We have yet to find any bid tool that really works well with Google. 1P exports are nice, except we found that if we want to provide accurate click and cost data to the client, that has to be pulled directly from the engines.
You can use it with Atlas tags or 1P tags. The benefit of using Atlas tags is to prevent double counting of sales if you are running banner campaigns through Atlas DMT.
I'd suggest using straight 1P tags.
On Overture it usually makes sense to lower your bids at night to save money. On Google it makes sense to turn your keywords off, otherwise if you just lower them your CTR will take a hit and it will probably cost you more per click when they go back up during the day because your average CTR will probably have dropped. Not sure if you can turn off keywords in G within 1P at night and turn them back on in the am automatically.
I don't think G or OV passes the actual click cost at the time of the click back to reporting tools, especially with Google. Unless you have very few clicks you never know how much you really paid for a click because it is changing all the time. It doesn't seem like the reporting tools do either.
It takes a while to learn how to use the system. It will probably help more with ROI based bid adjustment on OV than G. Unless your keywords are more than $0.50 per click on average on Google, I'd just use a good reporting tool and am not convinced that the bid management stuff is worth it at least if Google is the primary reason to use one. If bidding is fast and furious in your market, then things like jamming on Overture and some of the other rules may be valuable.
Add up the projected cost of using the system, the time it will take to learn to use it and adjust it on a regular basis. If the efficiency it squeezes out of your campaign is greater than the cost, then its a good investment.
KeywordMax is similar & cheaper, but again, not sure if any of them can really do much good on Google bid management, though they all say they can.
IMO, better creative, more highly targeted keywords, and better AdWords copy are more important than bid management.
IMO, better creative, more highly targeted keywords, and better AdWords copy are more important than bid management.
I second that Skibum.
Is it really necessary to have something actively manage those keywords or just a bunch of hype to make you think you need to spend money to manage them? Definitely spend the money to track but evaluate the management part more carefully.
I'll wager that your idea of bid management is rules-based or portfolio theory constrained by human capacity limitations. Even if you consider Atlas or Did-It to be state of the art. the former methods are poor man's bid mgmt.
I'm wondering how many of you (professionals) use bid management software and for which goal.
Has anybody experience with the rules which can be set with OnePoint?
Thanks.
I'll wager that your idea of bid management is rules-based or portfolio theory constrained by human capacity limitations. Even if you consider Atlas or Did-It to be state of the art. the former methods are poor man's bid mgmt.
Shorebreak - Any chance of expanding on what you mean by this?
Are you saying in your experience clicks at night (or early morning) are less valuable?
I agree on your post about most sales occuring from 6am EST to 2am EST and I would like to add another spurt from around 6-8pm EST.
As far as turning your keyowrds offline overnight I am very skeptical for a few reasons. There really is no way of measuring the quality of traffic during these hours. Currently Overture and Google do not allow you to run cost reports for specific hours of the day so its quite impossible to do any type of useful analysis without hourly cost data. Also there is no way of knowing whether or not the night traffic is browsing in the evening and buying in the morning. I would find it more of a hassle to turn these campaigns on/off on a daily basis.
If you have any input on how this could be measured I would appreciate it because it has definetly crossed my mind.
A computer program doesn't know which you're bidding against. Having an indirect competitor in the spot above yours is far less meaningful than having a direct competitor above you.
My theory is that the user starts with a search term, "widgets". But what is it they really want? To buy widgets? Widget repair? Parts for widgets? Consulting on widgets? Books on widgets? A job in the widget industry? Etc. If they're looking for widget parts, they'll skim down the listings until they notice one that looks appropriate for buying parts.
Another factor is that much of the low-hanging PPC fruit is on low-volume searches. If you're paying on a per-keyword basis, then you're incurring a high degree of overhead on such terms.
If your goal in bid management is stay on top of direct competition, though, I can see you point. But isn't the goal to be the one with the most transactions and the most profit?
... the same type used on Wall St to trade equities.
