[0044] An illustrative implementation of an "insert before aid" rule will be
described with reference to FIG. 4. Turning now to FIG. 4, an initial code portion 410 is subject to a rule 412 tiiat has the illustrative format wn to produce a result 414. Illustratively, result 414 indicates that tag 416 appears before the aid of the body portion of the HTML source code, as delineated by the relevant HTML and tags, that latter of which is represented by numeral 418.

[0045] An illustrative implementation of an "insert after an rule will be described
with reference to FIG. S, wherein an initial code portion 510 is subject to a rule 512, whidi includes the illustrative format shown, to ptodacc a result 514. As can be seen in result 514, tag 516 is inserted after an aiding body tag 518.

[0046] Goierally, in the case of tags insertion, an insertion point will be defined by
the name of a tag, and optionally, die value of the "id" attribute. Thus, in some embodiments, all of ihe other non-"id" attributes of each tag can be scarded as well as the literal content inside HTML tags and the tags.

[0047] A lexical analyzer can be utilized to produce a sequence of tokens. The
lexical analyzer receives HTML text as iiqnit and will provide a sequence of tokens that correqwnd to the detection of an opoiing tag, a closing ta& an attribute name, an attribute valu a comment, and so on. The lexical analyzer can form a part of client application 130 or sava--side application 136. With reference to FIG. 6, the firuits of q>plying an initial code portion 610 to a lexical analyzer 612 is shown in box 614. It is not necessary to verify that the page is syntactically correct, that all the open tags are closed, etc. Rather, an embodiment of the present invoition will wait for a tag with a given name. When it is found, it is inserted into a queue along with information about the position of its opoiing and closing tag and the "id" attribute if present. At the end of parsing, a detomination can be made if there is an insertion point that satisfies a spedfied rule. If a tag is not found, the insertion fails in one onbodimoit Otherwise, the exact position in the docummt is known as to where the tag will be inserted. This exact locati(m can be passed as one of the parameters to an "insert" function.

[0048] Another parameter to an "insert" function is that of the "insertion rule id"
that can be used to write a maiicar for the inserted code 120. This functionality may be
useful to automatically iqxlate an oldo* version of a custom scr to allow usos to choose
one instrumentation provido* in place of aiuither, or to fiunlitate A/B testing, whidi is a torn
that we use to describe one aspect of facilitating experimoitation and variation of wsite
topology and interaction; for example, exposing a first set of options ("A") to some
customers but a second set of options ("B") to others, and u«ng the instrumoitation to
evaluate a comparative efficacy of both options. An illustrative format for an opening
marko* is as follows:

[0049J "."

[0050] An illustrative format for a closing marker follows:

[0051] "."

[0052] These are mere samples. Other items could be included, such as a time
stamp for an identifier, and some items rranoved, sudi as in table below:
Opening maAer.
Closing marker TABLE 1

[0053] We show a more goieric form of markers in FIG. 7B. But in the exanq)les
above, "product" is a fixed portion that identifies a fwoduct that is the subject of instrumenting (such as Google Analytics). "Rulejd" idoitifies a rule identifia, sudi as an integer that is provided by a configuration database. This can be retrieved from tag repository 140 in one onbodiment "Version" identifies a veracMa in tag repository 140 of a rule 152 applied for instrunMntation. An illustrative exanq is provided in FIG. 7A. And FIG. 7B illustrates a hiior-level example, whernn a bjjmung markor 710 marks the beginning of embedded code 712 (or 120), and ending maika714 marks the end. Detection and deletion of inserted tags 120 can be implemented using an HTML parser aiMl looking for one or more comment tags that contain markers (sudi as 710 or 714) with the sped&ed product and version. As metoned, this provides an easy way to detect currently embedded tags.

