Search Marketing» Adgroups Blog Posts – Epiphany Solutions Digital Marketing Blog

When Good Adverts Turn Bad

Steve Baker — Fri, 12 Feb 2010 15:18:23 +0000

You’ve been testing new adverts for years, making ever-smaller refinements as you close in on that holy grail, the ‘perfect advert’. Suddenly, and without warning, your clickthrough rate crashes. You try tweaking various aspects of your adverts, rolling back to your last few versions, but nothing helps. Your advert is officially pants.

What happened? Where did it all go wrong? What do you tell your line manager/client?

Adverts don’t exist in a vacuum. Whenever your advert is displayed, it’s being placed amongst a number of other adverts, generally for the same product or service. It’s easy to think that the clickthrough rate of your advert is simply a function of its position, and the advert text. But unless you look at it in the context of the adverts around it, you’ll have no way to tell whether it’s good or not.

Here’s an example. You’re selling the new Apple IWidget for £249, and your advert says:

Advert 1:

Is this a good advert? Or would you be better saying:

Advert 2:

You run an A:B test, and both have a similar conversion rate, but advert A has a clickthrough rate of 5%, compared to just 3% for advert B. This is fine and dandy, until one of your competitors realises that their clickthrough rate has dropped, and spots your advert. They write one saying:

Competitor:

Suddenly, your USP isn’t so compelling – you are more expensive, and are telling the world. Your clickthrough rate drops sharply, and your advert goes from amazing to pants overnight.

Of course, the opposite is also true. You may have tested the two adverts and found that the second one works better, because your price wasn’t competitive. But the moment your competitors stop advertising their price, they open the door for you to do so.

Writing good PPC adverts isn’t just about identifying the right USP, and pushing that. The USP that works best may change, and actively tracking what your competitors are saying (or aren’t saying) can be the difference between having to react when your adverts crash, and dominating the PPC results.

The price-sensitivity above is an obvious example, but there is a small danger. You need to ensure that large changes to your adverts don’t negatively impact the quality of visitor to your website. I said earlier that the conversion rate was the same, but what happens to the average order value? It may make no difference, but it may be very significant. Here’s another example. This time, you’re selling Digital Cameras.

Advert 1:

Advert2:

Suppose that Advert 1 has a better clickthrough rate, and the conversion rate is the same. It’s a no-brainer, right? Clearly, the people that click on the second advert and likely to buy more expensive cameras, so the better advert may well be the second one.

This all sounds fairly unremarkable, but something happened to one of my campaigns a couple of weeks back, and I became interested in this whole question in a hurry.

We were advertising a good price-point, and were mopping up in the market that we were advertising in. A competitor came along, with a lower price-point, and undercut us, taking half of our conversions. But the conversion rate didn’t change, and the value of the conversion went up sharply, since our competitor was taking all the low-value traffic.

To cut a long story short, the campaign became a lot more profitable overnight, because our competitor was taking all the unprofitable traffic. This led to a secondary impact, as the improved average conversion value meant that we could increase our bids, and move up the search results.

We are now getting almost as many conversions as before, but our competitor has made our campaign a lot more profitable, by qualifying our traffic in a way that we could never do ourselves.

How Not To Delete A Bad Keyword

Steve Baker — Mon, 01 Feb 2010 11:32:59 +0000

In a nutshell, my approach to managing my Ad groups has always been to group similar keywords together, and if I see a subset of these keywords performing differently, split them into their own Ad group.

For example, if I was advertising Sony Digital Cameras, I may include…

Cheap Sony Digital Camera
Cheap Sony Digital Cameras
Sony Digital Camera
Sony Digital Cameras

… all in the same Ad group (Ok – I wouldn’t, but for the sake of this example, suppose that I did).

If, after a month, I noticed that the conversion rate on ‘Cheap Sony Digital Camera’ and ‘Cheap Sony Digital Cameras’ was much lower, then I would split out these keywords into a separate Ad group. I would then assign them their own bid and advert, providing that they had enough traffic to justify their own Ad group (if they were generating almost no clicks, or they were generating a lot of clicks, but no sales, I may simply decide to remove them completely).

