Online marketing to monitor the effect of offline operation

The "Preface" is one of the most frequently asked questions on our QQ group, and is in fact one of my clients ' frequently asked questions. This is a very complex issue, not to ask the question itself, but to solve the problem of the feasibility of the practitioners have been plagued. This article will discuss several types of online marketing impact on offline actions and current solutions and remaining problems. This is the topic of the last article, see the following: Online marketing to the offline action effect monitoring (next).

Body】

Today's topic is a new challenge, frankly speaking, I know that in this field is limited, and has always been afraid, but this is an unavoidable topic, if you want to do online marketing, your boss will ask you what the end result. This effect, for many of the business models that are not directly online transactions, actually means that the audience is affected by network marketing and action taken online. This is not to be monitored by web analytics tools alone. As a result, practitioners in the online marketing community have tried various ways to try to solve the problem and will continue to explore it in the years ahead.

This article is not able to sum up their exploration, and similar to the previous article, but only to talk about my understanding and practice, hope to cause everyone's thinking and discussion.

Classification of

Online marketing effect on offline operation

Online marketing to offline action produced by the effect of a variety of types, according to the impact from the comparative primary to very strong, can be divided into four levels, as follows:

Product impression (awareness) to produce goodwill/potential purchase possibility (Preferences) Submit purchase intention, trial, or call Advisory/Enquiry Telephone (Leads) generate actual offline purchase behavior (Purchase)

In general, the ultimate goal of online marketing (Digital Campaign) must be to achieve purchase, but purchase implementations sometimes require a long period of time, and the whole loss from awareness to purchase, It would be nice if campaign could achieve leads or preferences, or simply impress consumers (awareness). In this way, people's evaluation of digital campaign effect is not limited to viewing the final actual purchase, but also to the whole process from awareness to preference to leads.

Effect monitoring of

awareness

Awareness refers to the audience's awareness of the brand/commodity advertised. Based on my previous experience in agency to monitor digital campaign effects, awareness monitoring relies mainly on advertising exposure data (impression). However, impression is a very weak metric (at home, I have expressed "contempt" for this measure many times), it is difficult to bring any valuable insight, so we need some better ways.

Here, I would like to thank my former clients, Kai men and Xu Fei, they have put forward a very good solution in this field, but also special thanks to Tenly (Shangjian) classmate, he put forward a very good view of these methods. So, here's a summary of the following methods:

advertising Impression/click Number of ads to the site visitor number of soft Wen total read number (soft PV)/evaluation/response number of products or brand keyword Search volume growth rate of products or brand keywords of the search engine results of the growth of the range social Media Sharing/Forwarding Quantity survey Questionnaire direct statistics Word-of-mouth monitoring

These methods are somewhat easy and somewhat difficult, but they are indeed awareness evaluation methods that can be implemented in practice at present. Among them, advertising to the site of the number of visitor I have been more attention, but, because of the existence of crazy cheating, this value is now a serious challenge. However, the network marketing effect analyst cannot be baffled by this, Quweicunzhen is our work. Therefore, we should change the number of this Visitor to qualified Visitor (Visitor). Qualified Visitor = Campaign patterns visitorxbounce%xstay longer than 30sec%xother criteria.

The total number of soft text reading is another commonly used evaluation method, soft text for many advertisers, is a cost-effective method. The total number of readings is the amount of soft page view in addition to the soft text release media, as well as all other reprinted total Page view numbers. However, interestingly, I suggest another quantity--------------soft reading number = Soft Text total read number-soft text on the original release platform to read the number. Why use this quantity, cough, because the media itself that data is really can't believe Ah, simply do not believe. How does the total reading of the soft text count? I use both precise and fuzzy methods. Accurate method, first with Baidu search all articles, and then click into the manual record of each reprint of the article provides statistical data (generally available). This method is troublesome, but can be as precise as possible. The fuzzy method is to set up a transparent picture of the 1x1 pixel of Google Analytics (or DoubleClick, etc.) in the soft text, and then pray that the other media will transfer the picture along with the reprint (but the odds are relatively small), Then according to the total number of reproduced in the volume of the number of reprinted to the proportional amplification. Basically, this method can tell you the approximate, but certainly not accurate. See the first few paragraphs of this article for how to add this transparent image.

Product or brand keyword Search volume growth rate, this is my preferred method. Use the method is simple, use Google Insight or Google trends can. Google has told us that Google Insights and Google trends use the same data sources, but Google Insights has more comprehensive and advanced features.

This tool is especially good for emerging goods/brands, such as Intel's Core i5, but it does not apply to those who cannot differentiate between search terms, such as "painless abortion". So, suggest a new digital campaign, worthy of shouting a loud and good note of the new name.

