Optimising your paid search copy can not only improve the performance of campaigns; it can also provide valuable research for your other marketing strategies, writes Nathan Levi.
It is widely accepted that search engine marketing is not ‘push’ advertising. However ‘creative’ a 25 character title and a 70 character description might be, it cannot ‘grab’ in the same way as a multi-coloured press ad or billboard can.
This is not to say search engine marketers should not and do not experiment with creative messaging, but the purpose of doing so is not so much to ‘grab’. If used effectively, copy optimisation can not only help to improve campaign performance, but also provide valuable market and customer insight which can be fed into other marketing strategies.
As a foreword, the analysis outlined below is Adwords-centric for three reasons:
- Pre-AdCenter, Adwords was the only platform which allowed rotation of more than one text ad against the same keyword (the imminent launch of Panama in the UK will accommodate copy rotation in the same way).
- Google’s market share is so large that it allows you to build up statistical relevance much quicker than the other search engines.
- Thus far, Adwords and its accompanying editor allows for a speedy technical set-up which means multiple text ads can be uploaded in a matter of minutes.
This being said, it should not be assumed that better performing copy on Google will translate across the other search engines. In the US, where market share is more even between Google and Yahoo!, studies have proven that better performing ad copy on one engine can be poorer performing on the other, and audiences between search engines can differ for certain vertical markets significantly.
Before I lay down some case studies, there are some technical considerations that should be taken for effective copy optimisation implementation. First, a pre-requisite is to track each text ad to conversion/s. An ad server or Adwords’ conversion tracking will enable you to do this. While some bid management tools will automatically optimise copy, a more hands-on approach is recommended since auto-optimisation removes the most essential benefit of copy testing - the qualitative insight and analysis.
Second, text ads should be rotated. While Adwords defaults ad serving to optimise click-through rates, it may not be cost effective to drive large volumes of traffic to your website, particularly from browsers or those early on in the sales funnel. Furthermore, the ‘optimise’ ad serving option often ‘learns’ too quickly and will not always build up enough statistical relevance to be a fair test. This is why it is recommended to select the ‘rotate’ option which shows ads more evenly, until enough data has been recorded for you to implement a rules based programme.
Third, you needs to consider which keyword to optimise copy against. This is where it is important to benchmark exactly what you are trying to achieve from your experiment; copy optimisation should stem from a question or an assumption.
On a general level copy should be tested against high traffic driving words, either branded or non-branded. There will be a cost implication on testing against some high value non-branded keywords so there may be a need to set a budget cap as the benchmark for statistical relevance.
Here are some case studies to demonstrate why copy optimisation is important....
1. This first case study highlights how copy optimisation can be effective in increasing ROI. We tested the following two messages against high traffic driving non-branded keywords for a month:
Description (a): Discover hundreds of healthy eating tips in our recipe book.
Description (b): Save £9.95 on our healthy eating recipe book.
We found that while (a) drove twice as much traffic, (b) had a much higher conversion rate. While relevancy was lower, it was still a lot more profitable to run with the price driven message. By including the price in the copy, one consideration in the buying cycle has been brought to the searchers’ attention before they click. In other instances it is better not to pre-qualify traffic in this way as rules will change for different brands or markets.
2. In this second case study, we tested the following two messages against non-branded high value keywords for a month:
Description (a): Compare and select from hundreds of insurance policies online.
Description (b): Get a quick online insurance quote now!
We found in this case that searchers were more likely to convert when they felt they were being offered choice rather than forced into taking an action. More importantly we found that whether or not searchers actually searched on a ‘quote’ related term they were still more likely to convert with (a); the semantic relationship between keyword and copy is not always as clear cut as it might appear. This contradicts Google’s recommendation to always include the keyword in text ads; often the term searched for is not what is meant.
In both examples above, the ad offering consumers choice drives the most traffic. In terms of delivering conversions, the results were very different. It is important also to consider that search will not always be driving the consumer to an end point. Efficiencies might be made by re-targeting searchers from the high traffic-driving (a) messages as search should always be evaluated as part of a wider media mix.
Paid Search is potentially the only advertising medium that allows you to build up statistical relevance quickly and transparently enough for you to make informed decisions about your products. This can be achieved by running separate promotions in ad copy concurrently and seeing which one consumers respond to. This can also be tested at a more subtle level.
3. We ran the following two messages, which were the same promotion but sold differently, against high value non-branded terms for a month:
Description (a): Get 10 months for the price of 12 when you buy online.
Description (b): Get 2 months free when you buy online.
Whilst there is only a very subtle different between these two messages, (b) drove a significantly larger volume of conversions than (a). The results of this test then fed into the brand’s other online and above-the-line campaigns. The question one might ask is why does description (b) perform better than (a). The answer lies in what competitors were offering at the same time. The ’12 for 10’ or ‘15% off’ messages were more frequent in this paid search landscape than ‘2 months free’ hence competitive analysis can be key to effective copy optimisation.
4. Paid search campaigns can have more than one set of KPIs since copy optimisation can be instrumental in delivering multiple objectives as our fourth case study shows. We ran the following two text ads on a branded term for a retail client who was both trying to improve footfall into stores and drive online sales:
Description (a): Book an appointment online to meet one of our consultants.
Description (b): Huge discounts on a range of products! Order online.
As expected (a) had a much lower click-through rate. However, its store appointment conversion rate was four times that of (b), which drove twice as many online sales as (a). This retailer then had the option of up-weighting appointments or online sales depending on which drove the highest profit. In this way copy optimisation became central to this campaign’s branded term search strategy. This highlights the benefits of copy testing against branded terms.
Copy optimisation is one example of paid search not only working as a pure direct response mechanism but also as a very powerful market research tool. By testing different messages it is possible to drive efficiencies and challenge assumptions about your brand, products and consumers.
Nathan Levi is head of search at DNA.