In the above case, I decide to remove them. What’s the result?

If I was using Exact Match only, then I have no problems – the keywords are removed, and my adverts don’t appear for these searches.

But if I’ve got my keywords on Phrase Match, all that will happen is that when somebody searches for ‘Cheap Sony Digital Camera’, my advert will appear, assigning the traffic to ‘Sony Digital Camera’ (Phrase Match).

If you delete a keyword from an Ad group, make sure that you have no other keywords that will catch it on Phrase or Broad Match. If possible, add it in as a Negative Match, to be on the safe side.

Of course, if in the above example, I had decided to move the ‘Cheap’ keywords to a new Ad group, and reduce the bids, I’m liable to get the same result. If the Phrase Match has a significantly higher bid, you are liable to see the clicks appear against this, and your new Ad group with the lower bids will be irrelevant.

So whenever you move a group of keywords to a new Ad group, make sure that the old Ad group doesn’t end up picking up the traffic – use Negative Matching at Ad group level. If you are removing a keyword from the whole campaign, it’s probably best to use Negative Matching at Campaign level.

You need to be especially careful when you have a particularly important keyword in its own Ad group, on Exact and Phrase Match. If you need to split the two versions into separate Ad groups, then you have to put a Negative Exact Match in the Phrase Match Ad group.

Small Budgets And Big Keyword Lists

Steve Baker — Tue, 18 Sep 2007 14:55:16 +0000

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…

A:B Advert Testing, A Cautionary Tale

Steve Baker — Mon, 09 Jul 2007 15:39:00 +0000

The conventional wisdom on PPC adverts on Google is that you should look to improve the click through rate, as it is generally accepted that this is an important attribute in the Quality Score, which determines the amount that you need to bid to get a certain position (or how high up the rankings you appear for your bid, if you prefer). And this is probably true, and isn’t a bad idea. But it’s definitely not a good idea to focus on the click through rate to the exclusion of all else. The click through rate is an indication of how interested people are in your advert, but if your advert does not accurately represent the content of your site, you’ll be enticing traffic that doesn’t convert very well, and may be putting off exactly the people that you should be attracting to your website. This sounds like an easy thing to avoid, but it’s not quite as straightforward as it sounds. Suppose that you are a company that offers free marketing advice via a weekly e-mail that people have to sign up for. Your initial advert may read:

Free Marketing Advice Get Free Advice From Marketers Inc Free E-Mails Every Week MarketersInc.com/Advice

The advert does quite well, and gets conversions occasionally. But you’re concerned that the second line is a fairly weak call to action, so you decide to try something different.

Free Marketing Advice Get Free Advice Here! Free E-Mails Every Week MarketersInc.com/Advice

You run it for a while, and it doubles the click through rate, so within a day or two you bin the old advert and go forward with the new one. Then you look at the third line. It doesn’t really extol the benefits of the e-mails, so you try another line.

Free Marketing Advice Get Free Advice Here! Learn The Tricks Of The Trade MarketersInc.com/Advice

Even better click through rates, so you keep this one. But the changes in the second line may lead people to believe that there is free information on your website, rather than from a marketing company. Whilst you’ll get more traffic to your site, it’ll be of poorer quality. And the change to the third line reinforces this. But surely you’ll see a fall-off in the conversion rates, and keep the old adverts? Not if you’re changing your adverts as soon as one appears significantly better than the other, based on click through rates. Suppose that the campaign above starts out with a click through rate of 3%, then increases to 6% and 8%. At the same time, the conversion rate falls from 10% to 7% to 4%. Finally, assume that the cost per click moves from £0.30 to £0.28 to £0.25 If you accept a 90% level of significance, your results look something like this. 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.

A:B Advert Testing – Is Statistical Significance Over-Rated?