[0054] Of course the above is moely one example of how to denote ented tags.
There are a variety of difforait ways to denote tags, but what is CQnq)arativdy m(»e rdevant is that in some onbodimoits tags are somdiow daaced so timt those notations can be seardied for so that embedded tags can be easily fouixi.
[0055] Regarding tag detection, we note that web-analytics instrumentation soipts
can co-exist with otho- analytics products. In some embodiments, &ese cannot be inserted twice in the same page. In such embodiments, before inserting new tags in a page, it is fruitful to verify that sudi tags are not already presmt in a page. Pre-existing instrumentation may have been inserted manually, by the owner of the website content, or automatically by another tool, or even by clioit amplication 130 itself. Tags that are inserted as previously described are relatively easy to identify because they are surrounded by markers as previously moitioned. These markers define the boundaries of sudi tags, (e.g, 710 and 714). Thus, these type of tags can be found using a method that seardies for &e markers.

[0056] But in cases where markers or equivalents are not provided. Hie source code
of a web page can still be inspected to see whether a "code dgnatiue" qiecific to a code
portion to enter can be seardied for. For ocample, cotain wd}-analytics products bear a
code signature. In fbt context of detection, it is not particularly impottaat to perfectly
define the exact boundaries of a code snippet An illustrative code snippet follows:

[0057] "Script language - JavaScript" SRC == http://x.maLcom/saipts/acasanpt.js.

[0058] A goal is to correctly identify the rig code aiippd. To fiidlitate a desirable
levd of flexibility, a code signature can be spedfied through a regular expression. In such a
scoiario, tiie detection function will be implonoited by {flying a regular expression to ihc
content of a page. The fimction will return a Boolean result *% tagged or not" togedm* with
information about the posititm of the tag code in &e page if needed.
[0059] Regular ex|ffessions do not dways woric wdl in the context of wd> pages.
The searched patton may appesat in many diffoent places in a page. For example, the text or might appear in a string contained in a script tag before &e actud body tag. This is shown in the table below:

[0061] To help avoid errors, some prqnxKessing stqis can be carried out Before
looking for a regular-exfHession matdi, tbe page can be prqmioessed to remove all the parts that can cause spurious matdies. From an inlemention point of view, a similar preprocessmg sdieme with different parameters can be used to implement any tagging operation, including detection, replaconent, insotion, etc. Given a page to be tagged, the preprocessor will return a strii)ed vosion of the page that only contains the minimum s of tags and attributes that are required to >ply the specific tging operation. This stripped version will only be used internally in (me onbodiment to jnedsdy dete the presence of a tag or to identify the correct insotion point and avoid ptsible fidse matdies. The actual operations of insertion, deletion, rqplffiiiement will be done on the cniginal page.

[0062] In the case of detection, to find pre-existing instrumentation code, the pre-
processing step will discard all of the conunents as well as all literal contoit inside HTML tags in one embodiment To make this process easier, pi caiwnicalization can be employed. It is easier to caiK>nicalize the string we matdi against so that the rilar expression stays simpler. White spaces such as spaces, tabs, and new lines can be transformed into a single space during the prqnocessing. Thus, lines that have various spaces in betweoi expressions can be simplified so that tbey are separated only by a single space.
[0063] Regarding tag replacement or deletion, an HTML parser can be employed as
previously mentioned to look for comments whose contoit matdi the text of a msAsac in cases where a marker was provided. When a matdi is found, all the maiiaq>s that are included between two markor comments (such as 710,714) can be ronoved fivm tiie page. The replacemoit of marked tags can be implemented by ddeting all existing marked tags finm the web page and then re-inserting, possibly in dififaatf positions or with difiGaent parameters, new tags.
[0064] For unmarked tags, an alternative process can be employed. Gamrally, the
tagging of a single product (like Google Analytics, fw exanle) generates a number of changes in different parts of a wd) page. These changes would need to be detected, and a set of reilar expressions would be defined to specify tl» poile code snippets to look fiw. Thus, in one embodiment, a list of popular analytics programs can be inspected for code that is either unique to that providor's implonentaticm or is found in other jnovider's implonentations and serve as a )od indicator that a {dirase or torn indicates the presence of a tag 120.
[0065] For example, consider the illustrative tag 162 of FIG. IC. It can be seoi that
the iirase "URCHIN.JS" is pcesoit. This phrase, alcmg witii as many others as is desired to be searched on, can serve as the basis of a quay to seardi for in a set of wd> pages. Because it is relatively difScult to actually remove all of the code associated miSx the snippet such as that of 162, such code is neutralized in one enibodiment. For example, all of die calls to functions of known products can be commented out. Because there is a potential for essentially breaking a web page, one emboment of the present invention {Kovides the ability for a user to test his page hefote puUisliing it to ensure that it has, in fiict, not been broken. Accotfy, in one embodiiiMnt, all changes that are made to specific web pages are tradced. These can then be cancelled aiKl the original page can be restored in the event of an error.

