I blogged a while back about the sweet spot for your campaign, and how to find it.Basically, you estimate the conversion rate, cost per click and clickthrough rate for each position that your advert can appear in, and calculate how profitable each one is. You should find that one position is more profitable than the ones above or below it, and so this is where you should be putting your advert. 
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.
PPC Campaigns Blog Posts
When The Adwords Sweet Spots Turn Sour…
Thursday, October 18th, 2007Advert Text, Bidding, Content Network, Google Adwords, PPC Campaigns, Pay Per Action, Testing
Small Budgets And Big Keyword Lists
Tuesday, September 18th, 2007Adgroups, Advert Text, Content Network, Google Adwords, PPC Campaigns
I recently saw somebody describing their campaign on a forum. They said that they had 250,000 keywords, and were concerned that Google may ban them.
Really, this was the wrong question – what they should have been asking was how they could possibly manage a campaign with 250,000 keywords.
Do a few sums, and you see what I mean. Suppose that a typical click costs £0.20. How long do you need to run a keyword before you can hazard even a rough guess at its conversion rate? 100 clicks? If you’ve got a low conversion rate, even this may not be enough. But to get 100 clicks on 250,000 keywords, at £0.20 per click would cost £5,000,000. And how long would you have to wait to get 25,000,000 clicks anyway???
Realistically, the majority of these keywords would get no traffic at all, and 90% of the clicks would come from 1% of the keywords. So you can still optimise the keywords that get the vast majority of the traffic, so the problem isn’t that big an issue.
So what about the other 99% of the keywords? If you can’t optimise them, then what’s the point in bidding on them – they may never be profitable! On the other hand, people keep saying that ‘the long tail’ is the key to successful PPC campaigns.
The above example is quite an extreme one – most campaigns won’t have 247,500 keywords generating very little traffic. But the 90%, 1% issue is probably true of most campaigns. If a handful of big keywords eat your entire budget, how will you ever make the other 99% profitable? They’re supposed to be the most profitable in general, with their low cost-per-clicks and their high conversion rates…
One option would be to pause the big keywords, and spend your entire budget on the smaller keywords. This will, in turn, lead you to find that 90% of your traffic is STILL coming from 1% of your keywords, as the largest of the keywords you didn’t pause take most of your budget. These keywords are probably more profitable, but it doesn’t really feel very optimal!
If you’ve read through my case study you can see how I would go about this problem in most cases.
My keywords are generally grouped by product or service, with extra Adgroups for the more generic groups of terms. So in the case study, I had one Adgroup for each printer, one for each printer type/manufacturer etc.
Then I optimised at Adgroup level initially. I optimised this by trying to equalise the ROI from each Adgroup, such that the total daily budget lasted (on average) just until the end of the day.
This should maximise the number of conversions that you get per day.
Having done this, I look within the Adgroup that’s getting the most traffic, and start adjusting the bids on the keywords that generate the majority of that traffic, looking at the ROI again.
If you have a situation like the printers, where the products are largely similar, once you’ve got a few Adgroups done, you can see patterns emerging. Certain keyword formations will perform better or worse than others. So you can make the adjustments to Adgroups even without having enough data individually. This is quite good, as you can have a stab at optimising keywords that haven’t got enough traffic.
I would also do one more thing here. If a keyword’s had no clicks after a month, I’d delete it. Even if you get a 5% conversion rate, if a keyword gets a couple of clicks per year, it’s not important; it’s just cluttering up your campaign.
Just be aware that there are clear limitations to this approach. Just because a group of keywords works on printers doesn’t mean it’ll work on photocopiers, telephones or PC’s.
Consider again the campaign that I mentioned at the start of this post. He was promoting a worldwide hotel booking service.
Clearly, the approach is likely to be valid here. If “Hotels in Moscow” converts better than “Moscow Accommodation”, then it’s likely that “Hotels in Durban” will convert better than “Durban Accommodation”.
It’s likely that each city has exactly the same keyword list, with just the city name varying. This is a huge opportunity to save a fortune when optimising. Rather than just switching the whole thing on from the start, why not work out using a few cities which keywords are profitable or not, and how much to bid for each type of keyword? Rolling this out on the others would give you a huge head-start, saving you a lot of money.
But if it works here, why not use this approach for any campaign where you have the same keywords in each Adgroup with just a different model number/city?
Sadly, it’s not really something that I can do, in my position here. When a client asks us to start up a campaign on their behalf, they expect us to build it and switch it on ASAP. After all, one of the main benefits of PPC is the immediacy of the results. You turn on a campaign at 9am, and at 9:02, you’re getting clicks.
But if it was my money on the line, and I had a lot of keywords, and only limited cash, I’d probably use this method.
What do you think? Is this better than the ‘throw everything at the wall and see what sticks’ approach? Give me your thoughts or experiences…
Should You Put The Content Network Into A Separate Campaign?
