StatCave 7: Control vs. Certainty

Why is it that everyone who manages AdWords has some degree of difficulty getting the intended output? There's an underlying compromise, a balance, between two types of desirable account behavior, and it's happening regardless of the type of bidding solution you use, or whether or not you're even aware of it.

Click here to see Veritasium's video on Regression to the Mean.


If you have ever managed ad words, then you've struggled to hit a particular return on ad spend target, that's just fundamental to the entire experience of managing ads. But why is it so difficult? Well it turns out that there's this constant balance between the amount of control you have, and the amount of statistical certainty you have on your decisions. So we'll dive into exactly what causes this conflict between these two competing sides, and some recommendations for how to mitigate that problem.

To start with, we need to look at how bidding works fundamentally, and it's actually really, really simple. You start by understanding how much of your revenue you're willing to spend on your ads, and that is called a cost of sale target, and it's generally shown as a percentage, it's just the reciprocal of your return on ad spend target. Then you need to know how much revenue you're getting per click, that's simple division, and the resulting metric is called value per click. You multiply these two together, which is a very simple bit of math, and you find out exactly what you're intended target CPC ought to be.

From there you have to pick what layer of your account you're willing to apply your bids to. The larger the samples that you're bidding on, the more certain your decisions, but the slower you can react. So it's a balancing act between certainty and reactiveness. And so you've got account level data, and that obviously gets broken out into campaigns, which are then broken out into ad groups, and then any number of segments based on anything from product segmentation in your shopping campaign, to audience segments, to time of day or geography, and so on. And so you can take almost any account of any size and eventually sub divide it down to the point where none of the individual segments have enough data to produce a meaningful bid.

You can think of this as if it were a car, but a very strange car indeed, where there's a coin slot on your steering wheel, and the coin slot in your windshield. And you have a finite number of coins to put in, and the more coins you put into the steering wheel the more responsive the car is to your commands, the more coins you put into the windshield, the more clear the road ahead. How do you decide how many coins to put into each slot? Put too many coins into the steering wheel to ensure you have the ability to respond to changes, and you lose the ability to see with certainty what's in front of you on the road. Put too many coins into the windshield, and you can't respond to what you're seeing. Either way, the outcome is undesirable.

So how do you resolve this constant conflict between your sample sizes and your level of granular control, because you do need to find the right balance for your account, and no two accounts are the same on this front.

There are three major approaches that I'm familiar with. The first one is machine learning, now this is going to be the worlds most gross oversimplification, so if anyone watching this is actually building a machine learning bidding system, I apologize ahead of time. But this is fundamentally what you're learning a machine learning system to do. So, you start with an input, you start with an intended output, and then you either import some preexisting data, or you run a bunch of tests. And as you start to accumulate more and more data in the system, the system gets a better and better idea of the pattern that is emerging, and eventually identifies what is the "Best possible input."

Now, in our example I've designed a very specific type of algorithm to this machine learning bucket, and we'll get to other types of artificial intelligence in a moment. But if you want to learn more about exactly the kind of logic that's going into this kind of system, not necessarily the implementation but the output, I highly recommend checking out Veritasium's video on regression to the mean, I have put the link in the description below.

The distinction between machine learning and artificial intelligence in this case is one that I'm making, this isn't necessarily standardized, in fact the two are often used interchangeably. But I think that oversimplifies even farther than we're going to do.

And so I'm going to leave regression analysis in the machine learning bucket, and reserve the term AI for things that are built on top of neural nets, and other sophisticated learning algorithms that aren't just building a bunch of data on top of a pre existing and pre determined formula. For these, you once again have to start with a bunch of example data, but instead of just mashing it all together and finding some sort of average and banking on that, or variations on that theme, this type of system tries to understand some sort of underlying pattern that is true beyond just the individual points of data. Some sort of generalities about the way certain types of behavior can be tracked over time.

And so you give it a bunch of individual data points, and then later you say does this match or not match the thing you're looking for? And if it does match, take action A, and if it doesn't, take action B. But, sometimes that can be a little bit complicated.

So these systems are really only as successful as the amount of data that you can give it, and the quality of the resulting models. But they are really dependent on the sheer volume of data. So if you're an enterprise account, your site does more than a couple of hundred million in revenue a year, then this could be a great system. But if you're a very small account, say spending just a couple of thousand a month or less, then it really depends on whether or not your account is the only source of data that it's being trained against, and if it's being trained against other data, it may or may not be as relevant to your account as you'd like.

Now, that was way, way simplified, like barely scratching the surface of even the elevator pitch for an actual machine learning, or artificial intelligence based bidding solution. But I think it's important to recognize that I'm not just talking about third party vendors, I'm also talking about Google's built in optimization, like their OS bidder, or Enhance CPC. These systems just have a lot more inputs, because Google has a lot of data behind the scenes on shoppers and shopper behavior that they're trying to bring to bear.

