see it in any economic condition, let alone the one we have now, and no matter what your marketing budget, you want to spend responsibly. Bidding up a few brand terms for strategic purposes is fine, but I doubt any organization can bid for top position for 10’s or 100’s of thousands of keywords.

    $50K was mostly a made up number, but it is indicative of the sweet spot for solutions like WebTrends Ad Director. Despite my apparent animosity for them, bid management tools are perfectly fine for small businesses or any organization with a reasonable number of keywords and a modest budget. There comes a point, though, where the amount of ad spend coupled with the number of keywords (and ads, and possible landing pages, etc.) becomes too burdensome for a human with a bid tool – you either have to throw more people at the problem and/or accept sub-optimal results. $50K/month is clearly above that threshold where automated optimization is needed, though surely companies spending less would benefit too. I suppose in a future post I could go into greater detail on determining when an automated solution makes the most sense.

    I hope this helps.

    - Barry

Leave a Reply