Many search engines with a platform to place search advertisements dabble with this question. But as you can see, this question is very open ended and in order to be able to answer it clearly, it is very important to understand the component questions. Addressing each of them will ensure a better business performance. In an ideal scenario, we would know every customer’s goals completely to such an extent that we can fine-tune each and every advertising campaign. But this Utopia doesn’t exist. Even if we knew an advertiser completely, it is highly impossible to address every advertising campaign on the system. The most analytical way to proceed with this problem is to look at it from a general standpoint and determine general and automated solutions that can in the long-term lead to higher revenues for the platform as well as achieving the goals of the customer. So, the question now becomes – “What are the factors I need to consider before embarking on such an initiative?”
There are 3 aspects to this question
1. Demand – This part is very obvious.Demand in the online paid search industry basically refers to advertisers. The users of the search engine form the supply. There are two dimensions to demand – quantity and quality. Quantity can be influenced positively by promoting the search engine and also by having a good amount and quality of supply.This is a task for the marketing wing of the organization. Any features that the platform offers advertisers will go a long way in influencing the advertisers positively. Influencing quality of demand is a harder task and requires a rigorous analytical set up. Many companies don’t even have an analytical wing and it becomes the responsibility of 3rd Party analytical service providers to show value. It is imperative to understand the various aspects to a PPC(pay per click) model that need to be looked into and optimized to ensure that an advertiser performs to his best capacity. Let us discuss a few ideas in this section –
- Seasonality – There is undoubtedly seasonality in user searches. Searches related to certain industries spike at some times during a financial year. Various reports can be designed to assess the extent of this seasonality. These reports should be able to tell you which segment of keywords show an increase in searches, so that keyword recommendations can be made to advertisers. Business will also do well to understand bidding patterns during these seasonal time periods. An advertiser showing impressions for a keyword may stop showing impressions because of an increase in competition on seasonal terms. The business should be able to identify which advertisers are getting impacted due to competition and make bid suggestions to them for these seasonal keywords. Advertisers should be given the flexibility to set variable bids based on the season
- Ad Quality – Ad Quality is very important in influencing a customer to click on it. Ad quality can be determined on multiple factors such as the text, the ad position and the quality of the landing page URL. It is also important that the ad be directly related to the query that is causing the impression. If not, the system can get flooded by a lot of low quality ads and the credibility of the online advertising platform will be lost. This can do massive damage to the popularity of the platform. Terms like CTR (Click Through Rate = Clicks/Impressions) is a reliable measure of ad-quality.
- Advertiser Intent – Any advertiser wishing to advertise on an online platform will have to create an account and adjust settings to match his intent. Settings could be things like picking keywords to bid on, selecting targeted users, selecting geographical locations to display ads in, etc. Many advertisers are not sophisticated to understand how to make these changes to settings in the platform and hence end up creating campaigns which actually do something very different from the advertiser’s intent. If the business could recognize such advertisers, it could give them a demo of the tool and show them how to use it. This will help in building advertiser relationships as well as improve the platform as a whole. The benefit is mutual. It helps advertisers to achieve their goals more effectively and the search engine becomes more relevant.
To keep it simple, I have mentioned only a few dimensions above. For learning more, please get in touch with me at email@example.com
2. Supply – Supply is mainly driven by the quality and credibility of the search engine. It doesn’t take a marketing guru to understand why Google and Bing have such a large user base. There is more to supply than just having a large user base. It is also important to understand how they behave while using the search engine. In other words, like demand, supply too has 2 aspects – quantity and quality. It really doesn’t bring a lot of monetary benefits if you have a large base of users who do not click on online advertisements. Industry based distribution of search queries could give a fair idea of user behavior. Below are a few ideas to improve performance on a supply-based approach.
- Industry Analysis -This analysis mainly involves tying up the number of searches that happen on search terms belonging to certain industries to the number of advertisements available for these searches. Analysts can check to see how many number of ads were displayed on every search. This will help business recognize weather there are enough advertisements. If you see a lot of search pages with very few ads for a particular industry, you now know which field your customer acquisition efforts need to focus on.
- User Search History – Tracking user searches is highly controversial, but I still feel that this can be done while protecting user privacy. One can look for general patterns and understand the nature of user searches. One good analysis would be to identify search queries that users look for before they get the right result.