Product or brand keyword search engine results of the growth, directly in the search engine quotes can be added. This growth rate also represents an increase in product/brand awareness.

Social Media Share/forward quantity, this metric is very important for digital campaign of Social Media marketing, but how to get this value is actually difficult. There are APIs open to Twitter or Facebook, so there are a plethora of third-party statistics, and more advanced collaborations-such as the May 2009 Omniture and Facebook-can make data statistics deeper and stronger. But in our country this is a problem, on the one hand, Facebook does not come in, the entry is also "Must Die", Twitter is needless to say, everyone in mourning. On the other hand, the domestic most successful Sina Weibo has not made any good for Data statistics API, Renren has no net. You can count by hand, but I prefer to stay in the future.

Survey questionnaire statistics, this is absolutely omnipotent method, but also the most time-consuming, the lowest degree of freedom method. Not only for awareness, but also for preference, so put it back.

Word-of-mouth monitoring, similar to the survey questionnaire, also put it back.

Monitoring

of

preference

Preference is harder to assess than awareness, and I even think it's the hardest to assess. Because it reflects the inner activity of man. People's inner activities can be translated into behavior, but not backward. For example, I feel sad, I do not necessarily cry, even if I cry because of sadness, but can not say I cry must be because of sadness (perhaps because of excitement), so use "cry" to push "sad" error. Similarly, preference's assessment can be measured indirectly through behavior (such as referrals to friends, collections, etc.), but the error is obvious.

The relatively reliable method of evaluating preference is to ask the audience's inner feelings directly, this is the research. It is also because research can directly ask the audience's inner feelings, so the research methods are also used to study the audience's awareness.

Online research

The basic method of research is to divide the research audience (sample) into two groups, one for control group and the other for exposure group (or test group), and control group is controlled by technical methods and cannot be exposed to advertisements. , and exposure group is definitely exposed to advertising.

For example, Adindex introduces the most commonly used online research methods:

Figure: This is the basic method of online research (copyright belongs to Millward Brown).

The above diagram is the basic method of online research, Group A is determined to be exposed to the originality of the campaign advertisement, while Group B is controlled and certainly does not see the advertisement (what is the control method?). In fact, it is very simple, for example, before the release of campaign to conduct a survey, then the control group will not be able to see the audience ads.

If you want to evaluate the impact of a campaign website on visitor preference, you can also use this method, just to ensure that the control group's audience did not visit the site before the answer.

For the above campaign Web site questionnaire, you can set up your own web site program to achieve a number of visitors to your site before the completion of the questionnaire (control group), while the other part of the site to leave the time to fill out the questionnaire (exposure group). If you don't want to write your own program, there are third-party services in the country that provide similar functionality, such as Millwardbrown, or Admaster. You can also use the Free 4Q questionnaire, which also provides a customized questionnaire (although the number of topics is limited).

There are two places where special attention is needed for the use of questionnaires to investigate preference. First, the questionnaire content design, the second, the number of recovery effective questionnaires and the impact of the statistical significance. I dare not say more, because I am still a complete newbie in these two fields.

Finally, we show you a practical example of a questionnaire survey:

The questionnaire is a survey of online gamers whose statistical conclusions are divided into several major parts, including awareness, motivation (purchase motivation), relevance (relevance), and preference. It can be seen that the exposure group in most aspects of the performance is good, indicating that campaign has a certain role. But it is more regrettable that sample size is slightly smaller and whether it can represent the general audience.

Preference's other monitoring method is to use Word-of-mouth monitoring, this is definitely the forefront of the current field, but also a very difficult to solve the problem, especially for our Chinese.

Word-of-mouth Monitoring

Word-of-mouth monitoring is not as direct as the questionnaire, but Word-of-mouth also reflects the audience (mainly netizens) on the brand awareness and preference feelings. The importance of Word-of-mouth is that if a netizen directly in cyberspace reflects on your brand/commodity concern/love/dislike, then he may not only represent a considerable number of users of the views, but also can rapidly affect other people's views. My own spending is done massively through the web, and negative word-of-mouth is more important to my spending, and I believe most consumers do.

In the article "Sidney's Iwom Monitoring and analysis: Understanding and Practice", I talked about the implementation method of IWOM monitoring, but in fact it is limited to the analysis, so I will prepare another post to talk about my understanding of iwom analysis. Here I would like to mention one of the key metrics in iwom analysis, the sentiment ratio.

The ratio of campaign refers to the proportion of positive word-of-mouth and negative word-of-mouth, as well as the proportion of positive word-of-mouth, neutral word-of-mouth and negative word-of-mouth. For example, there are 10,000 positive word-of-mouth, and 1,000 negative, then sentiment ratio=10:1. This value means the overall preference of your campaign commodity/brand to the general audience.

