Let me know what you think
http://connect.icrossing.co.uk/paid-search-higher_902

I fully stand behind everything my esteemed former colleague Mr. Beeston has said about how AdWords *can* be modeled, understood and that knowledge used to automated *and* bring significant lift to campaigns. The fact is, EF wouldn’t have gotten to the position where they are today without pleasing the majority of the $100K-$5M/mo spenders they manage; pleasing them means driving both lift and strategic control over SEM outcomes.

That said, I’ve had my own epiphany over the last few weeks, which is that best practice really does lie between automation and manual management; sounds obvious to say that given other comments on this thread, but I think it’s worth going into in more depth.

I believe we’ve already covered what it is that a human approach can’t do for modern-day campaigns = large keyword sets, multiple search engines, multiple metrics. What’s equally interesting to cover, though, is what automated approaches such as EF’s can’t do today, and how that will shape the development of keyword management platforms going forward. ‘Bid management’, incidentally will end up being synonymous with the Model T Ford as pertains to search.

Let’s talk about a couple scenarios to illustrate what automated systems such as EF’s can’t do and why other solutions and/or evolution are necessary:

Scenario 1: Large retailer buying 100K+ keywords, spending $500K/mo, inventory fluctuates signficantly for 30-40% of products. The issue here with a cross-keyword data modeling approach is going to be that of not having visibility into inventory, and thus spending money sending traffic to a product that is out of stock starting today and for the next four days. Unless inventory levels change in a predictable fashion (any retailers care to chime in?), a data modeling approach is going to mess up. So what’s needed here is a combination of data modeling *and* integration with inventory systems. But let’s be honest here – for 95% of retailers, that integration with inventory systems (which *must* happen soon) is going to happen via their web analytics platform, not their SEM tool, and so the role of a central analytics platform in keyword mgmt will become critical.

Scenario 2: big travel site w/low conversion rate (CR) relative to peers. The problem with EF’s data modeling approach is that it can tell you what keywords and ad groups are converting and not converting, but it can’t tell you *why*, yet that’s the most important question. If your CR only allows you to profitably bid to avg position 6.5, for example, you’re gonna get crap volume unless you improve your conversion rate, and there’s not a damn thing EF can do to help you other than free up your time to focus on it.

Great, so now what? Are you supposed to engage yet another vendor to help you with systematic, thorough CR optimization? That’s an option, but wouldn’t it be better if you had web analytics, SEM mgmt and A/B testing in one integrated system? I think it would, because
a)workflow efficiencies would allow you to get a lot more done;
b)conversion funnel data relative to your search adgroups could be easily analyzed and acted on;
c)A/B testing could be conducted *as a direct result of SEM campaign analysis*

This is what marketers want, and they want this level of SEM/analytics/testing integration because best of breed only works if you’re a huge marketing dept and even then multi-vendor ’solutions’ kill productivity. I’ll bet a dollar against a penny to everyone on this board that 5 years from now we’ll look back on <2009 and wonder how in the h$%l online marketers got *anything* done before integrated online marketing management platforms.

As the above scenarios show, data modeling alone is not enough. There must be a conjoining of data modeling, scalably applied human oversight and integration with central web analytics platforms over time for SEM to evolve where it needs to go. Conversion funnel analysis for SEM must be integrated for its applied output to be scalable across an organization. Multivariate testing must be integrated with both SEM mgmt & analytics for the hundreds of tests each organization *should* do annually to actually happen. Lastly, web analytics must be the conscious eyes providing the de-duped, channel agnostic, paid and unpaid unified view of everything.

Why do people who have accepted the need for transparency fail to acknowledge the ability of the tools designed to provide this very feature?

Q: How can one account manager successfully deal with 100,000 keywords across a magnitude of products and product groups, with seasonality effecting the level of interest, whilst fighting off the competition day and night? One man can not cope, nor can a entire office of account execs, without incurring huge overheads and massive communication breaches. Trust me I know more than most.

A: By forecasting the search landscape you are one step ahead of the competition.

Ask your self another question:-

Q: Are you doing anything different to your competitors? Or are your competitors doing something different to you?

The search landscape is warfare – plain and simple and as with every civilization; it is in times of warfare that new and exciting technologies emerge driving our capabilities forward.

Another point if you enter a bidding war for your target terms, why bother bidding at-all, unless you are achieving your optimum yield? None of us are bidding for the sheer fun of it.

Of course technology will never replace the human interaction, landing pages, copy creative, engaging creative, off line campaign drivers etc.,

I say if someone thinks they can do it and and possesses a proven track record of relative achievements wheres the problem? it is really a no-brainer.

But this is of course one mans opinion.

Would it surprise you to know that math-based investment firms account for over a quarter of all the world’s daily securities trading, or that competition for quant jocks is so heated that investment houses don’t even wait for these guys to graduate to start recruiting them from India, Russia, the U.S. and other countries?

My colleague Jonathan is, in fact, making claims that seem far-fetched to many of you. It will take us all of 5 minutes, though, to prove that we can effectively model impressions, clicks, effective CPC, resultant bid position, revenues and margins for our clients within Google’s opaque, QS-laden AdWords system, both within each keyword and across large keyword sets.

I’ve been spending the last 4 years getting advertisers to *believe*, so to speak, and it’s no easy feat. For starters, try convincing an advertiser who has no pain that they will, at some point in the future, hit a ceiling in their PPC campaign beyond which they can’t extract additional revenue and/or margin. Human nature is such that most of us (including, scarily enough, most people at publicly-traded companies that depend on online marketing for growth) need to feel pain before we’re ready to step outside of the box.

The reason Mr Beeston and I spend most of our time dealing with advertisers spending $100K-$1M+/mo on ppc is that those large ppc advertisers – and not the recent or even mid-size ppc advertisers – are the ones who are at or near that ceiling.

They know from experience that the limitations of manual keyword management can hurt a *business*, if not the SEM/SEO consultants who get paid more for the manually-intense ppc mgmt of that business.

They also know from experience that lack of strategic visibility into what efficient ppc campaign operating points are possible eventually results in a big miss and a 10-30% hit on their stock price. THAT’s when all the religious wars and marketing B.S. go out the window and the CFO steps in demanding the actual forward visibility he was told search and online marketing would bring.

Who can argue that marketing is quickly becoming a never-ending math calculation? Sure there’s the work humans will forever be better equipped to handle – ad copy creations, landing page creation, knowing what bets to make, etc – but keyword bid management is becoming a world of algorithms both at Google and at the advertiser.

And for the naysayers about EF’s technology, then I’m happy to respond to your calls of BS. Perhaps this isn’t the right forum, but feel free to get in touch: jonathan dot beeston at efrontier dot com.

But there are so many factors involved (not just your own but those around you) that are almost impossible to calculate and are constantly changing, which means these tools will ultimately still always be just be that – an estimate. Which is not saying they are without use (as I mention above).

The way you have expressed it, makes it sound a bit like this particular tool has it nailed down to a simple bid to position model – although I haven’t used the tool, I am willing to bet that isn’t the case.

Moving on – What is really going to make 3rd party bid management solutions stand out?

Most already offer a simple integrated system for monitoring and management, but I think a lot of advertisers and agencies want more. I certainly think the future of 3rd party bid management solutions is more about delivering ROI / CPA rules that really provide significant performance increases over manual optimisation.

Will Google and Microsoft try and develop these further and intergrate? Or will we have to rely on the 3rd party independant offerings / in-house systems?