Archive for May, 2009
May 09

This great article from Advertising Age describes how Pay-per-click is only one of the metrics advertisers can track. I was suprized to learn that 2/3 of Internet users never click on a sponsored link and those who do tend to be younger and have lower income.

Pay-per-click is a model I like in the sense it gives me invaluable and measurable data on how customers react to a particular ad. It gives me the ability to change the format, editorial and offer and measure results almost instantly. However it does not provide much insight into customers change in attitudes or purchases intention.

Like any metrics, it is useful to have but has to be complemented by other behavioural data. One metric the article refers to is from the car industry. Research shows that the number of test drives booked online can be a good proxy for car sales. This metric can of course not being used on its own but provides a good complement to other Pay-per-click data.

May 09

Here’s the interesting part. As soon as the e-newsletter was sent out I started monitoring response rates 1h, 6h, 12h and 24h after the newsletter has been sent. This was sent to America, Europe and Asia so the 24h time span was necessary to guarantee all members had a chance to receive the newsletter.

The intention was to use the 3 most common email marketing metrics:

- Bounce rates
- Open rates
- Click-through rates

Typically, email-marketing software (such as the G-lock example below) provide these metrics. They offer a high level of granularity (to the extent you can have data for each recipient). Other metrics I look at are: forward rate, unsubscribe rate etc …

I also added 2 behavioral online performance metrics, drawn from traffic to the various blogs and websites we promoted in the newsletter.

I would like to go through each of these 5 metrics and discuss some of the limitations, as I stumbled upon some legitimate roadblocks, which makes success measurement quite complex to perform.

Metrics and limitations

1. My bounce rate was excellent (close to absolute zero). Only one or two emails were returned to sender. This did not come as a surprise since the database I was using was spot-on. The community was launched recently so we would expect all email addresses to be accurate. However one can argue the bounce rate is misleading in one way since it does not take into accounts those emails that might have been deleted silently by anti-spam system. On way around this is to use delivery monitor services such as Delivery monitor. This service allows you to track what happens to your emails to major ISPs.

2. The second metric I looked at was open rates. The open rate is often referred as a percentage of the numbers of emails delivered to recipients. It varies by industry, company, how well has the segmentation was performed (loyal customers are more likely to open all emails they receive form their favorite brands). The headline (in particular if there is a monetary offer) can also boost this rate.

Some of the limitations related to the open rates are:

- First it is a rough proxy for reader’s interest. It does not really measure whether the recipient read the newsletter or recall its content. I would use click-through rates and other measures to determine the ‘engagement’ of community members.

- Second, some email clients have a preview function where part of the email is being displayed (one the side or bottom of the email client). In other words, an high open rate might not take into consideration these recipients who did not actually read the newsletter.

- Third, open rates are calculated differently. The most common approach is to look at open rates as a percentage of email being delivered. Others looked at the same percentage over emails being sent.

3. The third metric I considered was the click-through rate. A CTR is the percentage of recipients who clicked on a link in the newsletter. I like CTR in the sense it provides a better measure of recipient’s engagement.

However, the primary limitation to CTR is it does give an accurate representation of positive shift in recipient’s attitudes and behaviors. In the case of my newsletter, having a number of CTR to let’s say a blog does not indicate whether recipients enjoy the blog content or have a stronger affinity for the community we are building. In the case of a purchase/ re-peat purchase driven newsletter, we will need to link CTR with actual purchase to evaluate the overall performance of the email campaign.

4. The fourth metric I took into account is the number of hits the various blogs and web addresses got in the 24hours after the e-newsletter was sent. Along CTR, this is another way to make a connection between newsletter and traffic generated. In this case, we will need to subtract regular traffic from overall traffic to come up with a good grasp of incremental traffic. CTR is easier to compute.

5. Lastly, one of the call-to-action in the e-newsletter was a phone number community members could call to attend a seminar. This proved to be a reliable success measure (even though it is hard to differentiate the incremental measure form the regular measure since community members could have obtained this phone number by other channels).

an style="font-family: 'Trebuchet MS', sans-serif;">All in all, bounce rate, open rate, click-through-rate and other behavioral measures are great to have. Any email marketing software usually provides all of them. Just watch-out for some of the limitations that might bias results.

