Search Analytics: Internal Search, SEO, PPC

Hi guys,

It has been a while since i wrote about Analytics. I started reviewing “Web Analytics: an hour a day” again for my work over the weekend and i thought it would be a good idea for me to share what i ‘ve learnt.

I am most familiar with Search Analytics and i intend to share what i’ve learnt from Chapter 8: Search Analytics – Internal Search, SEO and PPC.

1. Internal Search:

From my working experience, this is probably the most ignored area when people talk about analytics or analyze website traffic (in SG and SEA context).
We normally analyze all the external traffic sources and completely forget about internal search.

Before we start, let’s make sure we are all on the same page.
When i mention internal search, i mean the search function within your website. This function allows users to search for results within your site when they are there. For example, when you visit IMDB, you use internal search for movie titles; when you visit Amazon, you use internal search for books etc…

Second of all, why is internal search important?

Well, the fact is people are getting more and more lazy, especially when it comes to the internet. (That’s probably why Google is so successful in the first place. :P)
In the mean time corporate websites are getting more and more complicated with thousands of pages. It’s become increasingly important that website has built-in search function to allow potential customers to quickly find what they are looking for.

However, the fact is not many sites in Singapore have internal search function.
Let’s take a look at some of the banks:

  • DBS: DBS is the most popular bank locally both in consumer banking and the corporate side. They don’t have internal search function
  • UOB: No
  • OCBC: No
  • Citibank Singapore: No
  • HSBC: Yes! They have internal Search function. I am not going to touch on the quality of their internal search since it’s beyond the scope of this post.

For B2C sites:

  • Singapore Food Guide (HungryGoWhere): They are supposed to be a so called “Search engine for food”. However, IMHO, the internal search function is not up to my expectation.
  • Flower Delivery ( reasonable internal search function.
  • Creative Singapore ( reasonable internal search function.
  • Singapore Airlines: It has internal search function. However, the quality is not that good. When i type in “Bangkok flight”, instead of showing me the promotions etc…, the top 3 results are for Boarding pass???

Avinash suggested a couple of useful reports/metrics when it comes to internal search analytics:

  • Measure Internal Site Search Usage Metrics
  • Report on the Top Internal Site Search Key Phrases. This can be a very useful report, especially if the information is shared between IT, production, marketing and sales department.
  • Analyze Click Density by Using the Site Overlay Report: This is a bit too much to apply in Singapore or SEA context for most companies. It basically gives you a visual way of looking at how users interact with the search result pages.

We can dive deeper into CTR, exit rate, etc…

  • Measure conversion rate of users who use Internal Search (if you have an e commerce site). This metric is obviously very powerful in justifying the investment for a better internal search function.

2. Search Engine Optimization (SEO):

SEO is basically about making your website more search engine friendly.
SEO have been a crucial part in the overall marketing strategies of many businesses (big or small, internationally or locally) in US, UK, AU or other markets. However, when it comes to SG or SEA, the adoption rate for SEO is still very limited.
Due to the scope of this post, i won’t go in details about common/simple techniques that you can do to optimize your side. But rather, i will go into how to measure SEO efforts.

As rightly mentioned by Avinash, it’s not easy to measure SEO efforts.

While the ultimate goal of implementing SEO is for your website to rank high on Search Engine Result page.
And it’s not just about ranking well. It’s about ranking for the “correct” keywords.
For example, there is no use to rank No 1 on Google in Singapore for the keyword “good restaurant” if you are selling flowers isn’t it?
That’s an over simplified example. However, I hope you get the idea because people’s intention behind each keyword search is different. People from Middle Class, C level are different when it comes to using search engine etc… So you need to be careful when doing keyword analysis.

  • Ranking report: choose 10 to 20 keywords and run a report to see how well you site ranks with Google/Yahoo.

As per usual, in most SEA countries, Google and Yahoo have about 90% of the search market share.
However, one thing to note is that unlike markets like UK, Europe or AU where Google has a lion share of the market, Yahoo is still relatively strong in SEA, especially Singapore. It has about 28% of the search market share <-this is truly remarkable. As the result, when doing SEO, you should optimize for both Google and Yahoo, not just Google alone.

  • Index report:

To me this is secondary compared to ranking report. Knowing how many pages of your site are indexed by Google is important because while it seems that you have hundred of thousands of beautifully designed web pages, if Google can’t index your site, you will never be found on Search Engine Result Pages (SERPs).

  • Pages with External Links

“See which pages on your site have links pointing to them from other sites.” – Google Webmaster tool.
This is a very important report. However, as i mentioned, going deeper into SEO is beyond the scope of this post.