But if you'd tried any of those systems, you see that sometimes they don't really work well, and hopefully this gives you a little bit of an idea of how those systems might actually be fallible.

Now those are the first two, I promised you three. The third one is obviously maintaining all of your bids manually, just using the ad words interface traditionally. Now the first two suffer from the same limitations that the third one does, namely that if you don't have enough data, they're not going to perform well. The same is true here, but it's a little bit more concrete.

So this is an exercise in determining how granular can you get? And I'm not just talking about say by breaking out your shopping campaign tree into finer and finer branches, but also things like how long a date range can you use? Well the answer is you have to balance all of those different types of segments against each other, so that the resulting sample size has at least a certain number of clicks and conversions. And the number that is appropriate for your click and conversion minimums is going to be dependent mostly on your conversion rate.

Let's for example take a look at a hypothetical product group in a shopping campaign that has a bunch of clicks starting to accumulate. You need to know whether or not you should start to back off your bid because of a lack of performance, or whether you should let it ride because any moment now the next click might convert.

Well there's two ways to go about that. The first is knowing what your click threshold should be. So, for example, one way that I've done it in the past is say take the conversion rate and then do a little bit of math and figure out how many clicks would I need in order to have a 95 or 99% certainty of at least one conversion within that pool? I've I get that number of clicks and I don't yet have a conversion, then something's probably wrong, and it's worth reevaluating.

The other approach is a little more speculative. Take any number of clicks that you have, with a much lower minimum, say like 20 or 30 clicks, and then you assume that the next click literally is that conversion that you've been waiting for. Well that gives you a conversion rate, and then if you know what you expect the average order value to be from that, you can work your way back to a value per click number, and you can generate a bid from that. If it's overly optimistic it'll run up additional clicks, but that'll suppress the perceived conversion rate, and lower your bid accordingly. If you do get those conversions, then that'll bolster the VPC, and then you can bid up normally.

Now, for date ranges there at least two variables to consider when determining how long you're willing to go back in time for data. The first is seasonality, obviously if you're diving all the way back to another period in the year where the behavior can't be expected to be representative of what you should expect tomorrow, that's a problem. You can use year over year data in some cases to help mitigate that, but it's a hard limit.

The other thing is what happens if the behavior suddenly changes? Say a new competitor comes into the market and disrupts the normal flow of shopper behavior. That kind of change is more quickly detected and responded in a shorter date range obviously, but that shorter date range comes at the expense of the level of granularity you can use across keyword bids or additional segments within your shopping campaign, and so on.

Imagine you have a text campaign, so we're not worried about the sub division within the product tree of a shopping campaign. You have the option to either bid at the [inaudible 00:08:46] level or the keyword level. Keyword level is way more granular, you have a lot more control. However, it also reduces your sample size by a factor of how many keywords you have on average per ad group.

So if you're going to do that, say you have 10 key words in an add group, you suddenly 1/10th as much data per keyword, on average, in order to make your decision. And you may end up then needing 10 times as long a date range, which if you're trying to respond to sudden changes in your market, or seasonality, and things like that, that can be more detrimental than beneficial.

So you might look for something that's a little less granular, that still gives you some of that control, say by lumping all of your broad match and phrase match keywords together, and all of your exact match keywords together. Then you only have two samples within the ad group, so you're only doubling the amount of date range you may need to make up for that additional sub division. But that's better than ten times.

Hopefully this helps explain why, no matter what you've tried, you're ad words campaign's always seem to be some kind of struggle to get them to behave the way you'd expect.

StatCave Episode 6: An Infestation of Affiliates

That's the right collective noun, right? Just kidding, but I am going after a few common types of affiliates that are shifty and I generally recommend avoiding. Coupon affiliates, I'm looking squarely at you.

There are a hundred ways to let others erode your margin, but I narrowed in on a few of the most common in this week's video. And yes, there are exceptions to every rule, and no, I'm not naming names, but if even one retailer reevaluates their affiliate program after watching this, I'll call it a success. :)

Up and to the right!

Video Transcript: 
I'm gonna get more hate mail for this video than any video I've ever done. But that's okay. We're gonna talk about shady affiliates!

Now affiliated marketing as a concept isn't necessarily evil, far from it. It's just a descendant from the idea of putting a sales person on commission. You wanna pay somebody based on a specific purpose like closing a sale. Nothing wrong with that inherently but in the past 10 years or so affiliate marketing has become an avenue for all kinds of other deceptive and manipulative practices that you should be aware of if you're gonna run an affiliated program. You can make sure that none of these parasites get in and start sucking away at your profit. 

Perhaps one way to come at this problem is to explain what a good affiliate looks like. I use a very specific measure of whether or not an affiliate is a positive contribution to my overall shopping funnel. That is, do they participate before or after the customer actually decides to make a purchase. Any affiliate that interacts with the customer before that point where the customer has decided to make a purchase is contributing to that shopping cycle. Then it's sort of really up to your attribution model how you're assigning credit for that based on where they are in that multi-touch chain. But, if they're touching that shopper only after they've made that decision to purchase, then really what good are they? What are they actually contributing. The shopper's already going to buy.

