Several test steps for SEM keywords without transformation

Hewei in the article "Indirect transformation" commented: "If the customer in addition to brand words have transformation, other words are not transformed, then do?"

　　

This question reminds me of a blog I have seen before, titled "100% Guaranteed loss account two weeks to improve ROI 50%". The author's main idea is: If your SEM account two weeks consumption of 500 yuan, generate income of 100 yuan, it means that the loss of 400 yuan. Want a quick stop? No problem, put your account stopped, send 500 yuan to me, two weeks I promise to give you 300 yuan, so you only lost 200 yuan, 50% of the improvement Ah! Bowen used the tone of irony, point out the problem is very practical: there are many customers in the knowledge of the loss of the case to continue the SEM launch, At the same time also suffer from the inability to achieve ROI indicators, which in itself is contradictory.

"Key words are not translated" is a certain universal problem, I would like to list some of the possible test steps I have encountered such problems, most of them believe that Hewei has been done, but for other readers may still be meaningful.

1, check the tracking means. If there is a "completely without transformation" of the normal cognition in the SEM advertisement, the first thing to eliminate is the technical obstacle. The effect evaluation of network marketing is based on the effective tracking. or through JS to write cookies, or in the Web site log read reference, are through the display-click-Results of the process tracking to achieve. In the case of Hewei, there are no other words in the brand name, and you need to make sure that the tracking deployment of the brand word and other keywords is consistent. If the tracking method is not a problem, go to the next step.

2, check the landing page. Most customers have a different approach to branding and other keywords, but landing pages are common. Brand words tend to point to the home page, while other keywords may point to related pages. Is there a problem with the loading of the relevant pages? Is there a problem with the process of arriving at the purchase page from the relevant page? In practice I have not encountered the problem of non-brand keywords landing page, but some of the keywords landing page and the next level of the problem is not uncommon, This is often caused by adjusting the site structure and URL format. Thousands of pages in a large web site often involve different departments, and adjustments to certain departments or functions may have a ripple effect. If the site structure is not scientific URL format is not uniform, adjustment will be prone to errors, resulting in some page failure or subordinate link failure. If the check finds the landing page, continue to check the conversion page.

3, the conversion page is actually a relatively easy to go wrong link. One may be the conversion of the payment link itself, unable to complete the transaction; another may be the conversion payment itself is no problem, but the payment completion page loading problems, resulting in JS can not be effectively written to cookies. It is also easy to distinguish between the two, since the former can be reflected directly in internal income movements, which are only monitoring failures, and real and reported income gaps. In the question of Hewei, brand words have been transformed and other words have not been transformed. We usually think that if the problem is converted, we should have all the keywords together problem-– here is the premise of the entire site has a unified transformation page. However, many sites will use different conversion pages to record different KPIs, such as different brands/Registration/download/installation/purchase/return/recommend/Add, and so on, may have a separate conversion page. If you are using a third-party payment, there may be a jump problem between the different domain names. So look at the specific situation to determine whether the conversion page is a problem.

4, all the above three excluded, that is, the technical level have been checked, is not the brand keyword does not produce transformation. Then compare the branded words with the non-brand words, compared to the display, than the click. Generally speaking, the conversion rate of brand words is necessarily much higher than the non-brand words, because customers in search of brand words have a stronger incentive to buy, skipping the purchase funnel in a long period of search behavior. This is why the SEM management should be the brand word and non-brand word separation management, the ROI standard of brand words should be different. (about brand Word management, I have discussed before, please see here.) It is probably reasonable to assume that brand words have 100 displays, 50 clicks, 10 conversions, and not brand words with 100 displays, 10 clicks, no conversions. Brand Word is 20% conversion rate, the non-brand Word may conversion rate in 5% or even lower, to 20 clicks before the conversion will occur ... If the brand name is not the overall display of the situation, the temporary need not to tangle why did not transform, should be through various means to increase the display of non-brand words and clicks, accumulate more data.

5, before spending more money to buy shows and clicks, you need to do a sanity check. That is, the launch of the non-brand keyword is at least reasonable. If the customer is selling QQ currency, but the key word is "washing machine", or put the advertisement is "quick to participate in the ipad lottery, click on the award", which obviously does not help to bring transformation. Do not think this is a myth, the example may be slightly exaggerated, but a similar situation is real. Many customers put the first keywords, the first advertisement is to pat the head out. We cannot completely rule out the possibility of conversion from low relevance keywords, but they are not suitable for the first batch of launch. At the beginning of the SEM launch, you need to focus on the core keywords and the most relevant ads. Can be through the customer's website, product analysis to obtain these keywords and ads, can also be from the launch of competitors to steal some of the most relevant. If you see enough "crazy" keywords or ads in non-branded keywords that don't get converted, you should remove them from your account to avoid waste. Only the keywords we expect to bring in returns are worth the extra effort or more display and click.

6, if we see the transformation on 20 or even 50 clicks, then the remaining question is how to get the right conversion cost. But if you still can't see it? We often have to face the unknown in SEM management. Obviously, with the existing data, there is a basis for optimizing decisions. Can be in the promotion of a new product, a novel advertising words, a batch of new keywords, how much cost should be invested? Before getting the first transformation, all the input is loss, learning costs, sunk costs, we have to prepare for losses, the problem is how much loss. Suppose a conversion is 100 yuan, we should invest 100 yuan, 200 yuan, or 500 yuan, 1000 yuan cost to obtain initial data accumulation? Unfortunately, without fixed pattern, the amount of learning you intend to invest depends entirely on your business decisions. No matter how much the amount is, a tolerable cost limit must be set. When the learning cost of the non-brand word exceeds the upper limit, it will shout stop. Choose the relevant keywords, not the possibility of conversion is not high, but the possibility of not making money is too big. In "Lao Zhang's Moving Company" a article, I discussed, small individual moving company Baidu Phoenix Nest launch may not be cost-effective. Whether it is brand keywords or non-brand keywords, whether there is a transformation or no transformation, as long as the plan to implement a thorough examination of the SEM test proved not to make money, then do not play.

Word, if the SEM delivery can not produce benefits, should not continue to launch. Business behavior is not a charitable activity, money is to earn money, not to make money things must not do. Before you come to this conclusion, you need to identify the reasons for not producing benefits, because the implied premise of this conclusion is "the implementation of the SEM is correct", I think this is the focus of the Hewei problem.

Original address: Http://semwatch.org/2011/03/non-conversion-keywords