In equities markets, all the competitors are direct and there are no differences in offer other than price (i.e., there's no ad copy).
If your goal in bid management is stay on top of direct competition, though, I can see you point.
That isn't the goal.
But isn't the goal to be the one with the most transactions and the most profit?
It is. That's why I don't turn it over to a computer program that's unable to evaluate the key variables.
Once the site, the keywords & the messaging are in place, bid mgmt would probably be much more efective.
A lot of what I manage has a heavy mix of indirect competition.
Cline, a good keyword management system should be able to keep track of, and optimize against, all the key cost, revenue, profit and branding variables that are important to the advertiser. I hear the type of statement you've made all the time - it seems like almost *all* the keyword management systems out there are incapable of reacting to the data in the same way a human would, and force you to translate business objectives into static rules that quickly become stale and ineffective.
However, if a system *can* track all those variables, integrate them into a model that encompasses past and current performance of the whole keyword set, and take the appropriate actions based on the model and the changing data, then there's a very strong argument for such a system to be able to do a much, much better job than any number of humans could ever do. Case in point - if you have 10,000 keywords running on Google & Overture and there are 5 bid positions of meaningful volume on each keyword & each SE, then you have (2 x 5) to the 10,000th power set of calculations to do every hour of every day in order to be aware of all the different bid choices you have.
The point is no human can do all those calculations, which is why advertisers and agencies trying to keep humans in the middle of it all are missing the boat, IMO. I DO AGREE that most bid management systems end up being of little value because they force you to hard-code rules and can't react to a dynamic marketplace, but there are systems adapted from the world of quantitative hedge funds on Wall St and which do what a human would try to do, only better because of superior modelling and regression analysis capabilities.
1. A large proportion of indirect competition
2. Low volumes of traffic, clicks, and conversions
Because of #1 the response curves are going to vary substantially due to market conditions difficult for a statistical model to identify without large amounts of historical data. Because of #2 there's not going to be much data to plug into a statistical model without large amounts of judgment being inserted into the model in order to pool data.
I'm not sure I understand what impact the competitors have on decision-making for your own ppc campaign; it's Monday, though, and I'm probably still slow from the holidays.
On your second point, I hear you. We've been trying to address the sparse data issue by applying algorithms that can cluster data from multiple keywords. Each keyword in the long tail may have sparse data, but there are often meaningful correlations that can be made amongst those keywords and which allow us to bid intelligently and efficiently. That sparse data issue is one reason why getting several months' worth of historical cost and revenue data is so important prior to putting a mathematical system in place to manage bids. That historical data helps build an initial model so that you don't have to wait for data to come in to have an initial model that accurately describes how all the keywords perform, not just the high-volume ones.
While clustering data from common keywords is an interesting theory, we found that it doesn't hold up in "real" world analytics. It also lends to the fact that some subjective analysis must be applied.
What you refer to as "Sparse Data" keywords, really shouldn't be evaluated on a ROAS/CPA basis, until they accumulate enough clicks/cost to warrant profitability evaluation.
We have experienced that these keywords don't generate enough cost during short time cycles, to justify evaluation, before some type of threshold is met. (We experimented with clustering for several months, and found that one keyword will perform dramatically different, from another common keyword, with no logical explanation. Anomalies occur in data samples, that is a fact of life.
What we have had success in "Sparse Data" samples is to build our algorithm to push keywords to garner more traffic so that a proper data sample can be evaluated against target goals for ROAS and/or CPA.
This model has been very successful in both variable CPC paid search campaigns (Overture/GAW), and Shopping Comparison Engines (BizRate/Shopping.com)...
As many eCommerce advertisers may have several thousand products and also several thousand keywords, an automated system can increase MaxCPC or position across the entire set of keywords/products, and perform real time evaluation against target ROAS/CPA.
Most of our advertiser base has experienced an increase of up to 50% in ROAS, while still optimally decreasing CPA (Cost per Sale/Action/Lead).