So these good affiliates often take the form of some kind of content site for somebody whose doing research before they get to the purchase. That's not the only example, that's just the first one that comes to mind when I think of what I want to see in my inbox coming from new applicants. There are plenty of other examples out there. But there are more examples of shady things that are intended to sort of misdirect your attention and just skim off a portion of all of your revenue.

So, why don't we take a quick look at some of the known really common actively evil versions and I'm just gonna take these straight from Wikipedia's affiliate marketing page.

Now this is kind of a mixed bag. What I mean by that is some plugins and extensions to browsers and the like are perfectly fine from the users perspective, from the shopper's perspective. But, they can still be effectively useless and irritating on the retailer perspective, having to pay those commissions. So, there are basically two types of these to watch out for. The actually malicious ones will do nasty things to the shopper, cookie stuffing, things like that. I've literally had somebody at an affiliate trade show tell me "is not hacking. Is toolbar." I didn't mention the word hacking. I don't know what conversation he was used to having as people came by his booth but it didn't give me a very good indication of what that particularly vendor was providing. 

Now, there are toolbars and things like that that will have some kind of feature that a shopper might like like automatically showing coupons that are applicable to the site that they're already browsing on. But, that doesn't really help the shopping site. That doesn't really help the retailer. They're already on the site. The only thing that the toolbar vendor has been able to tell me is that well if they don't find a coupon on it then they might not shop with you. What? You're holding customers hostage. The number of people using a toolbar is relatively tiny and the number of people who don't even realize that they have that toolbar enabled and don't even necessarily use those coupons from that toolbar are still considerable.

Now, there's a particular flavor of this kind of extension driven affiliate that is a little harder to get really upset about and that is one that's using an affiliate commission to fund some sort of donations to charity. It's hard to beat up somebody trying to generate money for charity right? Well, the nice thing is that the shopper gets to choose where the charity money goes but when it comes down to it, if you are a business decide that you wanna send a portion of your revenue to charity you can do that already. You don't need an affiliate to bridge the gap. The argument is that, oh well people are going to preferentially shop where this toolbar or this browser extension works because they wanna see that charity go and I'm not sure I buy that. People are shopping if they happen to in their normal course of their shopping behavior kick some money toward a given vendor then cool. So it's really up to you to consider whether or not you wanna give them a portion of it.

The first one is really pretty obvious. Those are trademark bidders. So trademark bidding as a concept is if you say were running then whenever people search for that add that might show up is a trademark term because I'd be advertising against people searching specifically for my brand. Not everyone does it but those who do tend to capture a little bit more of the traffic because some of it gets distracted before it gets to the organic results and the cost should be relatively tiny because your quality score is outstanding and generally you don't have a whole lot of competition for your own trademark. So there's no reason not to be doing this yourself. It takes so little effort it's ridiculous. I mean we barely end up charging anybody for managing that campaign because it just is so tiny as a percent of spend. Now, if you're looking at it as a percent of revenue, which is exactly how affiliates work, it becomes a different game. Now the only argument I've ever heard in favor of affiliates being allowed to bid on trademark terms that wasn't just bald face absurdity was the classic refrain of "oh but we only get paid if you get paid."

Yeah well the same is true of a tapeworm. Come on. Now the other kind of PPC Parasite is a little more insidious and it's actually built right in to merchant center. Check this out. Some of these sites are effectively comparison shopping habits right? So the experience would be you have listings with that comparison shopping engine. It's powered by one of your feeds, then they're going to be bidding on products that you carry in shopping against you. They're competing with you. You're driving up each other's costs and then at best the shopper then clocks on one of these ads from one of these affiliates, lands on a comparison shopping site that has a bunch of other site stuff on it. Lots of opportunity to get distracted and not click through and then eventually land on the same landing page you were gonna land them directly on. That's a terrible user experience and it's going to hurt your overall conversion rate for that slice of traffic. If you are running a really incomplete shopping campaign and you're really not doing anything sophisticated with your catalog within that campaign then they might be able to fill in some gaps and that's the argument they're gonna give you.But, my recommendation is just run your ad words well and you won't need any of these people.

Think about the way the shopper interacts with your website when they're browsing around, they're finding the products, they're adding them to the cart and then when they get to the cart they see that little coupon entry field and they go out to the web and look for a coupon to get a few more bucks off and who could blame them. Right? It's not really a problem with the shopper. But then this coupon affiliate brings the shopper back to you as if they had found them in the wild and was delivering it back to your doorstep, when that's just not what happened. They came in after the decision to purchase has already been made. They didn't contribute to that decision in the slightest. Further, if you don't pay them, what's the difference in the customer's experience? They are on your site, they go to a coupon site, the coupon site still has all these coupons because their traffic is dependent on it and then the customer still gets a discount. So instead of giving a discount to a customer and then giving a commission to an affiliate as well and eroding your margin from two different directions, why not just do it one.