[0066] Similarly, an HTML-verification process is provided in one embodiment
Such a process can verify the validity of an HTML aiment before and aftor a tagging
opoon. This can be UMdfcnr testing purposes as wdl as to verify tiiattiietaggii has not
brokoi a page, at least fix»n the point of view of HTML syntax. This functionality can be
implemented using &e aforementioned HTML prepfocessoiJparser.
[0067] To explain a variation of prqm>cessing stq[» to carry out, tiie i»resent
invention can employ page or document encoding. Information of this is provided at
mia-osoft.com/globaldev/getwr/steps/wrg_codage.mspx, whidh is incorporated hoein by
reference to explain encoding, and is also provided in an IDS filed with this docummt In
such an embodimoit, a codepage can be detected. This can be accomplied by detecting
Unicode BOMs (byte order mffi:ks) at the beginning of the docwnoit (see, e.g.,
unicode.org/unicode/faq/utf_bomJitml, whidi is incorpored by reference heron to explain
detecting BOMs), looking for the ptesace of in the page, aiul/or
even asking the rit codq)age to the user. After detecting, the page can be transformed
into a normalized format such as Unicode UTF16. An HTML parso runs oa the
normalized version to hdp &cilitate the inserttcm/deletion/modification of tags. If desired,
a resulting (UTF16) documoit can be converted back to the (mnal oo<>age.
[0068] Turning now to FIGs. 8A-8M, greater detail of user interfile 132 is
provided. In one embodiment, these are the screens fbai would be jnresented on climt
device 126 by way of browser 128. If climt {plication 130 were a local iq)plication, then
similar screois would be provided but not necessarily presented by way of a network
browser 128, although they could be evm if the tUcatimi woe a local, stand-alone
qyplication. Screoi 8A depicts an illurtrative welcome aanm that usor 126 woidd see in
preparation to instrunMnting a desired wcbate, whidi for {Niiposes of ttiis exanIe will be
wd)site 116, wifli pages 118 being a set of pages that ettfaer could or will be instrumented.
As can be seen by item 810, this process allows for automatic instrumentation.
[0069] In one embodiment, another soeoi, as shown in FIG. 8B, can be inresented to
allow a user to select m option to automidcally insert traddng scripts into potmtially all web pages 118. As is fAxnm in FIG. 8B by reforeoce numoral 812, automatic tagging will insert tracking soipts in each wd) page if desired. Turning now to FIG. 8C, soreen 814 inx)vid«i a vtset interfoce to receive locaticm infinmaticm iiuA enables identificaticm of wd) pages 118. It is moy illustratively iwn that m FTP paradgm can be used, vtiurein a server and security credoitials are provided. But it may also be the case that a firfder 816 can be browsed to by button 818 to receive location informaficm that oiables web {Mes 118

