4. Tags and Cookies, and how they work
Cookie
• Is a file placed on your computer that helps identify your
computer/browser
1st Party Tags
• A script the site owner (i.e. the site domain) uses, that places a
cookie on your computer and collects data. Some examples are
when sites save your password or login info.
3rd Party Tags
• A script the site owner places on the page but the data is collected by
a 3rd party. Some examples are tags used for web analytics tools
like Webtrends and Google Analytics, and also ad serving tags that
track paid media (banners and search).
5. 3rd Party Tag Data
Impressions
• Tracks if a page or ad banner loaded. Note, an impression doesn’t
always mean the user actually saw the ad or all of the content on the
page. Content or banners that fall below the fold (i.e. you have to
scroll down to see them) typically count as an impression even if the
user doesn’t scroll further down the page.
Clicks
• Tracks a click on a page.
Other Examples
• Video starts and completes
• Page interactions like scrolling, filling out forms
• Survey responses
• Shopping cart and purchase info
6. 3rd Party Ad Serving
Individual publishers
(web sites)
Ad Networks (which sell inventory for groups of sites)
determine their ad inventory and sell that space to
Agencies or Advertisers.
7. 3rd Party Ad Serving
Using Atlas as an example, the process is very similar for DoubleClick (Google) and
other ad serving companies.
Atlas Data is passed to Atlas:
• User Agent String
• Time/Date
• CookieID
• PlacementID
• Adtag
Ad delivered based on placement/ad
assignment and other rules such as:
Atlas • Schedule
Ad Server
(img server,
CDN)
• Frequency Caps
• Weighting
• Sequencing
8. 3rd Party Ad Serving
Data is passed to
Pub/Network: Network Network
• User Agent String 2
• Time/Date
• CookieID Publishers/Networks
• PlacementID often resell their
• Target inventory to other
Networks. When
preparing to deliver an
ad, there can be many
“jumps” from network
Atlas Network
3
to network.
9. Ad Conversions
Atlas can then tie each ad impressions and click back to the action tag, (per cookie).
This is data is then used to optimize the ad campaign.
When a user clicks on an Ad If the site has an action tag on the Atlas
they re-directed through landing page the visit can now be
Atlas to the destination page. directly tied back to the ad.
1x1
Atlas
Atlas
Ad Server
(img server,
Atlas records the click CDN)
and re-directs the user
to the destination page.
10. Action Tags
“Conversions” don’t always have to be landing on a
page. An action tags can also be placed on buy
buttons or pages deeper in the site.
11. Reporting
All Atlas reporting derives from three basic metrics: Impressions, Clicks and Action Requests, as well as
the properties of the placement and action tag (as entered into Atlas).
Atlas can report on the following:
Atlas Data Atlas Reporting
• Impressions & Clicks
Centers Processor Atlas Reports • Conversions
• Cost Accrued
Placement
Properties • Extended Data
• Reach & Frequency
Imps
Placement
• Anything derived from the above
Imps Properties
Clicks
Clicks Actions Actions
We use Atlas reports to:
- Determine effectiveness of
campaigns
- Optimize Placements & Creative
15. Reporting - Conversion
1x1
A conversion can be based on
anything that can be tracked
with an action tag, and is used
to optimize the campaign.
16. Reporting – Extended data tags (Atlas)
An extended data tag acts the
same as an action tag but can
also collect additional data
passed to it from a page
variable.
18. Direct response campaigns
DR campaigns focus on driving users to an “end action”, typically a product purchase trial
sign up or some other action that can be taken on the company’s landing page or web site.
19. Brand and Product campaigns
Brand and product campaigns focus on awareness and reinforcing positive brand and
product perceptions.
20. Remessaging
Remessaging, also sometimes called retargeting, targets cookies that have
taken a previous action. For example visited a site, abandoned a shopping
cart, or didn’t complete a sign up form.
...ever feel like you’re being fallowed all over the
web?
21. Personalization
Remessaging, but with targeted content. Personalization not only retargets
cookies but includes relevant content in the ads.
...ever feel like you’re being stalked on web?