The picture below is the sentiment ratio of the mainstream laptop brand in the second half of 2008, Neg. Represents the number of negative (negative), Neu. Represents the number of neutral (neutral), Pos. Represents the number of positive (front). At the time, I was amazed that Lenovo had achieved the highest level of adjustment, while HP's performance was rather poor. More than a year later, HP's "Cockroach gate" incident to make it all over the mouth and can not debate, let me feel that the adjustment of the monitoring still very reasonable. :)

The "Preface" is one of the most frequently asked questions on our QQ group, and is in fact one of my clients ' frequently asked questions. This is a very complex issue, not to ask the question itself, but to solve the problem of the feasibility of the practitioners have been plagued. This article will discuss several types of online marketing impact on offline actions and current solutions and remaining problems. This is the subject of the next chapter, see: Online Marketing to the offline action effect monitoring (above).

Body】

On the last article we talked about the online marketing to the offline action effect of the two categories: awareness and preference, this return we continue to the other two categories: Leads and purchase. Monitoring these two analogies makes awareness and preference much more difficult. Here's what I know. But I know limited, perhaps the industry has a better solution, so I hope that friends in the article after the generous enlighten.

Tracking Purchase intention (Leads)

There are generally two ways to track the purchase intention, one is online tracking, that is, let the audience in your online directly submitted to the product consultation or purchase intention, the other is offline tracking, that is, leave the phone number to let the potential customer call you. The previous monitoring is a typical web site analysis, rather than the monitoring of offline action effects, so this article is no longer to repeat. The latter is more important in this article, it is not only because it implements leads tracking in a offline way, but also because it is a dialogue between people (not between people and machines), so it is less likely to cheat (not excluding cheating, of course, everything is possible in this world), is widely used in China.

How to track?

The path that leads to campaign on line is usually as follows:

Potential customer-> See the phone on the AD/website-> call-> product advice, offer purchase intent or contact information

Therefore, if we can track the audience to see the source of the phone number (which is to see which ad or the site of the call), and the resulting effect (product consulting and purchase), it will certainly help us very well to evaluate the value of these ads or campaign sites.

However, offline phone and online advertising/web sites are two different systems, what can be done to make them related?

Do you remember my previous article: direct traffic is really directly flow? How do I identify the source of real traffic? In this article we mention:

use multiple home pages or multiple domain names

In some cases, your site is not to be promoted by the website, but by the offline promotion, or by the Internet spontaneous reference and friends recommended. Now that you are no longer likely to mark every source, how can we discern the spontaneous flow?

A good way is to use multiple home pages or domain names to identify. For example, if I want to do the offline promotion of my website, I will appeal to those friends who see my offline ads through the domain name "www.wac.cn" to visit my website (www.chinawebanalytics.cn) and you will find that these two URLs are different. If I have more than one line to promote the channel, such as television, magazines and building ads, I will give them different domain name: "WWW.WAC.CN/1; WWW.WAC.CN/2; Www.wac.cn/3 "to greet them-so although I cannot subdivide them in the traffic source report in the Web Analytics tool, I can subdivide them through the top landing page report, though not necessarily 100% pure, But it has been able to do a fairly precise subdivision. This approach, not only helps us Open the black box of direct traffic, is also a good way to help us indirectly measure the spread of the line.

The following joke lets us know that there are many good uses in this approach:

A method of tracing information leakage

Just see a way to track information leaks, suitable for wide-spread: registration site I have a habit, the requirements of the photo name, registered Sina I will fill in Sina, registered Yahoo I called Shang, registered Baidu I wrote Yuan Baidu, registered Google I will change the yuan Google, the tax is called Sham collection. Today, I received a call asking: is Miss Jiang Jianxi? I know, CCB sold my personal data ...

Similarly, if we can assign different numbers to different ads and the phones advertised on the website, then we can know which ads the customer is looking for. But the problem is, we might have to set a lot of numbers for a campaign!! Can you do that?

There is a company providing related services, called "off-line Pass", fully follow this idea.

Offline Pass

Off-line pass is a current filling online advertising and offline phone linkage blank products, is a very ideal tracking by the on-line campaign bring sales leads tool.

This tool utilizes the trunk provided by the Telecommunications department, each of which can support 1,000 telephones, so theoretically it can provide thousands of different phone numbers for a customer's campaign activity, each of which can be assigned to every campaign ad or different Web page, So that each ad can correspond to a unique phone number one by one. This makes it possible to monitor how many telephones the different ads bring, and the quality of the calls they bring.

Based on this idea, this product mainly achieved two functions: one is to be able to fully monitor the number of customer calls to enter, and the resulting leads and purchase; second, it is the source of monitoring the telephone, and accordingly assess the efficiency (ROI) and effectiveness of marketing activities (transformation, Conversion). Therefore, it can be positioned as a line of online marketing monitoring tool.

