What is Truly Automated, Adaptive PPC Bid Management Technology?
If you invested in bid management software within the past few years, chances are high that you were probably duped in one key area: automation.
Now, before you think I’m about to split hairs here…let’s walk through a few basics to make sure we’re on the same page…and…wait…who am I kidding? This might make you mad – or it will make ppc software companies mad so I suppose it really might split a few hairs.
The definition I think fits best for what I’m talking about is this one:
“the condition of being automatically operated or controlled; ‘automation increases productivity’”
My main argument in this post is simply this: One should have little to no adjustments to make themselves if a software solution claims to be ‘automated.’ One would simply have to provide any business metric (or combination of business metrics) – revenue, cpc, cpa, ROAS, margin or anything really…and the solution/software provider should customize their automated technology to meet or exceed those goals.
Myth: Most PPC Bid Management Software IS Automated
I spent a lot of time in Q1 this year researching over 2 dozen ppc software companies that provide bid management functionality within their software – or it’s the core to their solution.
Almost every single company claims to:
- Automate bid management entirely with pre-created rules that you choose to set per keyword/campaign/adgroup/whatever (in some cases calling them ‘algorithms’ that bid to position, cpc, etc)
- Give you full control to change bids or create custom rules for automated bid management
- Automated portfolio based bidding that is totally controlled by you
I dare you to go look at every ppc bid management solution out there and find one that does not claim ‘bid management automation’ in some way. I’m convinced that you won’t find a single one that is not claiming ‘automation’ as a feature/benefit to their solution…
A company that claims automation, while putting all of the control into a professional paid search manager’s hands, can’t really be considered automated at all!
Well…answer me a few questions…and you’ll get the idea…
Can a normal professional paid search manager understand the nuances behind advanced predictive statistical models and create/apply bid rules that match those models with any significant degree of confidence?
Can said paid search manager understand the algorithms and math behind the programming that manages the (custom or not) bid management rules/methods?
Can said paid search manager rapidly compare and model today’s results from yesterday’s results, same date/last year’s results, same date/last month’s results, and other date OR day specific results relevant to that day’s bids and adjust bid rules accordingly – even through sales, vacations, 24/7?
Now that you have answered those questions honestly…let’s run through a scenario on a very very small scale:
1 non-competitive (let’s say…5 advertising competitors), uber low search volume, location based keyword in an ad group
2 text advertisements in the ad group
$1 to $2 bid range
And we’ll have to agree on a few things:
Ad position is irrelevant to conversion rate
Each competitor has the same cpc range
Goals are different for each competitor
Each competitor tries to day part by day or week or whenever they feel like manually adjusting their bids
Ok, there are probably a few things I am leaving out…like the dynamic fluctuation of competitor’s bid patterns and consumer’s search patterns; but, I want to run through the basic math of the sheer number of variations to consider in this scenario where I believe manual bid managers can not accurately predict results and implement changes within a decent amount of time day in and day out:
1 keyword x 2 ads = 2 tests, right? You have one keyword with one ad…against the same keyword and the other ad. Of course you’re still following, because that’s easy!
Now…because we have 2 ad tests running in a non-competitive environment, it might take several weeks to accrue enough impressions/clicks/conversions/’x metric here’ to manually conclude an ad a winner and write another ad to test against the winning ad. Make sense?
Let’s make this problem a little more complex because you agreed that your competitors are probably changing their bids at regular increments via day parting to some degree. You will probably need to adjust your bids too when your competitors adjust theirs or vice versa. Remember your bid range is $1 to $2, or 100 pennies:
1 keyword x 2 ads x 100 penny bid range = 200 tests…a fairly large multivariate test, wouldn’t you agree?
I am confident that you, Mr/Ms manual ppc manager, will be adjusting bids every once in a while without much consideration to the bid change you made several months ago because you’re typically just concerned with the past day’s/week’s/month’s cpc/conversion performance and the last 4 or 5 bid adjustments you made. And some advertisers seem to get by with acceptable results that way.
Consider this: What if your competitor is using a real automated solution that uses an adaptive technique similar to that used on landing page MVTs. (I really like hi conversion’s illustrations when describing their testing technology. Please replace their word ‘conversions’ with other metrics that could be used in low volume spaces – maybe EPC or eCPC.)
With this adaptive/automated technology on their side, your competitors will beat you to writing new ads more consistently and frequently. And writing new ads should naturally lead to improving their target metrics more consistently and frequently.
If you agree with the above scenario, then why, as an industry in general, do we believe most bid management software companies are really automated? Is it because we have been conditioned to accept the definition of ‘automation’ to be “set and create your own bid rules or use ours and the bids manage themselves but you are always in control?”