There's one reason I can imagine giving an affiliate a share of that kind of shopping experience and that is as a stick to beat them when they aren't compliant with your coupon rules. So if you have coupons that are allowed to be used among your partners and coupons that are exclusive to your internal channels, if you don't pay them any commission then there's no reason for them to respect that delineation at all and every time you send out an email one of your list is going to add to that site. That site may be hidden in your email list. So maybe worth giving them a tiny tiny commission just so that you could threaten to take it away if they don't obey the rules. Some affiliates will claim that the value that they produce is due to the fact that they have your competitors' coupons on their site as well and there's some opportunity to end up with a shopper that decides to shop with you instead of a competitor. 

Few problems with that. One, nothing stopping the shoppers from going the other direction so then it just becomes a coupon price war and two, it just doesn't happen often enough to matter. Perhaps this is their most critical error, that's a testable hypothesis and we tested the crap out of it. We demanded that our coupon affiliates generate at least one percent new traffic. So if only one percent of the shoppers that we were paying them a commission for were coming from other shopping experiences rather than just from our site, we said good enough. One percent was a passing grade. I don't know any other situation where that ends up being the case. None of them could meet that threshold. They're not doing it. They're not doing it at least in any significant volume. Some of them will use paid advertising to encourage that sort of cross-brand shopping so I'm not really criticizing that potentially but just claiming that you should be paying them a commission because of this possibility of a shopper coming from your competitor's funnel is ridiculous.

I look forward to your letters. 

StatCave Episode 5: Ad Budgets and Pie

Working from your P&L to your AdWords budget is a complicated process. Pie isn't complicated, it's delicious. So, we use one to fix the other!

COS Worksheet link here!

Video Transcript:
There's a common mistake made when reading a P&L for figuring out how much to spend on a paid channel like AdWords but, to demonstrate the error, I'm going to have to buy some pie, which I'm heading to do right now. With our pie safely acquired, we're ready to take a look at how this is usually done. Imagine this pie represents the entirety of your revenue and from this, we're going to start subtracting all of our known costs. We're just going to look at fixed overhead. 

This includes rent on your office, salaries for your employees and other costs like software and IT expenses. Next, the big one; cost of goods sold. In order to sell something, you have to buy it originally or have it manufactured and that's going to make up a large portion of this pie. Then there's always going to be additional fees such as; merchant fees or credit card processing fees that are going to be taken out as well, at the end. 

Now, we are looking at that slice and that slice represents two major things that are left. That is profit and our marketing budget and so, when you're slicing up the pie like this, you end up looking at those two as adversarial. You're considering, "Well, the more I spend on marketing, the less I'm going to keep in profit." That's totally misunderstanding how marketing budgets work and what they do. They're intended to drive top line sales and therefore, you have to treat marketing budgets as an input, whereas, this entire approach treats it as an output. 

Now, a P&L is totally an accounting tool and is very important and critical to understanding how your business is turning a profit but it's not a predictive tool. It really should be used forensically against months that have already closed and you can use it to sort of get an idea of what's coming if you assume that the future will be like the past but if you're actually planning budgets, it's much better to work the other way around.

Take the same ratios that might be in your P&L but treat them as a way of assessing any piece of given revenue. To start off with, we're going to need to start with a new pie. Now, this time, instead of taking the pie, which still represents all of our revenue, and trying to extract out costs, we're going to assume that some of those costs are just baked into every single dollar that is this pie so things like fixed overhead aren't a slice, they are the crust and so, it's present throughout the entire pie, no matter how you slice it. 

Now, we're going to simplify things a bit and pretend that there are just two types of traffic and therefore, two types of revenue. Those are paid, such as PPC, email or social and then unpaid, such as, organic, direct or phone sales, depending on how your attribution works. The first thing we're going to do is separate this pie into those paid and unpaid segments. For simplicity, we'll assume that you have 50% of your revenue coming from paid channels and 50% coming from these non-paid channels. 

Because you have to pay for products before you can sell them, regardless of what channel may generate the traffic originally, we're going to remove cogs from both sides. Finally, we still need to take care of those transactional fees like credit card processing and potentially shipping if it wasn't part of the cog slice. These two slices combined, represent your gross margin. That puts an obvious ceiling on how much you should possibly spend on any marketing activity because, if you spend more than your gross margin, you will lose money by definition. 

However, you also don't want either of these two slices to go entirely to marketing spend and the reason is, within just the paid slice for example, if you spent more than just this sub slice, then you are going to lose money on that segment and would have been better off not running those channels at all and so, we end up with a pretty good, hard, fast ceiling on the maximum amount we would ever want to potentially spend on paid channels. 