I submitted a trial to get the dashboard of this tool in the following figure:

The interface itself mimics Google's way of showing the trend and average length of phone call data similar to the data that call center can provide.

The call source report is a more meaningful report that shows where the phone comes from. I think that this is the highlight of the tool, it seems more suitable for the offline pass dashboard directly displayed.

Search marketing also seems to support, but I did not try, so all call data are "0". But this function itself is quite meaningful to advertisers who conduct search marketing.

In addition, this tool provides ROI analysis, which is the input of advertising and final output. This tool seems to support the cost of online marketing itself, such as the records of the search marketing CPC (but not knowing exactly what the rationale is, not knowing whether to do a dedicated upload), and by its own setting, can also define the value geometry of each phone, this feature and Google The monetary value that each goal represents is similar in analytics's own setting. While it is not possible to capture directly the value of the purchase that each phone brings, but the custom approach still makes sense, because by estimating the average return on each phone's value, you can approximate the actual revenue they bring, after all, business is your own, and you must be able to understand how much money is behind each phone. The following figure is the ROI data interface.

The following figure this interface is a custom for each phone value.

I think the most important thing about this tool is to solve the problem of call source, which is the core obstacle of online marketing to monitor offline action effect.

Monitoring of other purchase intentions

In addition to telephone inquiries, the purchase intention may also be expressed in the online store for inquiries, or mailing the purchase intention form. These submissions leads the way in the information today, slowly towards the edge, but still not extinct.

There is no way to monitor these buying intentions, the most common method is to provide a network discount code (coupon code), and according to different coupon code to locate the real source of online marketing. However, the information on the online and offline conversion process, there will be blurred and lost. Therefore, this method is available but not accurate.

The way I know about leads monitoring is this, and welcome friends to offer better methods.

Purchase (purchase) tracking

Finally, finally to our marketing activities most want to see the results-buy. Do countless marketing, the final product no one to buy, is definitely a cup. Don't tell me the biggest coffee table is 9.6 million square kilometers, I believe that as long as there is a purchase, your business every day is washing. But how to track purchase is the real source of your online marketing results, not because customers are affected by other factors?

This is the most fundamental problem and the most difficult issue in the whole process. There are 3 available methods.

uses coupon code to use a Post Purchase survey (survey) to use a third party to track

Coupon Code

The role of Coupon code is very similar to the phone tracking method above, you can provide different Coupon code for your different campaign/AD/website, and the customer can get the discount by taking this code to buy offline. This allows you to know which online sources are more capable of promoting sales.

But the problem is that consumers don't always record coupon code (which is, after all, a hassle); another offer may be a costly cost.

Survey

Please the sales terminal to help you to find out what channel visitors are found in your product and final purchase is a common method, but the effect can only be said passable. After all, after people buy goods, there are other things, they generally have no patience, so the best you can still provide a certain discount or discount. Also, the ability to execute sales terminals is uneven and difficult to manage. You need a strong enforcement guarantee. However, this is indeed a commonly used method.

Another kind of survey is the way to send emails or phone calls to customers after buying survey. For commodities, for small items, both cost and customer interruption.

The last enemy to be overcome by two kinds of survey is insufficient sample size. It is difficult to receive tens of thousands of responses, so the design of the questionnaire needs more careful.

Third party tracking

Third-party tracking is my most surprising approach, but it does have this approach, and it is said that there are companies in Beijing to provide services! The idea is that before you do a network marketing (campaign), you can extract probably 50~200 a cooperative audience (partner audience) without telling them any information about your background or product in advance. Third-party companies will then track the consumer behavior of these cooperative audiences over the course of your campaign and subsequent periods of time.

This method is--it's too powerful! What I mean by being strong is that they really need to have good execution, cooperation and record (monitoring), which is really a big project! How far can this approach be? What effect did it achieve? I do not know. If you know this method and the company's friends, please be sure to contact me, my email is sxwuda# #gmail. com (change # #为 @).

Summary of

method

Finally, we summarize the online marketing approach to offline action effects monitoring, as follows:

Well, that's the end of the series. I hope you can talk more about what you see, or just thinking. Because this blog, I have to share, so also got a lot of friends to share and correct me. For example, a friend pointed out that control group is not a It should be translated into a "control group"; and a friend pointed out that I was statistically "messed up," and I totally accepted it, and I was trying to make it up; and a friend told me SAS tools actually provide the model and functionality for Web analytics. I have benefited greatly! I hope everyone can say reserve, thank you!

Last notice: All our QQ groups are full. I will gradually clean up zombie users and inactive users within a week and ask friends to try again later. Thank you!

Source: http://www.chinawebanalytics.cn/online-campaign-influence-offline-part-ii/