These alleged ‘automated’ solutions typically don’t employ predictive/rapid statistical modeling in their algorithms. These ‘automations’ are almost always tied to an ad position, cpa, cpc or other easy to calculate average over a specific (or custom) date range!
Manual control is a common theme among these ‘automated’ solutions. This manual control negates automation given the definition I referenced earlier.
Too much human control of a robot creates a lot of room for errors…especially when one is relying simply on rules or portfolio based solutions that are easy to understand and normally consist of using rolling averages to make bid changes.
Bid management is way more complicated than rolling averages!
Fact: Truly Automated, Self-Learning PPC Bid Management Technology EXISTS Today and it Trumps Manual and Pseudo-automated PPC Bid Management Software 24 Hours/day, 7 Days/week!
In my research I found less than 5 companies offering truly automated, self-learning bid management technology across the major search engines (at the time of my research these were Google, Yahoo! and Microsoft’s properties.)
The case studies found at each of these companies are truly impressive and you get a great idea about the real scalability, efficiency, modeling and metrics that these companies can optimize toward successfully.
A few key characteristics of these companies that offer true, automated, self-learning technology are:
- They employ or work closely with a well qualified team of PhD holding mathematicians and PhD holding statisticians.
- They have to (or should) manage their complex technology with strategy direction from an advertiser’s in-house paid search team or agency. Remember, this technology is so complex that it requires PhDs to help program and develop and improve it…most pro paid search people aren’t even close to that caliber of math because we’re marketers!
- They should NOT replace an advertiser’s team nor should they encourage it. The biggest mistake with bringing in these expensive, high cost automated companies is letting go of the pro team that can and should manage the automated tech company relationship/strategy/creative portion of search marketing.
- The technology is so advanced that it should be highly adaptive/self-learning toward cycles, Holidays, trends, sales promotions, and more. In some cases – i.e. sales promotions – communication is going to be key between the advertiser and the technology company team.
- They should have a fully functioning, transparent reporting system that may allow a very limited degree of control…for example, you might have a few branded terms or a campaign that you absolutely will never ever automate to custom metrics so you can keep it out of the pool of automated campaigns. (Yes, manual ppc management MAY be required at this point if you really really want to do it yourself…)
- They should allow control in the sense that you can still go into your AdWords, Microsoft, Y!SM accounts and change things manually yourself…but they should adamantly request that you never ever do that because it will mess up the millions of constant tests and statistics running through the automated solution and that could lead to massive losses/big problems. If you must do it, provide adequate advance notice to the tech company so they can prepare and perhaps even withdraw certain data based on your manual adjustments…
- Their sales copy is freaking annoying and as vague as possible because they have no clue how to describe their technology due to the dilution of the definition of ‘automation’ by all other pseudo-automated bid management software in the marketplace.
- They’re cost prohibitive for advertisers spending
So what are advertisers spending less than $100,000/mo supposed to do to compete with this insanely awesome sounding technology?
I guess you’ll have to settle for pseudo-automated software…while I might argue that one should go manual to $100k and skip the pseudo-stuff if possible because pseudo-automation leaves so much room for error, bugs and misunderstanding. In my mind it is more risky to adopt pseudo-tech than it is to wait until you can scale to using real advanced tech.
I wanted to cover the solutions I might pick for advertisers spending under $100,000/mo in this post…but I’ll cover a few companies in my next post because this one is getting too long! Time to wrap it up!
What is Truly Automated, Adaptive PPC Bid Management Technology?
The only technology that should ever call itself automated is the one that puts the entire bid/testing/learning process into a position of largely “being automatically operated or controlled”…meaning little to no direct input (bid changes) from a human. One acceptable input would be sharing business strategies, promotions, etc with the technology manager who would ‘move a few levers’ to adjust the technology accordingly. Another might be keyword expansion, writing new ads (the software should auto-pause poorly performing ads based on your business strategy/metrics!), organizing new ad groups, and more.
True automated, adaptive ppc bid management technology does not need bid rules input by a human – it simply needs business metrics to optimize and scale toward. This technology learns and optimizes along the way by performing complex predictive analyses that humans can’t even come close to doing at the same speed.
What do you think? Am I making sense regarding the differences between what I call ‘real bid automation’ versus what most ppc management software companies are claiming as ‘real bid automation’? Do you agree or disagree? Why?
PS – In my next post I’ll cover a list of the truly automated solutions I found + the pseudo-stuff I might consider using and one or two manual ppc management solutions to think about while you’re scaling your way toward the truly automated stuff.If you enjoyed this post, make sure you subscribe to my RSS feed!
Manual PPC Bid Management is Dead
When I started semvironment as an agency* several years ago, I was proud of the crazy results we were able to achieve with manual pay...