We know we don't want to spend more than that because it doesn't produce any profit. We also don't want to spend too little because then we won't generate enough revenue in the first place and so, there's an economy of scale that we're aiming to balance, obviously. Spend too little, you don't generate enough revenue. Spend too much, you don't generate enough profit. I'd love to tell you that there's a magic formula that finds that point of diminishing returns for your business but it varies so much. 

Ranging from what your margins are, to your conversion rate, to what your competitors are doing in the space. Within retail, I generally see about 30% of the gross margin allocated as marketing budget. That's a great number because it's less than 50%, half of the margin was what our little paid slice was. We knew we didn't want to go beyond that. Now obviously, because 30% of the gross margin is less than 50% of gross margin. That's where our profit comes from that channel. 

Then we look at a potential gross margin after advertising cost, of the remaining slice of the paid, plus the unpaid side. That's all well and good, assuming that your paid and your non-paid channels are about equal, which is a nice, healthy ratio but, if your organics are smaller than that, then all of a sudden, your overall profitability goes down because that organic gross margin slice is gone. What do you do? You don't have as much organic as our little pie example had. Then, 30% of your overall gross margin is going to be much, much smaller because it's going to be 30% of a slice that's mostly that paid channel. 

There are two paths from here. One aggressive, one more conservative and finding the right one for you is going to depend heavily on your business context. The more aggressive position is to spend a larger percentage of your gross margin, perhaps something that's comparable to what you would have been spending if you had had the organic there to pad your total gross margin number. Instead of just 30% of your gross margin, you might need to use 60% of your gross margin in order to bolster the budget based on the fact that you don't have that organic revenue coming in. 

The more conservative approach is to continue to stick to that 30% number, knowing it's going to be much smaller, knowing that's going to limit your top line revenue growth but if you're spending that time to short the defenses on your unpaid and organic side, then it can be worth the trade off because once you get down to whatever that slice is, if you then compare that slice that represents your advertising budget against the size of the total pie, you get a percent of revenue that you're willing to spend on ads. 

We refer to this as a cost of sale number but, it's effectively the reciprocal on the return on ad spend and is represented, as I mentioned, as a percent. This makes it very easy to compare it to a margin percentage and it communicates much more effectively than a row-as number, exactly how much you're spending of your revenue. Now, if you look at this pie, you see that it has both organic and paid mixed in. If we're just looking at a paid target for example, in an AdWords account, you need to take into account, the full context. 

If you can target 8% of revenue overall, that can be subsidized by that organic slice, to double it, effectively. Right? Because if half of it is not paying for the traffic and half of it is, and you can pay 8% overall, then that means you can pay 16% and still come out spending the correct amount overall. Another way to go about it, instead of doubling your paid target is to treat your organic channel as if it were a paid channel and instead of allocating dollars to an ad spend for it, you allocate those dollars to content generation to, technical SEO work on the site. 

Basically, anything that goes back into trying to make that segment grow and, while that might be limiting what you could have potentially spent on AdWords, it can result in a more interesting and healthy marketing mix. If you're ready to get started on the next step, which is applying this to your actual business with something a little bit more precise than me badly cutting a pie, we have a COS worksheet that's just a spreadsheet, that you can actually work through and identify what different changes to your marketing spend would have on your overall profitability, not just revenue. It's a little bit speculative but we found it to be very, very predictive for most mid-market retail accounts. I will include the link in the description and I hope you take a look. It's quite a handy tool. Now, if you'll excuse me, I have about six pounds of pie to eat. 

StatCave Episode 4: Competitors & Trademarks!

AdWords is a naturally competitive environment so, should you be bidding on your competitor's trademarks? Some people think that you're not allowed to bid on your competitor's trademarks and that pretty much isn't true. You're not allowed to use their trademark in an ad, but as far as keywords go it's way more of an open field. It's a perfectly natural instinct for a competitive retailer to want to go attack their enemies on their own territory. That's one way you might be able to capture more market share, right? Well, the problem is that laying siege to your enemy's castle has many of the same downsides that sieging an actual castle has. Namely, they have defenses. In their case it's in the form of relevancy. The fact that their brand is on their site and highly relevant to their site means their quality score is going to be fantastic. You might not have their name anywhere on your landing page and therefore your quality score is going to abysmal and right out of the gates are going to be at a significant bidding disadvantage on those keywords if you're able to show up at all.

But let's even set that aside for a moment and assume that you were able to get past that CPC disadvantage say through being willing to take a loss or something like that which is almost never a good idea. You're still not out of the woods. The problem is that those people are looking for that competitor's website. That is the most relevant search for them and so therefore your bounce rate is probably going to be terrible because they're going to land on the page, go "Ugh. This isn't what I had in mind," and bounce back. If they don't, they're still probably not going to convert because you still aren't what they were looking for. Remember ad words is best deployed when you're trying to match the shopping intent with your offering, but their shopping intent doesn't line up with what you're offering. You're offering a totally different site and so your relevancy, not just to ad words in terms of your quality score, but your relevancy to the shopper is low and bad and it's going to result in an abysmal conversion rate. You're going to be losing money pretty quickly. 