to be identified to be instrumented. Such a folder does not need to physically reside in the same piece of hardware as fbat of client application 130 OT l»owser 128. Ratho-, it can be readiable by way of a networic drive.
[0070] FIG. 8D indicates that the invention is ready to start insoting tracking scripts
into each page of website 116. A setup button 820 is provided to allow greater customization of instrumentation sudi as those previously described and mote. In FIO. 8E, it can be seen that the presmt invention is iq>proximately 33% complete, having processed three files as indicated by reference numeral 822.
[0071] FIG. 8F depicts a screenshot blowing that all nine files 824 were processed
and tagged successfiilly onbedded, whidi is reflected by refermce numeral 826. As shown by comment 828, ti-addng scripts 120 have now been iniMrted such that wsite 116 is ready to gather visitor or user information.
[0072] An illustrative trackdng-soipt-installation repent can be seen in FIG. 8G,
which is referenced by numeral 830. This rqwrt shows eadi page as well as an indication of whetho' a script was insoted and optionally by what way, such as "banc instrumentation." FIG. 8H dqncts a soeen 832 that illurates an ability of the present invention to allow custom setiq> aiul tracking scripts to be inserted. Here, a usor is allowed to select certain folders and files to have custom content embedded in. The illustrative custom content shown is analytics-tracking scripts.
[0073] FIG. 81 indicates that the files in the indiotfed dnectories are searched for.
As shown by referraice numoal 834,4,215 files are shown to have been analyzed.
[0074] Turning now to FIG. 8J, a screen 840 of the user inter&ce illustirates }K>W
options such as an ability to refivsh a set of traddng soipts 842 is available to a user. Also shown are various features such as downloads tracking, page raamin ai events tracking 844 and tiiey are available by an embodiment of the present invention. A user can click one link 846 to refiresh his entire wd)site with tracking scripts. Following sudi a link would allow all of the scripts that have been embedded in all of tiie pages to be refi«shed according to, periu a new set of rules defined in rules database 152.
[0075] Aspects related to downloading files can also be tracked. Turning to FIG.
8K, a screen 848 shows that files with certain extensions or file types can be selected to be tracked as downloads on wsite 116. Moreover, a usor is free to add additional files having customized extensicms by following link 850. hi this way, an onbodimait of tiie present invention allows for downloads to be tracked.

[0076] Another illustrative feature of the present invoition is shown in FIG. 8L.
H&K, >er page alias" selecticm can be made. This allows files with potentially long file
names to be roiamed with an alias or nickname that is ema to work with. Illustrative page
names that are shown include "index," and 'y page," shown by numoals 854 ai 856,
respectively. These pages can be roiamed as shown in FIG. 8M. Note in FIG. 8M that iUm
858 is now associated with an alias 'purdiase" which was the desired name dhange, as
shown in FIG. 8L by numeral 858A. Similarly, item 860 in FIG. 8M is renamed
"checkout," as can be seen witii referoice to item 860A in FIG. 8L.
[0077] To recapitulate embodiments of the present invoition jnovide an extensible
set of embedded services that can be included. The invention fecilitates modifying a set of web pages so that contoit ads can be displayed in the hiest numbo- of locations. Also, in the gadget/widget maiket, some make a business model out of embeddable content sudi as web 2.0 items. An onbodiment of the invoition oiables sudi itons to be easily and automatically embedded. In a community-type contact, community contmt genoates more visitors if it is distributed in third-party web content (e.g., Flidcr images, Soiq>box videos, LiveEartfi maps and the like).
[0078] Analysis of HTML content is also fnovided to determine what embedded
services can be included: &nbodiments of the invention mable for each target ccmtent an
analysis of &e compatilnlity with the requested embedded service.
[0079] Tag cust(nniz8tion is also me possible. Eh t 120 may contain
parameters that can be customized. Embodiments of tfie present invention expose a uso* interface that helps a contoit publisher 144 or 126 to adi tl» embedded services to their needs.
[0080] Tags 120 insertion in the right location in HTML file is also i«ovided.
The invention can detect the right locati(m(s) in the tmget ooiltent 118 to insert the tags 120.
Alternatively, for specific type of contents 148 and/w 142, an embodiment of invention
can expose a way for user 126 to visually select the locaticm. For example, in onbodiments
whoe advertising-content-rdated code 148 is to be onbedded, an onbodimoit of the
presoit invention can receive vsat input that either confirms or dictates an insertion point
for the code 148/120. Because fliis type of insertion will usually change the layout of &e
page, an embodiment includes die option of presmting a preview of what the modified pa
will look like. The user can review the layout of a page after tiie ad is inserted.
[0081] User-configuration information can be stored for each embedded service and
wd) destination or otho* location information. This makes iqxlating easier. Illustrative