For more information on this topics, check-out this amazing resource from Email Marketing Report.

May 09

Recently I sent out the first issue of a monthly e-newsletter, which aims at enhancing member loyalty and relationship for a particular group of interest. The primary goal of the newsletter was to encourage members to attend a particular seminar, sign-up for our blog services and discover a new community web site. Members tend to be digital savvy and opted-in to receive newsletter. I used proprietary email marketing software and sent one HMTL version only to close to one thousand members. Due to time and cost constraint, neither segmentation nor newsletter customization was performed.

The lay-out of the newsletter was distinctive, simple and a continuity of the brands we were promoting. It allows members to gather information quickly. The headlines were designed to capture member’s imaginations and there were several call-to-actions spread-out throughout the newsletter (a combination of links and phone number to call). As I developed the newsletter, I have found the following sources really useful in providing some guidance on content, design and call-to-action.

- Email anatomy > Email experience council
- Email design > Email marketing reports
- Email copy tips > ClickZ Copy tips
- Email copyrighting > Email marketing manual
- Best practices > ClickZ Marketing Excellence Awards
- Email call-to-action > Email marketing report

In addition, you can check out these organizations web sites: EMarketing Association, Direct Marketing Association. I personally also like the CMO Council and Marketing profs which regularly publish white papers on this topic.

May 09

Historically I have relied on Public Relations firms to provide a daily scan of articles on anything related to my company, competitors or industry. I also have set-up Google alerts. Recently I discovered a number of blog monitoring tools that are free and easy to use. I like blogs as a complement to my daily traditional and digital publications scans. Major publications like the Wall Street Journal have a blog that does not always comes-up in Google for instance.

Today, I would like to talk about Technorati.

Technorati claims to track more than 112 million blogs and over 250 million pieces of tagged social media (posts, photos and videos). You can perform keyword searches across blogs and conversations, in more than 20 languages. A service to rank blogs was added recently.

When performing searches, I usually use my company name and its competitors. Sometime I will track down some particular words that are related to my industry. I can modify the date range in order to compare who is talking about what in a given time period.

Unlike its major competitor ( blogpulse), what I like about Technorati is I can also refine my search and select only articles published in the most popular blogs. This is referred as authority. The authority ranking gives me an idea as to how many blogs link to it. The Wall Street Journal has a ‘lot of authority’ for example. From a PR perspective, the authority is probably as important (if not more) than the number of impressions.

Finally Technorati also offers URL and tag searches. I particularly like the URL search function, which lists blogs that link to my company web site (you can also find out what these blogs say).

May 09

I was recently tasked to measure brand awareness for various sub-brands and decide which ones were small enough to be included in our umbrella brand.

Without hard data on hand (and very little time to perform primary or secondary research), I discovered I could use Google trends and adwords to calculate the number of Google searches for one or more keywords.

This will ultimately give me a rough idea of how a brand is searched by customers. One could argue it is a simplistic digital view of measuring to what extent a brand is known by customers. Clearly the example below is for directional use.

Let’s begin with Google trends.

Google trends compared the relative number of searches between various keywords. In the example here, I used two luxury car brands: ‘lexus’ and ‘infiniti’. Google trends ranked the keywords and it clearly show a ‘higher’ brand awareness for lexus.

However Google trends only gives a relative scale ( 2 to 4). To actually get the number of exact searches, I had to use Google adwords. Google adwords gives us the number of searches per month over the last 12 months. This is an average and with a bit of maths and using the relative scale above, I was able to create a graph showing the number of searches for both Infiniti and Lexus over the last five years.

This is a rough estimation but with the data I collected, I was able to have a conversation with our regional brand managers. And decide whether to keep the sub-brands independent. It turned out there is a lot more to brand awareness than the number of online searches. But again, the proxy was a good starting point.