Jan 102008

PPC Copy Changes Can Hurt You

Success or failure in PPC advertising depends on a number of factors: the keyword list, the bidding strategy, the landing page choice, the selection and price of the offerings, the ease of shopping the site, the reputation of the company, market forces in the economy at large and more. As I've mentioned before, the ad copy is a smaller factor than any of the above, yet often receives disproportionate attention because it's visible and relatively easy to manipulate.

However, tinkering with copy too much can cost you. Here's how:

  • Quality Score Erosion: The three biggest components of quality score are Click-Through Rate (CTR), bid, and historical performance (CTR) for the account. Landing pages are not a factor for legitimate retailers. The CTR measured is based on historical data for that combination of keyword and copy. Interestingly, the data is not retailer specific; if Google knows that the combination of a keyword and copy block have not performed well for one retailer, they expect it not to work well for others, either. Every time a retailer changes their copy, they lose the Quality Score they've built up on the old combination. If the old combination was doing better than average, the change will immediately either increase CPCs or lower position on the page, or both. If the new combination turns out to be better, great, but if not you'll never recoup this loss. It's very difficult to materially improve copy after the first few tests, so the odds of doing harm are pretty good.
  • Hanging Offer Copy: You can minimize the Quality Score damage by running the new copy as an additional version, rather than a replacement, but that can result in two other problems. First, the new copy is randomly interchanged with the old copy, so if the copy change was dictated by the "corner office" this won't satisfy the boss. Second, it's compelling to constantly roll out new versions on the theory that the best version will win out, the problem is if offer copy is in the mix it is all too easy to leave those offer text blocks running after the offer has expired. That can really get you in hot water.
  • Opportunity Costs: This is often the most costly piece. Many firms get so wrapped up in frenetic copy changes that they stop paying attention to those other factors that have far more impact on the campaign's performance. Careful analysis of the data may not be as easy to demonstrate to the corner office, but the benefits that flow into smart, subtle adjustments to bid strategy, scrutiny of search strings and match types, and extra attention to the performance changes going into and coming out of a holiday generate much greater return on the investment of time.

When does it make sense to change and test copy?

  • We advocate testing copy blocks early to find the most compelling message for each category of product. Remember that you're not trying to sell the products here, you're trying to sell your website as the place to shop for the product. Periodically you might try to "beat the control", but pretty quickly you should find that that becomes very difficult. at that point your analysts time is better spent on other projects.
  • Test the merits of changing copy to reflect offers against running the standard copy while offers are happening. Note: you have to run the same keywords in different campaigns to make this work, using Google or Yahoo's copy testing system won't work. You'll find that the offer copy generates higher CTR and that will either lower CPC or raise position on the page, but you'll also find that conversion rate and sales per click drop. The quality of the traffic will be lower because offer copy attracts "bottom feeders". Whether the additional, cheaper CPCs plus lower quality traffic pays off will depend on the retailer and the promotion, but it should be tested. Don't assume that because the CTR is higher offer copy is better, we've found it's often a wash.
  • If testing reveals that offer copy does help performance during offers, then roll out several different versions to find the best one. More importantly, make sure the offer is sufficiently wide-spread to justify the time spent and avoid the opportunity costs referenced above.

Some bad reasons to change copy?

  • Because the overall performance of the program is heading south. If the big numbers are moving in the wrong direction the problem is not the ad copy, it's bigger than that. Look to your bidding, keyword lists, match types and landing pages to find the real problem.
  • Because you're "tired" of the old copy. Remember most of the folks who search for your products have never seen the copy before. They're not tired of it, and they're the ones that matter.
  • Because you don't know what else to do. Here are some higher value PPC projects for you to work on.
  • Because it's important to look busy. We think it's more important to generate results, and you're more likely to make progress by analyzing data and adjusting the big levers as the analyses suggest.

Your time is valuable, use it wisely!!!