configuration infonnation stored, peiliaps in tag rqwsitory 140, includes selected onbedded
services, user parametos, target content, page-spedfic infonnation and like.
[(N)82] Additional features can also be provided. lUustrative other featiores indwte
subscription to proposed embedded sovices (utilizing a c(»nmon user identifier acavMS various services such as onail, analytics and the like); syndmmization (verifying that the current tagging implonented in the wd} contoit matches the tagging timt die user required in the customization); refiresh (automatically iq)dating tiie web content with the configuration required by the user); and ronoval (removing some or all the "onbedded tags" fiom the website). These features are illustrative in nature, and do not compose an exhaustive list, as is the case with other lists contained herein.
[0083] As previously mentioned, onbodiments of the i»eseQt invention can take (m
a variety of implementations, niustrative implemmtatioos include taking &e form of an
online tool, a w content building tool plug-in, a wd>-sarver enhancemoit, a content-
editing platform extensicm, and/or a software developer kit (SDK).
[0084] Turning now to FIG. 9, an illustrative {nocess accordii to an embodimmt
of the present invention is provided. As a prqnx)cessing or ptecxasan fumtion, location
information of a wd>site to be outfitted wi& instrumoitaticm aod/or onbedded-content tags
is received. The URL of a wd)site, a folder location, IP address etc., may already be known
by the system, or it it can receive tbs data fi:om a user. The source code of a first web page
is analyzed at a stq) 912. Analyzing the source code of a wd> page may actually include
various substeps including analyzing a copy of the web page, or a stripped vosion of the
web page as previously mentioned. At a stq> 914, a detomination of an impropriate tag to
embed is automatically made. In some embodimmts, an fprqniate tag is a definilt tag to
be entered. In ofhor embodiments, user-im>vided information is considered to determine an
fq>propriate tag to embed, hi still other embodiments, an advotising-ocmteat code portion is
determined at this stage. At a step 916, an appsopnate onbedding locatioo is detennined.
This process has beoi previously desaibed above. With a tag determined ai insertion
point known, the ti or tags are automatically embedded into the web page at a step 918.
This process can ccmtinue for as many pages as are to be instrummted. Thus, at a step 920
a determination is made as to whether tha« are any additimud pages to instrummt. If so,
processing revots to stq> 912. If not, the proce» may end in one embodinMot
[0085] At a step 922, a verification st can occur that allows far the verification
that the embedded contoit did not cause an undesired altmiticni of the wd> page(s). If the content that was added was an analytics soipt, than the qipeaiance of the web page suld

not have changed. A comparison of the way that the web page ypeared before the insotion
and after the insotion can be made. This can occur in a variety of ways. In OOB
onbodiment, size or other dimoisions of the wd> pe can be bld, flieQ an assumirtion can be made
that an undesirable change has occurred. If advertising contoit was aed, thai, provided
the size of &e advertisemoit was known, thai anotho: sizing comparison can be made. For
example, if an add was 60 pixels hig then the height of the web page can be diecked to
see if it increased by 60 pixels, hi still another embodiment, a pnsview of a chan can be
presented to a user to recdve ah indication of whetho* there is any noticeable change.
Exposing that option to a user is valiiable. Functional checks can also be made. Adding
analytics code or advotising content should not alter the fimction of a web page, at least as
observed from a viewer's 124 perspective. Things sudi as links and media renderings can
be automatically tested, as well as the rendering of the mtire page.
[0086] We also previously explained aspects of verification earlio- (see the section
desaibing document encoding). In smne embodiments, Hat invration does not exiidtly
dieck to determine if the inserted code 120 breaks aesthetics cur existing functionality per se.