Now, this is a great point to jump to a related topic which is what to do when the enemies come to your castle walls. So if your competitors are bidding on your trademark terms, what do you do? The fact is that your castle has all of the same defenses that their castle has so advantage you in this case. There are two types of instances where this will come up. One, large retailers like Amazon who just cast wide nets across the shopping space and the classic example of this was Ebay a few years back. We all saw tens of thousands of those looking for any key word here, find it now on Ebay. Amazon's doing the same kind of thing. It's more sophisticated, so it doesn't suck quite as bad, but it's still pretty garbage and so their relevancy isn't going to be outstanding, they're just going to be an always present also ran if they show up at all. And they don't generally try and go after your trademark so unless you're trademark is a fairly generic term that is descriptive of the products and your retailer then they're probably not going to go after you. 

The other type of competitor that'll show up in this kind of combat is a direct similar kind of weight class competitor, so if there's another online retailer that's selling a similar catalog to, they might try and chip away at your castle walls thinking that they're going to cleverly capture your market share. Now, just as we were talking about this is a bad idea for you to attack their castle, same thing here. You can deploy, you know, hot oil and throw rocks at them from on top the walls and it'll be hilarious. They will however have to stop at some point. Just like an actual siege once again. The limitation is how many resources can be deployed on either side of the wall and you have the advantage of the better conversion rate, the lower cost per click, and they're running up against some kind of finite budget. They can only throw so many dollar bills at your wall before they realize this isn't making us money. Now if I'm wrong about that, that's actually good news because then your competitor is losing money left and right and isn't that just the best? I am a fan of letting my competition lose money. 

There is exactly one, count them, one use case where I think it makes sense to go after a competitor's trademark terms and that is if you are a considerable underdog. The example I want to share with you is actually someone from my own experience as a retailer. Before I worked at Leslie's Pool supplies where I was VP of digital marketing obviously, I was CMO at Pool Supply World, a much smaller company online peer play retailer which was eventually acquired and that's why I ended up at Leslie's, but when I was at PSW and before the acquisition, Leslie's was my biggest enemy. They were the only national brick and mortar chain. They had the largest online presence and so if I wanted to try and chip away at their dominance, I had to go for the jugular and I could do something that you shouldn't do if you're not a significant underdog. I created a direct comparison landing page showing, here's what Pool Supply World can offer and here's what the competitor target can offer and this makes sense if you are the underdog because people recognize the big brand and they don't recognize your brand and so that comparison is saying, okay well you know how legitimate this looks, well we are just as legitimate and here are these other advantages. 

You know, you take the similarities to draw legitimacy and then you use the differences to sell your superiority. If you are, you know, a major player in your market already, this can undermine your brand because by ... If you were the big dog and you're recognizing all of the little ones that are nipping at your heels, you're actually increasing their exposure, but if you're the little guy, there's a potential way to go after it and that is making sure that the landing page has a chance, any chance at all at having a decent quality score and that has to be going directly after the name of your competitor which is a trademark tactic. But please, once again, only if you're the underdog. 

So, if we're not gonna go after their trademark terms, how do we make a competitor sweat? My favorite trick is to go after their organics. If they are getting a significant portion of traffic for an organic search result, they are not paying for that which means it's more margin rich for them and your ads would show up higher on the search results page than their organic listings. So you get this dual benefit of gaining the sale potentially but also stealing some of their most margin rich revenue, a good combination if you can reproduce it. While there are tools like SEO Martian Spy Fu which are just two that I've used, I'm sure there are plenty more that help you with key word research, when it comes to this kind of strategy, it's not nearly as difficult or sophisticated as some things that you've seen maybe from SEO experts. For our purposes, all we need to know is what does this site think that is their primary category or categories, what do they think they're strong at and their navigation, their site layout, their merchandising, their email program are all just gonna tell you that because they're trying to tell the customers the same thing. You don't need access to your enemy's ad words accounts to know where they're probably investing the most attention because it's gonna map pretty closely to where they're probably investing the attention on their site and other marketing channels. 

Alright. That's it. Basically my advice is, don't bid on your competitor's terms unless you're the underdog and have a landing page where you can have some quality scores. If your enemies are coming at you, let them. They're going to lose and then if you do want to counter punch, go after their organic listings. So the last thing I wanted to mention is I happen to be wearing my Feedonomics shirt actually today. They're a partner of ours that do fantastic feed management work and they are co-sponsoring StatBid Summit on February 26th, the day before Etell West in Palm Springs. We are throwing a massive, well massively important, but small venue, small group event with PechaKucha style presentation, so we've held maybe a half dozen of these in the past where the presentation format is auto advancing slides every 20 seconds and there's only 20 slides each, so each presenter only has about six minutes, so nobody is bothering to check their phones during each presentation. They're fantastically fun to watch, action packed, full of amazing details and then the Q & A is just out of this world, so I can't wait. I hope to see you there and, because you're watching this video, I'll provide you with this discount code which'll get you in for free. Anyway, it should be a blast and I hope to see you there. 

