Google are currently beta-testing a new service, called ˜Automatic Matching’. The premise is simple , if you aren’t spending your complete budget, then they will show your adverts on searches that you aren’t bidding on, if they think it will be successful. On the face of it, this may sound like a good idea, but Google’s history with automated tools is, to put it mildly, poor. Looking back at their innovations over the past 12 months, there has been a steady stream of ˜improvements’ that will allow you to manage your campaign with less effort. And at the current time, I am using precisely none of them. Now, it is true that as a PPC account manager I should probably be doing things by hand, but the simple fact is that whenever I have used Google’s new tools, I’ve been disappointed. Here are a few examples…
Expanded Matching
When this first came out, it did so with little in the way of warnings. The first I knew about it was when my clickthrough rate collapsed on a few of my campaigns. Suddenly, where I had been bidding on leather beds on broad match, now my advert was appearing on searches for upholstered beds. In truth, I got lucky. Somebody who searched for upholstered beds clicked on my advert. As a result, it appeared in the Search Query report. If nobody clicks on your advert having searched for a term, then you don’t know about it. Even now, nearly a year later, if I search for upholstered beds, there are two adverts for leather beds. I guess some people don’t run search query reports very often… This option was designed to fill in all the keywords that are relevant to your campaign, but that you haven’t thought of. But the problem is that you can’t have Broad Matching without Google throwing in Expanded matches. Since Expanded Matching is totally unreliable, and you can never even tell what searches fire your adverts, Broad Matching has become, for me at least, unusable.
Pay-Per-Action
This sounds like a great idea. Basically, you tell Google how much you’re willing to pay for somebody to perform an action, and Google adjusts your bids accordingly. So I tried it on my two biggest, and most consistent accounts. Both of them had a very predictable cost per conversion, and I set that as my objective. The result? Virtually no traffic. In two days, a campaign that usually collected 250 clicks per day managed only 2 clicks (neither of which converted). But the point is that this is automated. It’s never likely to work very well. Either it’ll bounce your bids all over the place depending on the results from the previous day, or it won’t be sensitive to changes over time, and your bids will always be in the wrong place. It takes a human brain to work out whether the variation in the conversion rate is just random, or part of a more significant trend. Perhaps, with a lot of work, a computer could be programmed to draw reasonable conclusions, but as a rule, the tools implemented by Google tend to be relatively simple. I feel that my test could have been a lot worse , Google could have spent far too much per click. Underbidding may cost me potential sales, but at least I can’t make a loss!
Site And Category Exclusion
Alright , I wasn’t totally honest earlier. I am using this one for a few of my campaigns. The idea of this tool is that you may want to include or exclude certain categories of websites when you advertise on the Content Network. So, for example, you may decide that you don’t want your advert to appear on Forums, or on Video-sharing sites. And rather than input them all manually, this allows you to filter out irrelevant websites en-masse. This sounds great, but it raises two questions. The first one is that if these sites aren’t relevant to your adverts, why are they appearing there in the first place? This is Content Targeting. It’s targeted on the relevance of the content to your website. If it’s relevant, why do you not want to appear there. If (as is more likely), they aren’t relevant, why does Google show your adverts there? But the problem with the whole thing is that it’s just a shortcut to avoid you doing some work. Ever since Google implemented its Placement Report, you have been able to see which sites are working for you, and which ones aren’t. Armed with this information, surely it’s not that difficult to create a Placement Targeted campaign, and choose which websites you DO want to appear on. It doesn’t take very long, and you can always reactivate your other Content Network campaign occasionally, to make sure that you aren’t missing valuable traffic.
Budget Optimiser
This one’s just a bad idea. The idea is that you decide your daily budget, and Google adjusts your bids to get you as many clicks as possible for your money. That’s as many CLICKS as possible for your money. Exactly why would you want to do that? If you’re selling something, you want to maximise the value (or at the least, the number) of sales for your money. If you are generating leads, you want as many leads as possible for your money. So who would possibly want to maximise the number of clicks that they get for their money? I can’t think of many businesses that would find that useful. The simple fact is that clicks are going to be worth different amounts, depending on the conversion rate (and in some cases, conversion value). Google can’t take that into account, so it could easily bid a lot on low-quality keywords, simply because they are cheaper than the more competitive, more relevant terms.
Campaign Optimiser
There’s that word again… Optimiser (optimizer if you’re American). A quick look up on the Interweb tells me than optimization is defined as: œThe decision strategy of choosing the alternative that gives the best or optimal overall value The best value of what? Well, that’s up to you, really. In most cases, it’ll be profit, Google doesn’t know what a conversion is worth to you, But it doesn’t even take into account your conversion rates. Here are the changes it will propose, along with Google’s definition…
- Daily budget adjustment Budget changes can affect your ad visibility and bring you more targeted traffic.
