Human Vs Automated PPC Management

Who has the edge when it comes to paid search bid management - human decision systems or computer decision systems? Dixon Jones, director at Receptional looks at the two approaches.

Google Adwords and Overture (Yahoo's equivalent) are the two of the most successful new advertising mediums of the modern age. They give an impressive and (for a change) measurable return on investment. Administering pay per click campaigns, on the other hand, is complicated. Doing it properly in whatever way you define "proper" means that there is a cost associated. Leaving it to the Googles of this world, or agencies trying to cut the mustard on commissions only just isn't as cost effective as spending some of the campaign spend on enhanced bid management services. In other words - you get better returns by paying a surcharge cost per click for someone - or something - to manage bidding strategies on your behalf.

There is little doubt that in some areas of this process, automation is the way. Tracking clicks and key phrases from spend through to sale would be a crazy task not to automate, and the traffic providers themselves have taken the shine off of most automated bidding tools by automatically ensuring you only pay a penny above what you need to attain the position you deserve or pay for. But the real cleverness is in defining and implementing a bid strategy. Who has the edge - human or machine? Well - it depends on which human and which machine. In the final analysis, Kasparov still beats the best chess computers, but mere mortals would get thrashed by the average chess program. Here are some of points of difference between human and machine.

1. Choosing your keywords: I think here, a computer can certainly develop a keyword list faster than a human, but the quality needs a serious sanity check - both from the optimiser and the client, for different reasons. A computer can give you thousands of search phrases that your customers MIGHT type in, but is unable to make a qualitative judgment as to the relative value of each phrase to your business. This means that the computer can create a faster list, but the human has to cull the list manually and then give every keyword some kind of value to the client as to whether they think it can convert well or poorly. Alternatively, every search phrase could start with an equal "value" and then you could let the computer decide which ones actually work - but there is a hole in this theory, as we will see later.

2. Writing the creative: Definitely humans win here. There are a few mechanical tricks - like matching the title of the advert to the actual search phrase of the user, but these are built into Google Adwords and don't require a management tool. Automated management tools are so bad at understanding creative, that this is the largest failing for them, with unexpected consequences. For example, we compared two creatives for a block of luxury apartments in London. Advert A had a high click through rate and advert B had a low click through rate. However, Advert B was far more targeted. Deleting the advert with the highest click through rate is counter intuitive to a rules based machine, but quite obvious to a human understanding the client's business. This failing of the software solutions to interpret creative can (and we believe does) occasionally totally undermine the tools whole raison d'être. In another opposing example based on conversions to actions rather than click through rates, the creative may bear no relation to the content on the landing page. In this instance, the correct action is to revise the copy on the landing page. However, all the software can determine is that the phrase is a low converting phrase. The software in this event will reduce the budget spent on this phrase or drop it completely to bring the ROI on the phrase into line with other phrases, when the correct action was to change the copy on the landing page to improve the ROI from visitors arriving on that phrase.

3. Defining the strategy: Here the human has to take the first step but a machine can react quicker to environmental factors. Do you bid high or low? Do you need to change you bid price by time of day? Do you want to factor in the bid price your competitor is showing? Do you want to bid just above or just under your competitor's bid price? Is the search phrase likely to be a high or low converting term, and how will that affect your intended strategy? Initially, the human makes these decisions. Some strategies do not need software solutions to maintain "best price for 50 pence a visitor" is deployed at the coal face just as easily as through a software system. "Only bid the minimum outside working hours" is, however, best achieved by machine. The problem for the machines and humans alike comes when assessing if these strategies are optimal or can be improved upon. It is here that the machine and the human will again take different routes. A machine can only stick to the strategy it is given. However, if the overall strategy is "try to optimise the bids so that every search phrase converts to sale at the same overall ROI" then only a machine can keep up to date with this challenge. In theory the machine wins - but in practice there are two flaws that machines fall down on. The first is that the phrases that convert best are usually the niche phrases, where less people search on the phrase. This means that the data of impressions and click through rates does not have enough volume to be statistically significant. You may be surprised to learn that 50% of all searches on Google in a given month are never repeated again that month according to Google engineers. If that is true for the whole of Google, what hope does the minute subset that is your campaign have to make statistically sensible decisions? None. A fatal flaw. The second problem again relates to the machine's inability to diagnose the root of the problem for a poor ROI phrase. The machine has to make the assumption that all creatives and landing pages.

4. Measuring the success. One up for the machines here. But surely we digress, both in technology and in focus? Any mid-range tracking solution like www.hit-counter.net will ensure that you can track users who commit to a sale or call to action all the way back to the original search phrase they used and the campaign they clicked through on. With some more thought, the systems can tie these calls to action into revenue net worth, as can Google and Overture's own tools.

5. Reporting results. Now at last we have a clear winner for the added value tool. The problem facing the human is that without a tool to aid them, they need to look and interpret data from numerous interfaces you build up a single picture that can be presented on a single spreadsheet for management to review. A bidding tool that works across multiple platforms has the ability to also report across multiple platforms, saving potentially huge amounts of time aggregating and unifying data into a single report.

   In the final analysis, there is clearly a place for bid management tools, but for the people closest to the coal face, it appears that human decision making processes, supported by an integrated web analytics and ROI tracking system that can work across multiple campaigns and PPC suppliers is the most effective. But this only works best if the Humans have the not inconsiderable skills needed to interpret the subtleties of pay per click management and Creative copy writing. Look out for Google Adwords qualified professionals and Overture accredited SEM companies for reassurance in this regard

Back 03.02.2006.