How to evaluate the effect of SEM on brand customers

Author: Huanghao graduated from Rotterdam School of Management to study financial investment. Baidu engaged in search engine marketing consultant work for two years, customer service industry includes automotive manufacturers, finance, consumer electronics, retail, it and other services, the main customers have Shanghai GM, Dongfeng Citroen, a hi car rental, Pacific Insurance, Sony ...

　　

Often listen to brand-type advertisers said: "I am on the search engine keyword ads How not to the afternoon on the offline, the effect is so bad!" Then in search engine how to calculate the effect is good, how to calculate the effect is bad? is the ad's offline time as a criterion for assessing the effectiveness of search engines? This article will focus on the evaluation of the effect of some discussion, Hope that some brand customers can discard some of the only applicable to the traditional media assessment methods, from the understanding of the nature of search engine ads to comb suitable for this kind of SEM advertising (SEM in this paper, the search engine keyword paid promotional ads) assessment indicators.

The most common mistaken ideas of brand customers on SEM effect assessment

Advertisers in search engines can usually be grouped into two main categories: the first category is direct use of the network platform directly to achieve online sales, such as tourism on-line booking, E-commerce, car rental and so on, such advertisers are called effect class; the second category is no direct online sales platform, using the website to do brand dissemination promotion, such as car brand official website, Luxury official website, cosmetics official website and so on, such advertisers are called Brand category.

Some of the brand advertisers due to long-term reliance on traditional media development, the effectiveness of traditional media to assess more understanding, but the search engine this kind of new advertising platform relatively unfamiliar, so in the examination of SEM ads will also apply some traditional advertising, hard wide thinking. This will form a certain misunderstanding, the most common is to advertise the time to examine the effect of SEM advertising.

As described at the beginning of this article, many traditional advertisers and their advertising agencies often pay much attention to online advertising time. Some advertisers will ask for at least a few hours online and so on. In SEM advertising, in fact, online time should not be an important indicator of assessment. Reason to start from the principle of SEM advertising: SEM advertising the fundamental mission is to specific keywords as a bridge to help advertisers to bring the site accurate traffic; The cost of SEM advertising is to charge a certain fee for each of the traffic with a precise flow. So this kind of advertising effect assessment is to surround the search engine with how much precision flow and the cost of each flow, and the ads on the search engine to show how much time should be ranked after the flow. To give an extreme example, an advertiser in the search engine on the launch of an ad, the creative writing is very poor, and users search the key words of the relationship is also far from the results of 24 hours and no one to click. If we use advertising online time to assess, this is undoubtedly a perfect advertisement, the advertisement was placed for 24 hours without a penny; but the brutal fact is that this is a bad ad, not to The Advertiser's official website to bring any traffic, the search engine system will make the same judgment, and may not display this advertisement in the future. Therefore, the advertising online time to assess the SEM advertising is one-sided.

Advertisers who take the time to examine the effects of the ads often have to adjust their ad online time because of budget constraints. That's what we need to do to analyze this side. For example, a car advertisers bought "about 100,000 to buy what car good" this keyword, and its budget only enough to cover to 3:00 to the line, but the advertisers feel that after 3:00 ad online for him more important, so delayed the ad on-line time (the default online time is 7:00) to the noon 12:00. Let's take a look at the time-sharing distribution of the number of searches for this word (see Figure 1).

　　

Figure 1: Weekday and weekend "about 100,000 what car to buy" time-sharing distribution of search quantity

According to Figure 1, the search volume of this word has a trough, three peaks: the early morning 2:00~7:00 for rest period, the amount of retrieval is very small; morning 11:00~12:00 before lunch to reach the first peak, afternoon 15:00~17:00 before work to reach the second peak (this crest in the weekend did not appear); At night 20:00~22:00 reached the third crest. So, what time do we have to show the limited ads to the people? The answer is uncertain. Should advertisers distinguish between a midnight up search "around 100,000 to buy a car good" and a person to search the word before work? In common sense, it should not be treated differently, because these two people are likely to become advertisers ' potential customers. Unless the advertisers in-depth analysis of the different time period of the crowd from the SEM ads into the website after the depth of browsing, if it does prove that the depth of browsing is not the same, we have reason to adjust the display of ads, otherwise it is very subjective behavior. Figure 1 tells us that advertising at peak times is likely to cost the budget a lot faster.

The test logic of SEM effect on brand advertisement

So exactly how to evaluate SEM advertising? We may as well comb the SEM ads to help advertisers bring the user's process.

　　

Figure 2: Typical behavioral path map for potential car buyers

As shown in Figure 2, the behavioral path of the entire potential car buyer can be summed up as: "Requirements-Search-click-Browse-Decision" five big steps. With SEM ads are associated with the demand, search, click, browse four major aspects; SEM advertising can play a central role in the search and click, and the user through the SEM into the official website after the browsing behavior of the main advertisers and the quality of the site, product quality is related. The brand Advertiser can follow this behavior path to examine the SEM effect. Below I will be in accordance with this idea to elaborate the SEM effect examination way.