Our 3+ years of "real life" experience has been, keep the model simple, but make it scalable to handle unlimited number of keywords/products.
Let's assume we provide widget consulting. Consider the following 2 cases, where everything else is equal, except for the competing offers.
Case 1
8 firms offering widget consulting
Case 2
- books on widgets
- new widgets
- widget repair
- widget parts
- used widgets
- jobs in the widget industry
- class action suit for widget injuries
- your offer for widget consulting
Would you manage these two cases the same?
My theory is that who the competition is and what they are doing *is* relevant. Consequently I'd pursue different bidding strategies between the two cases above, ceteris paribus.
Other factors I take into consideration are the competitors' bidding behavior. For example:
- Does their ad belong in this space?
- Is their bidding rational?
- Are they new?
- What is their financial situation?
- What is the strength of their offer?
Based on this information I may decide to pursue a bidding strategy considerably different from what the numerical data would indicate.
For example, if a new competitor appears who is using an irrational bidding strategy I may counter with a bidding strategy designed to minimize the competitor's profit and maximize their psychological frustration, and do so at considerable short term expense. Often what happens is that the competitor eventually gives up on PPC and goes away.
In one case I know that I drove a major competitor nearly bankrupt (a mutual supplier confided to us how badly the competitor became in arrears -- so much so the supplier cut the competitor off). There was a cost to doing this. It trashed my client's profitability for a couple months (n.b., this strategy was of course executed with the full knowledge and support of the client). But the result is that the competitor's business was weakened and as a result we now have greater market share. CTRs are up; CPCs are down.
I see how your approach might work well in your space. Our clients typically spend $200K+/mo on ppc and have tens of thousands of keywords (or more), and so a strategy based on anything other than their own business goals won't be optimal (unless gloating at a competitor's loss is one of the goals).
I'm of the opinion that there's far more to be gained by focusing on your own game than there is to be reacting to competitors' moves. Of course, our clients (who typically have 10K-1M keywords) tend to have 20+ significant competitors each, so at some point competitors' particular strategies coalesce into 'the market'.
The other challenge with your approach is that is won't ever scale. You could do a great job with several clients, but you couldn't possibly, IMO, apply your approach to hundreds of clients and millions of keywords. We all know how hard it is to find SEM-experienced people these days. If your growth strategy involves needing to find SEM-fluent people who you then train to react to competitors the way you do yourself, then you'll quickly be in the rear-view mirror of the SEM firms that automate their businesses. There are simply too many calculations to do with large amounts of keywords; in your approach, my opinion is you're either leaving too much volume and/or margin on the table in order to spite competitors, or you're unable to scale your business.
My opinion is just my own, and nothing more; it definitely sounds like you know what you're doing better than 99% of folks out there, and I speak without any knowledge of the particular space you're in, so I may be making false assumptions left and right. I'm just convinced that the only way for an SEM business to both scale and maximize returns for clients, is to automate keyword management and not worry about anything other than their own clients' business goals.
I do, however, take a broader perspective on what my clients' business goals are. One key thing I've observed is that the profitability of PPC advertising is closely related to the number of competitors, especially direct competitors. Fewer than around 4 there's not enough competition to drive up bids, making it highly profitable for the advertisers. More than 8 and competition becomes intense. Bids get driven up high. Profitability becomes difficult for some of the players. Therefore, short-term strategies that convince new PPC players to leave PPC, or even to just bid less aggressivly, can have substantial long-term payoffs. It's not about gloating or spiting competitors. It's about achieving strategic business goals.
You're correct that my approach is seriously difficult to scale. There's quite a limit to what can be delegated with this approach. So be it. I'm not in this to become the biggest SEM firm around. I'm interested in making my clients as profitable as possible. I figure if I do that my profitability will take care of itself.
There's lots of business in this space, and a lot of niches. Marketing strategy is my forte. I'm happy to leave the massive number-crunching niche to others. Clients with $200k/mo spends tend to experience different market conditions from those with $5k-50k/mo spends. One size does not fit all.