- New keywords Proposals might include new targeted keywords that relate to your landing page.
- Deleted keywords If the Campaign Optimiser identifies poorly performing keywords, it may propose removing them.
- Changed keyword matching options The right matching option can help you reach customers more effectively.
- Keyword CPC bid adjustments Your cost-per-click bid (in addition to your ad quality) affects your ad position.
- Ad text edits The Campaign Optimiser may suggest changes to make your ad text more effective.
- Refined location targeting We will suggest that you target only those regions that perform well for you
So, it will tell you if your budget’s wrong, even though it doesn’t know if your campaign’s profitable or not. It will recommend new keywords , if you’ve ever used Google’s keyword suggestion tool, you’ll know how well that works. It’ll tell you which keywords aren’t working. Based on what? Clickthrough rate? Cost per click? These are not things you want to be basing your decisions on. So Google knows which matching types are appropriate for you? How? If it knows that there are keywords that you could pick up through Phrase Match or Broad Match, then they should be added in on Exact Match. If it doesn’t, then why would it recommend them? Bid adjustments are just a bad idea, for all the reasons discussed already. Advert Text edits??? Google briefly trialled an Advert Text writing tool. It was so bad, I lack the words to describe it here. I wish I’d kept some of the adverts it suggested, but sadly they are no more… Regional Targeting? It all comes back to what Google calls optimisation. They want to maximise your clickthrough rate. Which is definitely not your objective.
Preferred Cost Bidding
Haven’t tried this one, but it worries me a bit. The idea is that you decide how much you want to pay per click, and Google adjusts your bids accordingly. Sounds OK, but one of the few backups that you’ve got if things go wrong is your maximum bid , you can’t pay more per click than this under any circumstances. I’m sure that Google are going to be careful, but how well would this tool work? If you want to pay £0.20 per click, but yesterday, a £0.30 bid generated clicks for £0.19 each, will Google adjust your bids to £0.31? This won’t increase your cost per click by £0.01, as a rule. Either your advert will stay in the same place, and your cost per click won’t change, or your advert will move up a spot, and your cost per click could jump sharply. Extending this example, if you need to pay £0.19 per click to appear fourth, and £0.31 to appear third, then this change would increase your cost per click from £0.19 to £0.31. Which one does Google go for if it’s in charge? The point is that you can’t choose an actual cost per click , it’ll be one of a set of possible values, based on your Quality Score, and the Quality Scores and bids of your competitors…
Optimised Ad-Serving
Of all the stupid ideas I’ve heard in this business, none can compete with Optimised Ad-Serving. Here’s how Google explain it:
œOptimize (default): The system will favor ads with a combination of a high clickthrough rate (CTR) and Quality Score. These ads will enter the ad auction more often.
So, you’re running more than one advert for this to make any difference. Why would you be doing that? To find out whether the new advert is better or worse than the old one? So how useful is it is your new advert gets shown 1% of the time, and the old advert gets shown 99% of the time. You’ll never get decent results. I can only guess that Google want to limit your risk by showing the established advert more often. And it would be useful if you could do this, but a more useful tool would be to allow you to select the relative frequencies that different adverts show. That way, you could show your old advert 75% of the time, and the new one 25% of the time, if you wanted to. A number of times, I’ve done assessments of campaigns for potential clients, and they’ve been testing multiple adverts, but with Optimised Ad-Serving. The result was that they could never optimise their adverts, as one got all the traffic.
Standard Delivery Method
Of all the stupid ideas I’ve heard in this business, none can compete with the ˜Standard Delivery Method’. Except possibly the Optimised Ad-Serving. Here’s how Google explain it: œStandard delivery distributes your budget throughout the day to avoid reaching your budget early on. Your ads will show periodically throughout the day. So, if you’ve got £100 to spend per day, and you’re paying £1 per click, but your budget runs out by 10am, Google thinks your advert should appear and disappear randomly throughout the day. WHY???????? Wouldn’t it make more sense to cut your bids to £0.20, and get 500 clicks per day instead of 100 clicks per day, for the same money? If you only want your advert to run at certain times of day, select those hours for it to run. But randomly appearing? Is there any benefit at all in this?
Keyword Suggestion Tool
Since Broad Matching doesn’t work, it’s not surprising that Keyword Suggestion doesn’t work very well either. I decided to try it out this morning, on a website that sells conservatory furniture. That’s all they sell, and their entire website is dedicated to this one thing. Google’s suggestions? bedroom furniture, leather furniture, lighting fixtures, outdoor lighting, art prints, chandeliers, art education, school furniture, furniture rental, furniture shipping… Need I go on? None of the keywords were even slightly relevant , no mention of conservatories, cane, or anything else remotely useful. Maybe I have already added everything vaguely relevant, but I doubt it, somehow. The tool may have some value , perhaps if you run it, it may give you a new idea for a keyword. But it’s pretty clear that you wouldn’t want to use Google’s suggestions in general.