  • Segmentation:

Segmentation is the key in web analytics. In this context, it means separating organic traffic from Paid Search traffic and other traffic sources.
Define clear business goals (conversions for example) then compare organic (natural) traffic performance with other sources.
While doing so, please remember that organic traffic is Not Free.
If you engage an agency to do SEO, then the cost would be agency cost.
At the same time, don’t forget cost like “IT” cost; tons of money spent on branding offline/online, offline marketing activities for specific products (by the way after people see an ad, they will go online and search).
It’s hard to draw the line and say how much money spent on branding and other activities actually contributes to Organic/Natural traffic. Yet, it’s clear the organic traffic is Not free.
It may be cheaper to present your products/company info to a customer online vs offline. However, organic traffic may be cheaper than other type of traffic (both offline/online). Yet it’s Not Free.

3. Pay Per Click

As rightly mentioned by Avinash, Pay Per Click is hot. It’s hot simply because it’s effective and so called “easy” to get started.

A quick note: we must separate Search campaigns and Contextual campaigns.
The two types of campaigns are so different in nature that any kind of metrics have different level of relevancy or interpretation for each of them. So we can’t lump Search and Content traffic into one and use common metrics for both.

A few fundamental metrics of a PPC campaign include:

  • Number of Impressions = How many times your ad appears.

Number of Impressions is more relevant with Campaigns running on or Search Network compared to Campaigns running on Content Network.

  • Clicks: It simply means number of clicks from Paid Search to your site.

Please note that it does not equal unique visitors or total visits to your site as the result of the Paid Search campaign

  • Click through rate (CTR): the ratio of Clicks/Impressions

CTR is an important metric for both Search Campaign and Contextual Campaign. However, please note that for Search, the CTR is always much higher compared to Contextual and there is nothing excited/fancy about that.

Why CTR is important?
It’s because with high CTR, you have more clicks coming in from your Paid Search campaigns.
Not only that, normally with high CTR, you tend to pay less per click due to high quality score (Google term) / high quality index (Yahoo term) etc…

  • Cost per Click: CPC

Yup, those are the fundamental metric for a PPC campaign. However, as you can see, there are many loopholes working with these metrics in silo:

  • The first question you may ask: what these metrics mean to your business? to the bottom line?
  • How can you integrate these metrics/numbers into the existing analytics tool so that we can do more fancy analysis?
  • etc…

Despite the fact that there are many unknowns, unanswered questions, unclear relationship by just looking at these numbers in silo, there are still quite a number of agencies in Singapore/China/Hong Kong etc… that only provide clients with these data.
By this, i am not including those agencies that make an effort to explain to the client the full picture. Yet, due to the nature of the relationship, the constraints etc… that they have to work with these data alone.

Bottom line impacting metrics:

Avinash uses the term “Outcome”, which can be “orders, leads, page views, or anything defined as a goal for the campaign”
Then we have the following:

  • Onsite Conversion (OSC) Rate: Outcomes/Clicks. This should be a very common metric. It simply indicates how quality your traffic source is with respect to the bottom line.

However, please don’t read too much into this metric either. The reason is there are a lot of things that can have an impact on the Onsite Conversion Rate (OSC).

First and foremost, you need to clearly test to see if the way you measure Onsite conversion rate is correct or not!
It may sound simple but it’s actually not.
For the number of clicks, agency can get from Search engine so it’s the easy part. Yet, the client may use 3rd party tracking and if the agency don’t set up/test the tracking correctly then it can be a big problem. Even with proper tracking, the numbers from Search Engines and 3rd party tracking can be 10-15% difference especially if you use Search Network/Content Network.

As for the number of conversions, it can be even more complicated. The easiest and current ways of counting conversions use:

  • Conversion Rule: Last Click win. It means that a conversion is only counted towards a particular source if it is the last traffic source that visitor uses to visit the site and convert.
  • Conversion is counted when visitor lands on the confirmation page. And here lies a possible problem.

Often users may refresh the page or leave before the confirmation page is completely loaded. Either way it will affect the number of conversions.

Another common problem is one user may convert multiple times. It can be for multiple products at the same visit, different visits stretched across a couple of days or even weeks.
Besides, users journey is increasingly complicated. An user may interact with a banner (and so a cookie from the banner campaign is dropped on his computer), an eNewsletter (again another cookie), natural search and PPC. So if you don’t set the conversion rule carefully, different agencies will try to claim for the same conversions.
For example: agency in charge of media planning will count a conversion as long as someone clicks on the banner and later on buy something/convert. SEM agency will also count the same user because as far they are concerned, user visits the site through sponsored links.