- User segmentation – Segmenting users can be very useful in order to drive targeted advertising. User segmentation can be done using various variables like industry of search, frequency of clicking on ads etc.
3. Platform – Demand and Supply are external factors, but platform is a very important internal factor. I think that it is very necessary to have a very good platform and provide advertisers with as many features as possible that gives them more control over their advertisements. It is also important not to include a plethora of features that are not serving a very important purpose as this could result in information overload and leave the advertiser confused. I recommend building a solid dashboard showing the advertisers how they are performing at various levels of granularity. It could also be interesting for the advertisers to understand how their competitors are performing. The dashboard should also be able to tell the advertisers where they could do better and make recommendations on the fly. This kind of a feature will need a great deal of effort to build but I think it provides a very important value-add and all steps must be taken to put it in place if its already not up and running. Improvements to the dashboard can happen over time but just providing this feature tells the advertisers that you do really care about their business. This will help build good relationships with them. Platform also refers to all the complex algorithms which result in ad delivery starting from the time a user types in a search query. This algorithm is a black box to many advertisers and they may have some apprehensions on the fairness. It is highly important that the algorithm be flawless or else there is bound to be multiple escalations in the near future when advertisers begin to understand the failings of the system.
In this article, I have covered all the factors affecting performance very broadly. In case, you want a more granular approach or want to have a problem solved, please feel free to me at firstname.lastname@example.org.
Impression Share is basically defined as the ratio of the number of times that an advertiser actually gets an impression to the number of times he expects an impression.
When can an advertiser expect to get an impression?
Each time there is a search by a user on the keyword that an advertiser is bidding on, the advertiser can expect to get an impression
Why isn’t impression share 100% for always?
There multiple reasons for an advertiser not getting an impression on which he has bid, a few of which are listed below
a. Budget Exhaution
b. High Competition Bids
c. Ad not clearing thresholds for ad quality
At what grain is the impression share computed?
Impression Share can be calculated at any level – customer, account or campaign. It can also be calculated at the AdGroup level but Google Adsense does not go below the campaign level.
Why would I be interested in Impression Share as an advertiser?
Placing a bid on specific keywords shows your awareness of user base. It would be highly important for you to understand why your ads are not showing up whenever a user is typing the keyword. Once you know what is causing less than 100% impression share, you can take necessary actions to improve performance. Microsoft’s AdCenter will in fact filtering out your ads if not many users click on it. Thus, you would be able to understand whether your keyword choice is correct and reallocate budget, if not.
What are the other metrics which are closely related to impression share?
Click Share and Revenue Share provide another perspective to advertisers an can help by looking at the problem from a different angle.
Customer Attrition or Churn is a major cause for concern in the Paid Search Domain. Many a times, customers do not have account managers taking care of them, especially if they are low value customers. For the publisher, it becomes very important to quantify churn in order to understand the incentive involved in appointing an account manager for a set of low value customers. As the customers are of low value, their ad campaign portfolio is not very diverse and single account managers can handle lots of accounts. Is there a good ROI in appointing account managers to such accounts. The answer to this question lies in estimating the potential value that a customer can bring had he not churned out.
Churn refers to the notion of advertisers losing interest in the publisher. Analytics is all about quantifying notions and the notion of churn can also be easily made tangible. As an advertiser loses interest, his performance slowly begins to deteriorate. There are 2 definitions of Churn that I can think of.
1. Impression Based Churn – In this scenario, an advertiser’s budget may run out and he may not care to replenish. He would stop showing any impressions.
2. Click Based Churn – Click based churn is a super set of impression based churn. This is because it is not possible for customers not getting impressions to get any clicks. But, apart from this set of customers, there may also be others who receive impressions but may not attract any clicks. This could happen because these advertisers are bidding on the the wrong kind of keywords. Hence, they show up in contexts, that the user is not interested in.
In some platforms like Microsoft’s AdCenter, if a customer stops receiving clicks, he would also soon stop receiving impressions. The delivery algorithm is designed in such a way that a user gets to see only relevant ads and hence an ad listing not receiving clicks in spite of being served on the publisher, may soon be tagged as irrelevant.
Once we have established the above two definitions, the focus now should shift to how one can estimate Churn. For all my explanations, I would only be refering to impression based churn in order to avoid confusion. All that I explain below can be applied to click based churn also.