22. “Last Action”
Another example… Serve SKU-level
message based on what
she last looked at
Scenario:
“MDF”
Serve a vendor message
based on investment
(in this case, from Toshiba)
Shopper visits BestBuy.com “Category”
She browses the computer category Serve a category-level
message based on
Views a Dell Inspiron laptop… category she browsed most
23. After I made the last slide…
...the next there it
is!
27. Campaign landing pages
Often times referred to as a CLE (Campaign Landing Environment), is a page or micro site built
specifically for an ad campaign or promotion.
• Look and feel matches the banner, and campaign creative.
• CLE’s can be built outside of the typical company page dev. cycles.
• Agency or 3rd party can build the CLE with fewer technology constraints.
29. Ad serving Data vs. Web Analytics Tools
Ad Serving Web Analytics
Can handle large volumes of traffic Can track multiple metrics with one
tag
Little to no business rules Lots of business rules
Can get cookie level data No access to cookie level data
30. Metrics vs. KPI
Metric – is a quantitative measurement of statistics describing events
on a website.
Examples: Impressions, clicks, CLE page visits
Key Performance Indicator (KPI) – is a metric that helps you
understand how you are doing against your objectives. A KPI should
be relatively high level and tactical.
Examples: Free trial sign-up, online purchase
31. Avinash’s eight critical metrics
• Visits and Visitors
• Time on Page and Time on Site
• Bounce Rates
• Exit Rates
• Conversion Rate
• Engagement
32. Visit metrics
A visit typically means someone came to your site
and spent some time interacting and or browsing the
site.
• Web analytics tools should tie all of the pages into
one visit.
• A visitor can have multiple visits
• Most tools end a visit after 30 min of inactivity
Example: Every day during my lunch break I go to
MSNBC.com and read the news. I typically read a few
articles and watch a video clip or two. Each lunch break
should be counted as a single visit.
33. Visitor metrics
A Unique visitor should count the number of
people who visited your website.
• Cookies for most web analytics tools are unique
to the computer and browser.
• Browsers that block 3rd party cookies will be
seen as multiple visitors. (now 30-40% and
growing)
Example: In my MSNBC.com lunch example I should only
be counted as one visitor. But my wife and both use our
home computer, but we will still only be counted as the
same unique user, this can create some interesting
retargeting issues.
34. Time on Page & Time on Site
Time on the page or site should track how much time someone spent on your
website.
• Note: time on a page or site can mean a lot of things. It could imply people find the
site and the content interesting and are spending a lot of time engaged. But, it
might also indicate people on the site looking for specific information or help and
are getting lost or can’t find what their looking for.
10:00 10:01 10:05 Exit
35. Bounce Rates
10:00 Exit
Bounce Rate – percentage of visits that only had
one page view and left (no clicks).
• This is particularly a useful metric for tracking
how well a CLE is performing.
36. Exit Rates
Exit Rate – percentage of visits that left through a
Exit particular page.
• Metric doesn’t always indicate an issue. For
example visitors who exit via a purchase thank
you page is probably normal, or for example
they found the info they were looking for and
then left.
• Exiting via a product info page might indicate
people are shopping around, but it could also
indicate a pricing issue or poor content.
• This metric can help identify purchase path
issues.
37. Conversion Rates
Conversion Rate – percentage of visits or visitors
that convert.
• Conversions should be tied to a site or business
objective, like a purchase.
• Conversions per unique visitor is typically more
appropriate than per visit.
Example: Unique visitors who purchased a copy of
Windows 7 visited Microsoft’s Windows site 3-4 times
before making a purchase.
38. Engagement
Engagement – Is term used in the industry that can
mean a lot of different things, but typically it is an
attempt to measure how much visitors are interacting
with and consuming content on the site.
Examples:
• Reading content
• Watching videos
• Interacting with calculators, widgets, etc.
Note: hard to know if high engagement rates are
positive or negative, very similar to the issues faced
with the time on site metric.
40. Macro Insights
1. How many Visitors are coming to my site?
2. Where are my visitors coming from?
3. What do I want visitors to do on the site?
4. What are visitors actually doing?