17 Responses to "PPC Copy Changes Can Hurt You"
I agree with the overall points you make about testing ad copy early and not making too many changes at the expense of doing something more worthwhile. I’m not so sure about the quality score impact though. My understanding is that quality score is tied to keywords not ad copy. It’s the click through rate for the keyword (combined with the other QS factors) that makes up quality score not the combination of keyword and ad copy. This makes sense because otherwise, if there are several ads for an ad group, each keyword would have multiple quality scores. There’s already a different QS for minimum bids and ad rank but this would make it even more complicated. The adwords help centre says that “Quality Score for ad position is determined by a keyword's clickthrough rate (CTR) on Google”. No mention of keyword / ad copy combination. I’m not trying to be a smart ass here. I’m raising it because of the number of comments I see on forums asking whether you lose QS history if you change ad copy. I’ve seen responses along the lines that people are scared to change anything for fear of losing quality score. There are legitimate reasons to change ad copy on a fairly regular basis. Not just to test different messages but to promote seasonal offers for example. Obviously it’s important to know whether this has a detrimental effect on the ad position / cpc. My personal observations don’t back up a drop in QS when ad copy changes but I’m not 100% confident in this to determine the right answer. What’s the basis for the claim that “The CTR measured is based on historical data for that combination of keyword and copy”?
Hi Christine -- Thanks for your long comment. IMHO, George isn't advocating never changing copy; rather his post argues against running in circles changing copy too frequently. As to your point My understanding is that quality score is tied to keywords not ad copy. Copy is involved in QS. Indirectly, copy affects CTR, which is a large component of QS. http://adwords.google.com/support/bin/answer.py?hl=en&answer=21388 Directly, I've heard Googlers explicitly state at conferences that copy is a component of QS, so that is public knowledge. I couldn't find a link from Google itself for that on the public web, but that is confirmed ("New combinations of keywords, ad text, etc. lead to different Quality Scores") in a private Google agency powerpoint ("Ads Quality: An Introduction and Key Concepts, July 2007", slide 14). Focusing on the exact internals of this month's version of the QS algorithm isn't as important as focusing on good marketing and focusing on the user. Write compelling ads in the first place, use smart testing to improve them, and change copy when it makes sense (see George's "when does it make sense?" paragraph). Yes yes yes. But I agree with George: over-emphasizing the importance of copy can waste time, erode QS, and distract from higher-leverage PPC activities. Cheers Alan PS -- My comment here assumes you have decent copy in place as your control. If you have bad copy, that will harm you a great deal, and do work on it ASAP. But the economic benefit of "perfect" copy (whatever that means) vs. decent copy is typically darn small.
Hi Christine, The last time our Google reps visited us we grilled them on this very topic. This is the explanation we got straight from them. There are many heuristics involved, but the ranking is something like: keyword-copy combined history (across retailers), keyword history (for this account), account history. You're right, that people can worry too much about QS impact and never make needed updates for seasonality. On the other side though, lots of folks flip copy for the sake of flipping copy, and that's a poor and potentially costly practice.
Hi Alan and George, Thanks for both your replies. As I said I do agree with the points made about spending time on what is most effective. I also appreciate that ad copy is used when determining quality score, specifically the relationship between keywords and copy. It was the specific idea that changing ad copy resets the quality score that prompted me to comment. Especially as this is pretty important to know one way or the other. It doesn't exactly fall into the category of getting too involved in minor details of the inner workings. Some good points in the replies. Yours is one of the sem blogs I always read. All the best for 2008 Christine
ET says:
So, does changing your bids changed your quality score?
No, Quality Score replaced pure "Click-Through-Rate" as the other factor besides bids in the position determination. Where the ranking used to be determined by (bid x CTR), it's now (bid x QS). QS is in turn determined by all those other factors. Hope that helps! George
Linda Bustos says:
What if your changes to your ad copy are very minor, for example keeping the same text but capitalizing the first letter of every word (which resets the ad impression count) or adding a keyword to the display url after the .com/ or removing the www from a display url? Does this have the same impact on quality score?
Linda, super questions! The display url does not factor into the Quality Score, so changing that shouldn't hurt. I'm not sure about capitalization. I'm betting that changing the capitalization doesn't affect Quality Score, but I'll check with our rep to make sure.
Ouch! I was wrong on BOTH counts. Here's what our rep says: "Any change to the ad text indicates to the system that there is a new ad. Whether the change is as small as capitalization or Display URL or as large as a completely different copy, the system still views the ad as new. Any new ad has to establish a quality score in connection with the keywords in that ad group. That said, it shouldn't take the system very long to develop this quality score, and for minor changes, you'll find that the quality score ends up about the same as it was prior to the change. During this ramp up time, you could see a dip for only a few minutes or hours on high traffic keywords, and perhaps a few days on low traffic keywords." Again, the QS impact isn't the biggest issue, and certainly if you're fixing capitalization or adding helpful specificity to the display url, it's probably worth taking the hit. But if you're tweaking these things all the time, you might be hurting yourself both in terms of QS and more generally by pulling time and attention away from the bigger levers in search.
Bob says:
I feel as though this blog was tailor written about me! :-) Any proposed changes will have an initial, negative impact on performance. I don't think anyone disagrees with that premise. The larger question is whether that loss can be amortized, recaptured and potentially surpassed during the offering period? If it's a true test for perpetual or long term change, it makes perfect sense, however if the change is for a short-term offer, your outcome scenarios may be break-even, minor negative or major negative. There's a cost in making the change and when the period to benefit is limited (such as in a seasonal offer) and the outcome is not assured (it may be surprising to learn that copy for cheaper stuff does not guarantee better CTR's) all of this needs to be carefully weighed. As George said, there are other things going on in your campaign that require attention so, unless you have unlimited SEM resources (but don't we all???), something is going to get neglected for a time. This is an additional leakage which is probably never factored into the evaluation. If these changes are for a potential, long term benefit, it is the right course of action. Just knowing that an alternative message does not perform better is worth cost of implementing a test. Nothing ventured, nothing gained in this regard. Develop and plan your proposed changes carefully...THEN STICK TO YOUR PLAN (unless there are overriding circumstances). If I didn't know better, I'd swear this was written about me!
Bob, it's not about you...it's about Tania. Just kidding, I think your advice is right on target, QS concerns shouldn't prevent someone from testing a good idea, but should give the churners pause. As we know, there are plenty of other issues relating to merchandising, pricing, offers and execution that merit attention as they can have significant impact on customer retention and lifetime value. -- Grandpa Joe
Linda Bustos says:
Thanks for the follow up George :D As far as I've heard, making those changes to the display URL and capitalization greatly improves click through so I imagine it's worth it even if the positions are lower. ??
Absolutely, Linda. The hit to QS will be short-term, and if the improvements help CTR you'll have a better QS in the long term. The higher traffic keywords will have the shortest downturn -- maybe just a couple of days. The Low traffic terms will take a little longer to recover, but never worry about QS if you're sure you're making improvements to the copy.


Check out what others are saying...
[...] PPC Copy Changes Can Hurt You, Rimm Kaufman [...]
[...] PPC copy changes can hurt you: at least that’s what the Rimm-Kaufmann blog is saying! This post does a great job of explaining why too much copy-tinkering can be a bad thing, and offers suggestions for the right time to make changes.   Don't forget to subscribe to the PPC Hero feed!       Related Posts: [...]
[...] PPC Copy Changes Can Hurt You [...]
[...] PPC Copy Changes Can Hurt You [...]

Leave A Comment