Friday, August 31st, 2007Advert Text, Content Network, Google Adwords, PPC Campaigns
The answer to this question should be yes. After all, that way you can easily monitor it’s performance as a whole (rather than produce a separate report for it, or having to flick through your adgroups all the time). Also, you may find that different advert text works more effectively on the content network – creating a new campaign allows you to write an advert for the content network, and one for the search network. So that should be that then. But it’s not quite that straightforward. I recently took over the management of a campaign that was running on both the search and content networks. Thinking that this would be a good opportunity to assess the size of potential benefits from splitting the networks into separate campaigns, I did so. The result was disastrous! They had been receiving 80% of their traffic through the content network, and most of it disappeared! Slightly disappointed, I switched everything back. But the traffic didn’t return. So what went wrong? Well, it appears that Google’s content network uses different variables in its Quality Score, or at least puts very different emphasis on the different factors. Specifically, it appears to use the age of the campaign (or, to be more precise, the length of time that it’s been on the content network). So, the longer your campaign is running on the content network, the greater its exposure appears to be on the content network. And if you turn off the content network, when you turn it back on, you’re back to square one. This doesn’t strike me as particularly sensible, though I can see why Google might be doing it. If a campaign isn’t working on the content network, then it gets switched off quite quickly. So the fact that it’s been running for a while is a good indicator that it’s what people are interested in. Remember that the clickthrough rate is virtually useless on the content network, as it highly sensitive to things like the location of the adverts on the page. So what’s the correct move, then? I’d say that if you’ve got a successful, established campaign on the content network, then LEAVE IT ALONE!!! If you’ve just started out on the content network, or you are considering it, then you can certainly give it a campaign and budget of its own. My campaign has now recovered to about 45% of its former traffic. Don’t make the same mistake that I did.
Does The Quality Score Work?
Friday, May 25th, 2007So Adwords measures an advert’s quality primarily on its click through rate, then? And the objective, therefore, is to get as many virtual people through the virtual door as possible, then? But isn’t there a bit of a conflict here? If I was running a shop, and I had to pay people to walk through the doors, I’d try to be as selective as possible, and yet Google is rewarding you for doing the opposite. The result is that frequently, if you do what Google wants, and increase your click through rate, you’re being less selective, and the people that you are paying are less likely to buy anything. On the other hand, if you’re selective, and you are selective in who you target, then Google clobbers you for having a low click through rate. So, what do you do? Target everyone with vague or general adverts, and hope that the improvement in click through rate (and hence improvement in Quality Score and reduction in cost per click) will more than off-set the reduction in conversion rate (and hence cost per conversion)? Or run a tight campaign and get the best click through rate that you can? Part of the answer is in selecting the right keywords (mining the ‘long tail’ is a form of targeting people that are looking for what you have); and bidding on generic terms will cost you a fortune, but get you few conversions. But this can’t be the whole answer – for any given keyword, you can influence the click through rate by writing different advert text. For example, if you were selling expensive, high-quality widgets, then you could write either of the following adverts: Blue Widgets High Quality Blue Widgets Here! Exclusive Designer Widgets From £99 Blue Widgets Wide Range Of Blue Widgets Here! Great Prices, Free Delivery The second one may well get a better click through rate, but people looking for the elusive £49 widget will be disappointed – they wouldn’t have clicked on the first advert. On the other hand, some people would click on the second advert that wouldn’t click on the first advert, and then go on to buy (especially if the site is very good at ’selling’), so it may generate more conversions, albeit at a lower conversion rate. So which should you do? Is this just a question of testing, testing and testing some more? Probably yes, in the main part, though my experience tells me that in general, you’re better off being more specific and targeted, rather than trying to interest everyone, hoping for an improvement in your Quality Score. Of course, if you’re more than happy with your cost per conversion, and are looking for more traffic even if the cost per conversion is higher, this may offer an alternative to increasing your bids (which would mean that you’d pay more for all of your clicks, even the ones you would have got anyway). But this all raises perhaps a more general question – if Google is trying to reward sites that are relevant to the searchers, why is it penalising sites by encouraging them to attract visitors who are less likely to convert? Is it achieving the opposite of its stated intention? Perhaps it is to a point, but looking at it from the searchers’ perspective, they were unlikely to convert anyway – and the fact that they are unattractive traffic to the paying advertisers’ eyes makes them no less valuable in the eyes of Google. The basic premise that adverts with higher click through rates should be pushed into higher positions makes perfect sense in principle, but it seems to me that it makes a few big assumptions…
- That searchers can tell the relevancy of the site from its advert text.
- That advertisers don’t write misleading adverts to get more visitors to their site.
- That advertisers bid in such a way that their costs are less than their profits from the advert.
Allow me to explain. Points 1 and 2 are essential if you are to judge a WEBSITE by its click through rate – but if the advert doesn’t accurately represent the website, then it’ll get unwarranted clicks. As I understand it, Google will penalise adverts if the traffic immediately leaves the site (?), but I doubt that this off-sets these clicks completely. Point 3 is really highlighting the assumption that the better an advert’s Quality Score is, the higher it can afford to appear in the results. This is only true if they are monitoring their cost per conversion, and trying to set their cost per click to maximise their profits, based on the cpc and conversion rate. But I’m fairly confident that a lot of companies are still of the mindset “top position is good – be there if we can”. Their adverts may be poor, but I’ve managed a number of campaigns, and even with great Quality Scores, compelling advert text, well-designed sites and good products at good prices, they can’t get close to the top of the search results without bidding well over the odds. If Google’s model was working perfectly, then my advert would be top (or close to it, at least). The campaign is ticking all of the boxes, but the searcher is still quite likely to visit the other site. I’ve discussed this ‘numpty’ effect before, and it’ll probably come up again, but does this mean that Google are doing something wrong? Conspiracy theorists would tell you that what Google are really doing is pushing up the costs of advertising in the top positions, so that they can make more money, but as I’ve said, I don’t think they’d risk that if it resulted in poorer quality results. If people stop using Google to find relevant results, it’ll cease to exist. So what can Google do? Should a well-designed campaign run by an agency really outperform a home-made campaign, just because we know the tricks of the trade? Are we as much of a problem as the numpties? Or is everything fine? Google’s market share suggests that their search results must be fairly useful, and their advertising revenue suggests that they can’t be upsetting businesses too much. Is everything lollipops and roses, or are there fundamental problems with Adwords?
no comments