In short, I hope you get the ideas now. We shouldn’t read too much into Onsite Conversion. It’s critical to understand the number. However, there are so many other things that we need to take into consideration.

Next are other metrics related to conversions are:

  • Revenue
  • Cost per Lead (CPL)
  • Cost per Order (CPO)
  • Return on Ad Spend (ROAS)
  • etc…

4. Advanced Analytics Concepts:

This is chapter 14 materials but i think it would be beneficial to discuss some of the points here.

  • Statistical Significance: Please refer to Wikipedia for the definition
  • Segmentation:

It is impossible to segment any metric without putting in the effort to understand
what you are reporting and the business value that the metric represents.” Avinash wrote.
Personally i truly believe that Segmentation is Powerful and hard work. It’s also the sign of a good analyst.

In his book, Avinash presented many simple yet powerful examples to show us how segmentation should be done. I only discuss those that are related to Search here.

First of all, every time you try to report a figure, you need to take into consideration the context of that number, then numbers from Search Engines as a whole, then numbers for Google Search, Google Content and Yahoo Search.
(Yahoo Content is still very limited in SG and SEA so it doesn’t have many practical values to use that).

Don’t forget to do analysis for each country alone rather than grouping them together. It might be convenient now to group things together. However, please make sure that you have the raw data when you need it. This means no PPC campaign targeting two different territories please!

  • Trending Segmented Data:

It’s best to show these trends with bar graph or just lines. Further more, it would be very “actionable” if you show trends from different traffic sources together or mix it with competitor analysis/industry benchmarks.

Last but not least, below are the best practices according to Avinash:

  • Forget about Overall Site conversion rate: Totally agree

Many clients asked me this question before “What ‘s the benchmark for conversion rate?”
I always take my time to explain to them that every industry is different, every market is different, whether the client is using web analytics etc…
I haven’t tried something like “If i told you the number is 20%, are you going to scream because your number is 2%” yet, may be next time 🙂
“There is no other metric that will tell you less about your website than overall conversion” – Thanks Avinash!

  • Trend over time and Don’t forget Seasonality: Totally Agree

Forget about doing Paid Search for 1-3 months and start shouting loudly why it does not work.
Obviously there are people/agencies out there that are particularly bad…But for true professionals we know that seasonality matters as well as other factors.
Be patient because sometimes a SEM campaign can take as long as 6-12 months to be mature.

  • Understand your website/company/client’s company acquisition strategy

For example, it’s hard to start doing any kind of Direct Response Online campaign if a company has always used offline as the main acquisition channel and people in that market are so used to it.
(It doesn’t mean it can’t be done)

  • Measure conversion Rate by Top 5 referring URLs

One thing i find is that for big MNCs, marketing people often don’t have access to Web Analytics or no one shows them how to use it. Also, agencies are often lazy and don’t link up with Analytics team from the client side.

Whenever they (the marketing department/agencies) do media planning and buying for example, rarely that they think of finding out the top 5 referring URLs that bring the most number of conversions currently or historically!

  • Dont measure conversion rate by page or link

I don’t see this problem often in SG or SEA.

  • Segment like Crazy

We’ve talked about this over and over again.

  • Always show revenue next to conversion

This is only applicable to the advertisers/clients’ internal team i supposed. Revenue or profit margin per conversion is strictly confidential and clients normally don’t share this information with their agencies.
What agencies know (if they are not too lazy) is benchmark or industry average values only.

  • Measure conversion rate with a Goal in mind

There a quite a number of other best practices mentioned in the book as well. My recommendation is to get yourself a copy and spend time understand those practices in their well written context!

That should be all from me.

See you next time!


  1. Interesting post! I have been watching Pay per view, CPA, CPI… and this post gave a lot of details. Keep it up.
    My 2 cents: split the post into smaller parts. According to studies, people tend to stop reading after scrolling down for 2 pages, unless the article is _very_ interesting.Rgds

  2. Thanks for the comments Jacky.
    My only feedback is whether those people, who according to “studies” tend to stop reading after scrolling down for 2 pages, are part of my target audience…

  3. ———Singapore Food Guide (HungryGoWhere): They are supposed to be a so called “Search engine for food”. However, IMHO, the internal search function is not up to my expectation.———-

    Why the hungrygowhere search engine is not meet your expectation, any reason ? which keyword did you try to search ?

    IMHO, it’s my favourite site. The site search is nice, it’s correct some typos spelling.

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.