Steve Baker — Fri, 29 Jun 2007 15:53:37 +0000

On the face of it, probably a bit of a daft question. How can you be sure that your new advert is better than the old one, if you don’t wait to see if it’s statistically significant? And to an extent, that’s true. If you were to ignore significance completely, the moment somebody clicked through one of your adverts, you’d decide that it was the better advert, and bin the other one. It’s quite possible that only 50% of the time you’d select the better advert, and for every improvement that you make to your advert, you make another change for the worse, and you don’t get any overall improvement at all. But there’s a trade-off for statistical significance. Suppose that you have two adverts, one that generates a click-through rate of 5%, and one that generates a click-through rate of 10%. How long should you wait before you are sure the 10% advert really is better? If you get 30 impressions per day, it’ll take four days to be 85% certain (3/60 vs. 6/60 is significant at the 85% level). But if you want to be 95% certain, it’ll take eleven days (8.25/165 vs. 16.5/165 is significant at the 95% level). And to be 99% certain, it’ll take twenty days! So, in the time that it takes to run one test at the 99% level, you can run five tests at the 85% level. Clearly, you can get far quicker improvements in your overall click-through rate, if most of these changes are genuinely for the better. But what about the risks? You could select to keep adverts that are, in fact, worse than the existing ones (and you will, 15% of the time , any change to an advert will change the click-through rate; there are no ˜equally good’ adverts). But I would challenge that if an advert appears better at the 85% level, whilst it may be worse, the chances are very small that it’ll be much worse. So, if you run five tests in those twenty days, you’ll probably make one change for the (slightly) worse, and four changes for the better. Still an improvement on the one change that you’d make if you were determined to wait until you were 99% certain that you were making the right choice , this is advertising, not a clinical trial! Of course, this is a bit of an over-simplification. In reality, most of your advert tests will yield a much smaller return than doubling the click-through rate, and a lot of them will not be better than the old advert. The first point here is quite important , the smaller the difference between the two adverts (increasingly true once you’ve entered an ongoing process of testing), the longer it’ll take to get strong significance, and the less risk there is in taking the wrong option occasionally. For example, if you were getting 30 impressions per day, and had adverts with 5% and 6% click-throughs, you’d get 85% significance after 75 days, but even 95% significance is going to take 193 days , nearly three times as long. As for the second point, what if the new advert is performing worse than the existing one after a few days? It’s not significant, but, in a mirror of the argument so far, if it is in reality a better advert, is it likely to be much better? Is it worth waiting weeks to see if this advert, that’s probably worse than the existing one, is actually slightly better (remember that the smaller the difference, the longer it’ll take to be sure). Perhaps the time is better spent writing a new challenger, which may prove itself quickly? So what level of significance should you use? Personally, I’d say that 85% is probably sufficient, but I can see an argument for 90%. I feel that running a test for three times as long (as an 85% test) to get to 95% is excessive , yes, you’ll get it wrong less often, but it’ll take a lot longer to generate improvements, and lets face it, your rivals probably aren’t standing still! There is, of course, one problem that brings the whole process to a grinding halt. What if the two adverts are producing very similar results? It’s widely acknowledged that a small change to an advert can have a big impact, but more often than not, it has a very small impact. Everything stops until you get significant results, and the more similar the performance of the adverts, the longer it’ll take. The solution is fairly clear , sooner or later, you’ll have to stop the test. You can either keep the existing advert, since the new advert hasn’t proven itself, or you can take whichever is the better to date, regardless of whether it’s significant or not (this’ll be the better advert more often than not). I’d advocate the second option, although really, it doesn’t make much difference which you choose (since they are performing very similarly). An interesting claim , that under certain circumstances, you should take the advert that is performing better, regardless of whether it’s significant or not! So what process have we arrived at?

Decide before you run your new advert how long you are willing to wait for a result , this’ll depend on how long you’ve been testing (as you go on, the chances of finding a quick, big win decrease) and (obviously) how many impressions you are getting.
Set the advert live, checking regularly for significance. I’d recommend www.splittester.com, but any testing tool will do.
If, after a few days (longer if you’ve got little traffic), the new advert is worse than the old one, kill it, and write a new advert.
Once you’ve got 85% significance (or 90%, if you’re of a nervous disposition), keep the better advert.
If the deadline set in step one is reached without a significant result, keep the better advert, regardless of how small the difference is.