StatCave Episode 3: Daily Budgets

In AdWords, the daily budgets are set up for a couple of reasons. One, it gives Google an idea of how big you think a campaign is going to be. If you set it to $10 a day, that tells Google something. If you set it to $10,000 a day, that tells them something else. The other thing is obviously just a risk management thing. By having those set up, Google avoids having quite as many calls from frantically terrified people who've just spent more money than they've ever spent in AdWords in a month, in just a few hours.


There are primarily four different ways that people are using daily budgets, that I'm familiar with. The first is in conjunction with an automated bidding strategy. If you have an automated bidding strategy like target ROAS, or maximize conversions; Google's going to use your daily budget to identify how much it should be spending over time in order to capture as much of whatever the outcome is that you're measuring toward. If it's maximize clicks, they're going to try and get as many clicks as they can for that budget, within each day.

Then, the other thing that people will use it frequently for is managing to specific weekly or monthly targets. If you have a budget of $100,000 in a given month, then people will use the daily budgets to try and keep it on the rails toward that end.


Another thing people are doing, and this will be a little bit more interesting to get into, is they use budgets to shift investment around on the calendar based on anticipated differences in account performance.

Then the last thing is just managing risk, sort of in-line with the original intent of the feature, from Google's perspective. Many accounts, including the ones we manage, will use the budgets primarily just as a worst case scenario. If something goes off the rails, it'll hit this ceiling and go no farther. That sort of limits the liability of the account.


With automated bidding, there are a number of options available depending on your campaign and how much data is available. Some of these are obviously very highly automated, where you hand over the keys to Google for the campaign. Others like enhanced CPC, for example, still lets Google fiddle with the operation of the bids; but for the most part, leaves you in control of all the other variables. The thing is within these, only a few actually drive results that you want.


If you know what your return on ad spend target is, or if you're using the inverse of that as we do, cost of sale, which is your cost divided by revenue; that percent of your revenue that you want to spend on ads is a very useful metric, so similar to ROAS. If you know that you can only spend 10% on the ads, and still make a reasonable amount of money; then if you are running one of these other types of bidding strategies -other than target ROAS or manual CPC, assuming your managing it actively- you're not producing that kind of result. 

If we look through these, things like target search page location is not a metric that matters. It doesn't actually describe the value you're extracting from the traffic, and so it's a bit of a red herring. 

Target CPA. CPA is a stupid metric. Retailers should not use it unless they have a very flat catalog in terms of price, because otherwise you're treating a $2 product the same as a $2000 product; and that's ludicrous. 

ROAS is pretty much, in at least theory, identical to what StatBid tries to do using our tools; but your mileage may vary. I've seen it perform very well. I've seen it perform catastrophically badly; but at least philosophically it's in the right space. You're trying to generate as much revenue as possible at a specific return on ad spend; fantastic.


Target outranking share. Another red herring. It doesn't have anything to do with anything that actually matters. It might correlate with things that matter, but it is not actually itself a thing that matters.

Maximize clicks. Once again, how many clicks does it take for me to buy a burrito? That's a ridiculous question, because you need money to buy a burrito; and money does not come from clicks, it comes from conversions.

Which brings us to maximize conversions. Now this one sounds good on the surface, and this may work well in some use cases; but mostly, it does the same thing that the CPA target does. It ignores variability in your average order value. Therefore, it can't necessarily produce a consistent return on ad spend. Now, you may have an example where it has done so; but in general, that's not its job. Its job is to convert as many instances of transactions as it can for the given budget. If that means it's focusing mostly on $10 items, and ignoring your $150 items; so be it. That might be the best way to maximize conversions. 


Then, you have enhanced CPC, which is kind of like manual CPC, but where Google uses a little bit of its magic data from behind the curtain to increase some of your bids as much as 30%, I believe. The issue there is that it doesn't then discount bids, where it thinks there is a lower likelihood of conversion; so you end up with this hidden bias in your bids, on average, going to those auctions.


Then, obviously manual CPC; and that's how we manage everything. That's how most of our competitors manage things. You're using some sort of other logic, and you're just implementing it directly. Sometimes that's done by hand. Sometimes that's done via the API or scripts, but it's the low bells and whistles option among all of these.


Let's assume for now that you're looking at using one of these automated solutions that are more heavily automated, like target ROAS for example. If you're using these, then those budgets start to mean something different. They're no longer a liability control; they're basically like telling Google how big is the gas tank in the car that it's driving. It'll give an idea of how fast, how much fuel can it burn per hour if it's going to get a budget of that much per day. However, these are mostly bad ideas for retailers.

Now, they're built for the one-size-fits-all that is AdWords, not specifically commerce retailers like me and probably you. They're just a dead end, because they don't produce the actual outcome you want. You want as many conversions as possible, while preserving your margins; and that's not what those things do.