The first step in churn estimation is to select a time period of inactivity for the advertiser. Let us start with 1 week. and count an advertiser as churned if any of his listings do not get served for a period of 1 week. We could then take historic data of all advertisers available with us and then count for every month, the number of customers who did not perform (receive impressions) for a period of 1 week ending in that month. We could repeat this exercise for 2 weeks, 3 weeks and so on. It would not be sensible to go beyond a period of 24 weeks or 6 months in the paid search domain.
Let us assume that an Advertiser called XYZ banking served a listing on Google on a particular date in Jan 2010 and he stopped having any impressions till August 2010. This particular advertiser would be counted as churned, if we were considering a 4 week definition of Churn. Though we count him as churned, he has in fact not churned, because he started getting impressions in August 2010. Therefore this customer is not actually churned.
Consider the following graph
The x-axis represents the time period of inactivity and the y-axis represents the percentage of customers who actually churned. For eg. , we see that when we consider 2 weeks as the definition for Churn, only 60% of the customers who we counted as churned have actually attrited. The remaining 40% came back after having no activity for a period of 2 weeks.
There would be a point of saturation where, even if we increase the time period of inactivity considered for defining churn, the number of customers, actually churning out does not increase by much. This time period ,for which if a customer does not receive impressions, can be considered as the cut-off of inactivity for an attriting customer. There is little chance that he may come back after this period of no ad servings. The account managers can then take action on such customers much in advance of the end of this time duration of inactivity to understand the reasons and also make recommendations that are mutually beneficial to both the publisher and the advertiser.
For any further questions, please contact me at email@example.com
In this post, I would be discussing a methodology to calculate a publisher’s latent revenue opportunity in the paid search advertising industry. I would mainly be discussing opportunity coming due to inefficient budget allocation of the advertiser.
Spend and Budget data of advertisers on a time scale
Potential Impact :High Revenue Opportunity
- Insufficient Overall Budget
- Sub optimal allocation of budget among the different campaigns
- Find the time in the month by which the advertiser’s spend exceeded the budget
- Calculate potential monthly opportunity as what the advertiser would have spent on the campaign if we had infinite budget
- This can be done by extrapolating spend based on his budget and amount of time it took for his budget to be exhausted
- Simple Formula that can be used :
What is the Online Advertising Industry?
With the rise of the internet as an interactive networking medium over the last decade, companies have realized the potential that it has to offer in terms of bringing about awareness about their products. The Online Ad industry has undergone a lot of changes and it has become a good income generating medium for many websites. In fact, the bulk of Google’s income comes through online advertising. Listed below are some of the popular forms of Online Advertisements.
1. Display Ads – These are the huge flashy banners with pictures that you usually see online. They come in various shapes and sizes and are built to be visually attractive and catch the user’s attention immediately. You do not find such ads on search websites. The ads can be served by the publisher’s ad server. These include banners, leader boards, skyscrapers, large boxes etc.
2. Pop up Ads – These are the ads that open in a separate browser window whenever you click on a link or visit a website.
3 .Text Ads – Whenever you perform a search on Google or Bing, you will observe that apart from your search results, the search engine displays ads relevant to your search. These are the text ads. Most of the revenues of companies like Google and Microsoft (Bing.com) come from Text Ads. There is a highly complex delivery mechanism which ensures that every user sees the right ads and it is very interesting to study it. A lot of analytical work can be done in the area dealing with online text advertising. These ads can be served by the publishers directly.
4. Flash Ads – These are the animated ads that you generally see and they are a reply to pop-up blockers. They can come in a diverse variety of shapes and sizes and can be well integrated with their respective websites
5. Interstitial Ads – Ads that appear during a transition from one page to another. These user generally has to click continue to skip the ad.
6. Video Ads – With the increase in popularity of watching streamed videos on websites like youtube, these ads have become more popular of late.
7. Email Ads – These ads are distributed by the publishers to all recepients by email. The recipient generally opts-in to receive the ads from the publisher.