Conclusions
So, back to the new tool, then. Automated Matching. Google clearly can’t look at my site, and figure out what it’s about , you can see that from the Keyword Suggestion tool. It’s not very good at working out other, similar keywords , you can see that from Broad Matching results. It can’t measure the success or failure of a keyword, since it doesn’t use the conversion information. It can’t determine what you can afford to pay for a click, since it doesn’t know anything about your business. And yet, despite all of that, Google is willing to spend any remaining budget on keywords, without even telling you what they are. I believe that one person can keep an account well-managed in about 10 minutes per day, plus an hour every now and then to do a bit more analysis. Is it really a sound business-decision to give Google control of your money? Automated Matching? I think I’ll pass…
Which is all fine and dandy. But the other day, I was doing some forecasts and my profit curve looked like this: 
Clearly, I’d made a mistake! So I went back, and checked my forecasts for the clickthrough rate, the conversion rate and the cost per click. Here they are… 
I’ve changed the actual figures, but the result is the same. With a profit per conversion of £300, this gave me an inverted profit curve. Assuming that the cost per click is higher for higher positions, the conversion rate is lower or the same for higher positions, and the clickthrough rate is higher for higher positions, the profit from each conversion must be higher in lower positions. In my case, the conversion rate was clearly higher, the lower my advert appeared. If this effect outweighed the increased number of clicks that I got in a higher position, then it’s possible that I’d predicted that I’d get more conversions in a lower position than in a higher position. For example, if 5th place generated 5,000 clicks with a 3% conversion rate, and 6th position generated 4,000 clicks with a 4% conversion rate, then 5th place would generate 150 conversions, and 6th would generate 160 conversions. Clearly this is a danger when forecasting, particularly if you extrapolate beyond the range of your data. I can’t accept that you can get more conversions from a lower position in practise unless you have a restrictive budget (which I didn’t), so I looked at my data to see if this was the problem… 
So that’s not the problem. Finally, I looked at the profit per conversion, the number of conversions, and the product of the two (the total profit). 
The number of conversions is lower in lower positions, the profit from each is higher, and you get this ‘inverted’ profit curve – a ’sour spot’. So, the question is whether this is possible in reality, or if it’s just a flaw in the forecasting method. The answer is surprisingly simple once you think about it. If you advertise in a very low position (say, 100), you’ll get almost no conversions, and hence make almost no profit. The true shape of this curve would probably be something like this: 
It’s possible that multiplying these two monotonic functions (conversions and profit per conversion) can generate two turning points in your profit curve – a maximum and a minimum. I can accept that this is possible, and graphs of the above shape will have a sweet-spot of either 1st or the local maximum (in the above example, 6th). This raises one final question. In the above example, I looked at the top six positions, saw the sour-spot and understood that I needed to extrapolate further. But if I’d only run the advert in positions 3 to 8, I would have seen a sweet-spot, and thought no more about it. In this case, I’d still (just about) have the correct sweet-spot, but another time, I may have missed out on potential profit. And perhaps I have done. My conclusion is this – extrapolate your data as far as possible, limiting your graph only at your total budget. See if this kind of shape is a possibility, and investigate it.
There is no real falloff in the number of conversions, and a significance test of the difference in conversion rates is totally insignificant. In fact, to get significant results (even at the 90% level) for the conversion rates, you’d need to wait much longer. 
To put that in context, if you were getting 400 impressions per day, the tests for click through rates would take (1 + 3 =) 4 days, whereas the tests for conversion rates would take (38 + 8 =) 46 days. That’s quite a lot longer. So, what’s the conclusion here? Should you run your campaigns for ten times as long, to confirm that the new advert doesn’t hit your conversion rates? Bear in mind that the changes above are quite extreme , it’s unlikely that your results will show anything after waiting ten times as long , when do you draw the line, and say that the change is too small to matter? Even here, we’ve not taken into account the impact of reducing the cost per click (which will slightly offset a reduced conversion rate), or the impact of increasing the total number of conversions (even at a slightly higher cost per conversion, this could still be a good thing). Alternatively, should you just ignore the conversion rate, and hope for the best? Or try very hard not to change the meaning of the advert? You only need to write one bad advert to wreck your campaign. Perhaps the best approach is to physically look at the conversion rates of the adverts that you are dropping , if they are lower, then ask the question œhave I caused this to happen? The fewer conversions that you are getting, the harder it’ll be to stop a problem , so monitor the conversion rate, and if it starts to drop, check to see if you’re the cause.
(cookies - breakfast of champions)
Thanks for the fantastic write up!!
1 comment