It is worth remembering that Google's rewarded by you spending as much as possible, all the time. While they definitely have access to incredible engineers, and incredibly clever algorithms, and tons and tons of data; that is a conflict of interest to some degree. Just be aware of that when you're evaluating whether you might want to use one of those built-in automated bidding solutions; because it's basically like having an employee whose best performance is as bad as it can possibly get, but not quite enough to get fired.

The next use case is managing to set weekly or monthly budgets. This sounds like perfectly natural, but realistically it happens very rarely within retail accounts. I've seen it way more in direct-to-consumer brands, where some portion of the budget may be allocated to general branding activity, where you're supporting the awareness of the brand. For retailers, generally they're always happy to take the next order, as long as it's coming in at their intended COS or ROAS target. If you're making money, why would you ever turn off the spigot? 


For some people, they see that and go, well, there's a lot more traffic during the week than there is on weekends; and so I should jump in every Friday and Monday and jiggle those budgets around or build a script to do it for me. I would say that that is potentially a misdirection of attention.


If there's not a meaningful difference in the VPC between Saturday and Tuesday, then there shouldn't be a rational reason to bid differently on Saturday or Tuesday. If your return on ad spend is the same on Saturday and Tuesday, you'd want neither of them to run out of budget. To say that dollars on a Tuesday are worth more than dollars on a Saturday is kind of ridiculous. If there is a meaningful difference between value per click between say a Saturday and a Tuesday, you can take that into account; but the budget is not the right tool for that job. That's what the ad schedule bid adjustments are exactly for.


If you see a difference in the quality of traffic coming from a specific time slice, say day of week, you can already account for that using another tool that is specifically built for that. If Saturdays don't convert as well, and you want to reserve as much of your budget as you can for the rest of the month, where you do have days that convert more strongly; then why don't you just bid less on Saturdays, using an ad schedule bid adjustment for example. Then naturally, because you're bidding less, you're going to end up spending less; and that'll end up reserving a natural amount of budget for the other days. The budgets end up not being an issue, because you're always adjusting your bids to reflect the different quality of traffic that you might be getting across those varied segments.


You can definitely use daily budgets as a risk management tool, as I've mentioned. In fact, that's exactly what StatBid does. We have a system in place that watches the daily spend. If it sees that the campaign is spending a good portion of the daily budget, it'll actually just increase the budget. We can do that, because we have bidding systems in place that we trust to be producing the return on ad spend that we've been tasked by the client to hit. 


That is a great thing to have in place, in case there's this huge sudden increase in traffic available. Even though we do let our system increase those budgets automatically, it also sends an email to the account manager saying, "Hey, make sure this thing isn't burning the house down." Then we can go in and just make sure that the behavior is as expected. It's a good tool for just keeping your liability tolerable, as long as you're keeping it out of the way of the opportunity to capture more traffic and more orders.


There's one last thing that I wanted to mention, since we're talking about daily budgets, that I think is directly related. That is delivery options. Within a campaign, most campaign types anyway, you can set either accelerated or standard delivery method. Standard delivery method will spread your budget out over the course of a day, making sure that you have coverage across the entire clock face. Accelerated will participate in every auction that it is eligible to participate in, regardless of how quickly it's accumulating cost relative to your daily budget; and so you might run out.


However, if you're taking the advice that I've given you in this video, and the blog articles, and things that I've written; and you're managing your account to a specific return on ad spend, then you have no reason to want to set it up so that you're not capturing every single potential shopper that you can. So the standard delivery method sounds kind of silly. In fact, it's a little bit like that ad schedule adjustment thing we talked about between the Tuesday and Saturday, but within a single day.


You're assuming, by using standard delivery method, that say a dollar you receive in an order in the evening is somehow worth more than a dollar you receive from an order in the morning. The reason that I say that is because by spreading it out, you're saying I'm going to ignore these orders from the morning, and I'm going to favor these orders in the evening by shunting some of the budget farther along in the day. That's once again ridiculous. A dollar is a dollar is a dollar. I'm not a fan of that strategy, because I think it's sort of bizarre. 


I mean, it seems qualitatively kind of subjectively good, because you're covering the entire day and you always have ads running; but there's no difference. You still have ads running for some of the time, and not running for the rest of the time with accelerated; it's just segregated into here are the times that it is running, and then it runs out. Whereas standard, it's just sputtering the entire day.


Now, with accelerated the other advantage is that you can tell more quickly if a campaign needs a larger budget. If you're running it on standard, it'll still eventually show you the budget limited alert in AdWords; but if you're running accelerated, it'll definitely show you that alert, and it'll do it more quickly. So I find accelerated to be the better tool, but like most other things we've discussed, that's if you're confident that you're managing to a specific return on ad spend. If you're not confident that you are getting the return you want, then obviously do whatever it takes to reduce your spend and reduce your exposure. Fix your return on ad spend, and then go back and open up the flood gates.