Though there are a variety of online advertisements, most of the analytics work of late happens on the display and text ads. There are again many ways by which these ads can be delivered
1. Content based delivery – Content based delivery happens when the ads displayed on a webpage are relevant to the content displayed in the page. There are certain algorithms which assess how relevant an ad is to the content displayed and then display the ads. Companies like Google, Yahoo and Microsoft offer such ads
2. Search based delivery – Depending on the search query entered by the user on a search engine, text ads are displayed on the browser. The advertiser usually bids on keywords and based on the presence of the keyword in the search query entered by a user, the ads may be displayed.
The takeaway from this is that online advertising is still in its very nascent phases. There is still a lot of work required to be done in terms of delivering the relevant ads to the user, so that the user, the advertiser and the publisher benefit. Any analyst whose expertise lies in this domain would be in high demand. This said, there is still a lot to be done in this domain with regard to organizing the rich data that is available and reading patterns from it.
To explain what this term means and why it is important, let me provide you with a simple example.
Imagine two shops, A and B situated next to each other selling mobile phones. There is a high possibility that there might be some intersection in their portfolios which means that there may be some handsets being sold in both the shops. Given their proximity in distance, the price of the handset in shop A might affect volume sales in shop B. This phenomenon is called as sales cannibalization.
The above example is very simplistic. The dynamics of sales cannibalization are complex and there are many kinds of cannibalization phenomena. Let me try to list a few that come to my mind –
1. Cannibalization by a competing store
2. Cannibalization by a competing product which is similar to the product in consideration
3. Cannibalization by other brands
These forces of cannibalization are at play in every kind of industry and it is necessary to factor them into any driver analysis of sales.
Competitor price is generally taken as a variable representing cannibalization. The co-efficient obtained through OLS regression for any cannibalization variable must be positive. It follows from the logic that your sales volume goes up as the competitor increases price of his goods.
This concept is applicable to any industry where there exists substitute products.
Forecasting is an analytical exercise conducted to predict future outlook for any variable of interest. The variable can be sales, units, margins, market share etc. Forecasting is done for a future period based on historical data available with the analyst.
Driver Analysis is mainly done to identify causality. The variables for which a driver analysis is done is similar to the ones which can be forecasted.
Driver Analysis helps us quantify the impact of every independent variable on the dependent variable. It tells how much the dependent variable will increase with a unit increase in the independent variables upto a statistical degree of accuracy. Forecasting on the other hand cannot be used to measure impact of independent variables. The output from a forecasting tool would give us the predicted values of the variable of interest for a future time. Driver analysis can also be used to predict values of the dependent variable by providing it with suitable inputs but the accuracy of prediction for a Driver Analysis may not be as high as that of a Forecasting exercise. In a forecasting analysis, we include all variables into the predictor list as long as they are statistically significant but we do not check for correlation between the various independent variables. The assumption here is that no matter how correlated two or more variables are to each other, there is some additional information in each of them which can increase the accuracy of prediction. We do not include highly correlated variables in a driver analysis because doing so may cause the effect of one variable to be explained by the other variable and hence the statistical estimates that we obtain may not be accurate. The usual practice is to just include that particular dependent variable(out of the list of correlated variables) which makes sense from a business standpoint. This is the trade-off that happens during the variable selection exercise for the respective analyses. To put it in a simple way – A forecasting exercise compromises on causality to achieve higher levels of forecast accuracy whereas a Driver Analysis compromises on accuracy to yield better causality.
There are many tools available that aid us in doing the above mentioned analysis. ARIMA (Auto Regressive Integrated Moving Average) and ARIMAX(Auto Regressive Integrated Moving Average with exogenous variables) are popular tools today, in the hands of analysts doing a forecasting project. Driver Analysis can be done through OLS(Ordinary Least Square) Regression or MLH (Maximum LikeliHood) Regression techniques.
The problem statement for the above analyses can be like something mentioned below –
- The business user wants to forecast revenues for a future period in order to ensure efficient inventory and demand planning – A Forecasting Exercise
- The business user wants to identify drivers of revenues so that he can optimally allocate budget among his various marketing channels – A Driver Analysis
Independent variables considered for the above analyses usually fall under the following categories –
- Product Price
- Cannibalization from Competing Products
- Promotional Activities
- Competitor Promotion
- Product Life Cycle
- Simple Trends
The requirements for the above analysis would be a fair amount of historical data, data preparation and analysis tools such as SQL and software such as SAS with in-built modules for ARIMAX, Regression etc.