1: Demand

Take the potential car buyers for example, demand is divided into two categories, the first category is to see the traditional media advertising, to stimulate their special interest in a particular car, so through search to learn more; The second category is that they have to buy a car needs, want to search engine for help to find to meet their own needs. In view of this, the first step to examine the needs of SEM is to see whether the selection of SEM is in place, whether it covers all the needs of advertisers.

　　

Figure 3: Population analysis behind different keywords: bubble size indicates retrieval size

As shown in Figure 3, we see a lot more people searching for words like "about 100,000 cars" than those who only search for the June-Wei or Buick brands, although its interest in June Wei is not high, but the urgency of buying, the advertisers to fight for the original brand is completely not interested in the new customers are very good. SEM account contains this kind of keyword is our test of SEM launch of the basic standards.

2: Search

When the user needs to reflect the key words are covered by the account, followed by the user search, the Search engine results page ads show. The most important measurement of this quantitative indicator is the amount of presentation (impression) and the percentage of exhibiting (impression Share). Show the amount of advertising show how many times, but can show the proportion of the advertisers themselves to show the amount of advertising in the industry as a percentage of the total display. These two are the index to examine the effect of SEM, but should not be the core index. The reason is also explained in the above, because the core value of the SEM is to bring accurate flow to the advertisers, if only see not point, the significance is not big. In addition here to mention CPM, that is, thousands of indicators to show the cost, in fact, in search engine ads, the more attention to the ads, the more clicks, so thousands of display costs higher; and those unpopular ads, no one, natural CPM low. So it is of little significance to evaluate SEM with CPM.

In this category, advertisers should pay attention to their ads are displayed on the left or right, the left side of the ads are at least 20 times times the right, the degree of attention is also very different, as detailed in Figure 4.

　　

Figure 4: Search engine focus on hot-try

3: Click

Click is the most important index to examine the effect of SEM. The number of clicks indicates that the SEM advertising to the main advertising to bring the amount of traffic, should be all brand-type advertising main assessment of the core of a value.

Click Cost is also an important indicator, it determines the main advertising for each traffic pay how much money. So what kind of click cost is reasonable. This has to be examined from two aspects: first, this is to carry out a horizontal comparison of the industry, to see the average level of the industry, and then compare their own SEM account level; second, different parts of speech need different standards, such as the general word of the click Cost is naturally higher than the brand word click Cost, this is also very good understanding, The person who searches for the brand Word is the advertiser to be affected by other media channel before the person and make it interested to search, while the search for generic words is directly from the search engine to get the new customers, pay the price naturally smaller. From the search engine mechanism to understand the problem is that the general word is a lot of advertisers can compete, and the brand word system will tend to protect the owner of this brand advertisers.

Click-through rate is a relative indicator, also can not across. There are two factors that affect it: one is the matching method, the same situation, a wide range of matching hits certainly far below the exact match; the second part of speech, the universal word click rate is certainly lower than the brand words. The quality is a very good test of sem effect of the method, the proportion of high quality words can be used as an important account assessment criteria.

4: Browsing behavior

Later browsing behavior, in fact, with the search engine relationship is not too big. Because the search engine put the user into the website, after the Netizen's entire browsing depth depends on the product quality, the website quality and so on many advertisers themselves factor. But these browsing depth's performance actually also reflected to some extent the search engine to carry the traffic quality. Usually the main two indicators of the browsing depth are residence time and number of browsing pages.

ROI Assessment method for brand-type advertising main SEM

In recent years, brand-type advertisers to search engine assessment has approached the effect of customer trends. For example, there are car advertising owners to the SEM can bring the test drive number (Leads) and the cost of each test drive (CPA) assessment. Whether this is a sign that brand customers increasingly understand the search engine is worth further discussion. I personally think that the use of leads and CPA assessment or to separate parts of speech, the requirements of general words should not be so high brand words. In fact, since it is advertising, there is the role of publicity, whether the immediate impact of their purchase decisions should not be used as the core of advertising assessment standards. Because this product itself may also affect the user purchase decision, and many advertising functions have a certain delay, and does not necessarily immediately have an impact.

FMCG industry has formed a set of assessment of advertising on its product sales impact of the method, I think is more objective and realistic reflection of the role of the media theory. This method is to use a statistical approach, to take a linear regression model to analyze the various independent variables on the impact of the sales variables. This is called marketing Mix modelling, as shown in Figure 5.

　　

Figure 5:marketing Mix Modelling

In this model the sales volume of the product is dependent on the variable, which is the result. And all the reasons that may affect this result are divided into two areas, the first aspect is not related to marketing, such as the size of the product pipeline (such as the length of the shampoo product in the supermarket container), macroeconomic factors, these reasons for the formation of sales is base; the second is related to marketing activities, Simply speaking, it is the amount of the media and various promotional methods, these reasons for the formation of sales is incremental. This model has been accepted by some non-FMCG advertisers, such as some car manufacturers, using the model to analyze the impact of different marketing methods on their sales. If the SEM advertising industry as an independent variable, it can more objectively reflect its impact on sales. But use this model to pay attention to two points: first, how to consider all the reasons for the formation of base, the second is the delay effect of advertising, for example, September sales may be corresponding to the July of the amount of advertising input.

In general, for most of the brand-type advertisers without in-depth examination, in fact, the SEM advertising is not so complicated, the number of clicks is the most need to check and the core of the evaluation index.