In these embodiments, a chedc is porfonned on the DOM to analyzse if the existing DOM
remains the same after an operaticm of adding and then removing the soipt. Recall diat a
Document Object Model (DOM) pix>vides a structural rqnesentation of a document TIM
diecks poformed ensure that what is already in the page he&xe the insertion is not broken
as a result of the insertion. For exanqde: if a cUck on a button in die page changed the
source of an image on the same page, Hxai automatically adding traddng script 120 will ix>t break that existing fimctionaUty. Also adding traddng script 120 does not diange layout
of the page. Genorally, dianges do not occur to Has rendering, interaction, and
data/flication functionality of page. But adding an advertiseownt script will change
the layout of the page. Even so, an embodunoit of flie invention ensures that Hie
functionalities existing on the page pricsr to insertion are not Inidcen.

[0087] Changes may be made to tiie instrumeatalimi fimcticmality. "Reiklaingr
refers to the visual layout and appcmaux of the page. "Intenu:tion** refers to the bdiavicHr coded into the page for things like JavaScrifrt mouseover events, form controls, etc. (opticms are legion). "Data/q)pUcati(Hi'' teSas to die actual vmA. HM page does, e.g. process an ecommoxe transaction, rdrieve a user's account balance frnn a database and show it to them, etc.

[0088] One way of testing for errors or developing an exclusion rule includes
inserting tags into a set of pages (maybe thousands or millions), and thm removing them. A byte by byte and DOM comparison can be done on the page. If one of the comparison &ils to return equal results, the tagging is considered as fiiiled, at least to that page. The fiiiled tagging page can be saved and its content analyzed to idoitify what oror in the page broke the tagging. If a woric around is know, the a tagging fix can be created. If there is no woric around the we add the page ettor as one of the exclusion rules.

[0089] An alternative way is to tag on the fly random pages that are browsed. All
such pages are tagged. Later, they are untagged. A comparisons is made and action take as explained above.

[0090] Turning now to FIG. 10, another method for practicing an onbodiment of the
present invention is provided and referenced genonlly by the numeral 1000. hi this embodiment, at a step 1010, a uso* inter&ce is presented to a user to receive descriptive information that describes a desired website that is to be instrumented via tracking scripts. In one embodiment, sudi a user inter&ce may take fb.c form of (me or more screms of FIOs. 8A-8M. At a step 1012, tracking scripts are automatically installed into a set of wd> pages of the website. As previously mentioned, locations can be determined and scripte can be selected automatically, without user intervention, aiul embedded. At a st 1014, a user interface is presoited to receive events to be tracked. Tlxse stq)s do iK>t need to occur in any certain order. In some embodiments, it is hdpfiil to have scripts embedded in w pages before different paramers associated wi& those scripts are received. In other embodiments, these steps can be consolidated or evoi reversed in order such that a user into&ce is presented to a usor to receive indications of events to be tracked even hote the traddng scripts are installed. And afto: those iiHlications of events to be tracked are received, then the tracking scripts are installed. Examples of a user intar&ce to receive indications of events and the like to be tracked can be found in soeens 8H, 8J, 8K, 8L, and 8M.

[0091] At a step 1016, the tracking scripts tiiat woe entered can be autcmiatically
updated based upon evoit-tracking indications. Thus, in tiiis embodiment, a uso* may make actional changes, which are reflected in updated soipts that are again updated automatically without user intervoition.

[0092] At a stq> 1018, the effects of installing flie tracking scripts can be tested.
This process is similar to that of 922 in FIG. 9 in one embodimoit

[0093] Many different arrangements of the various conqwnents depicted, as well as
components not shown, are pos»ble without dqpaiting fiom die spint and scope of the present invention. Embodiments of the pnaaat inventicm have been desaibed witii the intent to be illustrative rattier than restrictive. Altenurtive onbodiments will become appaimt to those skilled in the art that do not depart fix>m its scope. A skilled artisan may develop altonative means of implemraiting the aforanenti