1.1 Online conversion pixels
 Return on investment – for example if the pixel is on the thank you page, it can reflect the profit from the conversions number and the product quantity
 Keywords – helps in deciding what is the best performing advertisement/phrase by measuring the conversions number on all ads
 Devices – based on the activities done on every device the advertiser can choose the best device
Predicting this conversion rate is thus essential for estimating the value of an impression and can be achieved via machine learning. One difficulty however is that the conversions can take place long after the impression, and this delayed feedback hinders the conversion modeling. This problem can be resolved by applying models that integrate the conversion delays. Such model is presented by Chapelle (2014). The model predicts the probability of future conversion by dividing the users in two categories – one that is treated as a negative sample and the other that should not be discarded yet.
Conversion tracking code often referred as conversion pixel can be set on any page of the site. These outcomes (conversions) are recorded using tiny images called pixels that allow the advertiser and the platform to know which users visit which parts of the advertiser’s website (Johnson, Lewis & Nubbemeyer, 2016, p. 12). A conversion pixel is normally a 1×1 transparent image initiated by a short line of code which is most usually placed on the affirmation page or the thank you page. It starts at the point when any individual gets to the affirmation page which helps in gathering information of the number of conversions over customers. The pixel enables the website’s owner to understand the user’s action through the site. With this data the advertiser would now be able to place new adverts promoting different products, retarget his customers, or try placing an ad to a different site. There are generally two kinds of conversions: view through conversion (this is related with the presentation that is given to the user. The client doesn’t click on the ad but returns on the page later. This rate is always higher than the other one) and click through conversion (when the user clicks on the ad). Primary, there are three critical benefits from placing a pixel:
 Retargeting (differentiating groups of people who have visited the site, retargeting all through a sales process) – furthermore, the advertiser can reach to people who have viewed the product listings, or specific pages.
 Optimizing (developing higher conversions regarding a specific activity) – the advertisers can additionally utilize custom conversions in order to adjust the tracking to their particular needs
 Tracking (monitoring the success of the promotion messages, for example through the number of sales resulting from a specific ad) – the advertisers can monitor how frequently a specific activity happens.
When the advertisers place a conversion tracking on their sites, they can characterize in practice the true essence of the conversions. The platforms that offer conversion tracking enable the advertisers to track basically any action on their site by setting a code, or a pixel, on the page that a guest arrives on after making a certain desirable undertaking. In this way, depending on the end goal, the advertiser can track the conversions of interest. When creating a pixel, advertisers can, most of the time, choose form a certain classification particular to their final objectives. Along these lines, tracker makers can track buying on their online stores, enlistments to an online course or event, memberships and subscriptions, or essentially anything they might need. Some platforms, additionally, enables assigning a monetary value to every conversion. These values are a critical component that many people disregard on the grounds that they don’t know how it functions. Including a value will give a more precise examination of the ROI. Each time a conversion happens, the advertiser have the capacity to perceive how much profit he has made off that conversion, in contrast with the amount spent to get that conversion. When the advertiser has adapted (changed) the code, he then puts the
pixel on the page. For instance, in the case when is important to track purchases, the code is placed on the page the client sees when he has already done the purchase, not on the items page or the cart page. Likewise, in case when the advertiser gathers subscribers, the code can be put on the ‘thank You’ page and not on the welcome page. With the pixel set up, it can be connected to the advertisements and it will start tracking conversions. One pixel can sometimes be effective for many campaign promotions, as long as the objective is the same for every advertisement.
The custom audience is an efficient tool that facilitates the way advertisers retarget the customers. When the seller has the pixel introduced, it will track the actions of every guest who
is signed into the platform. It will record which pages on the site they visit, which pages they don’t visit, and when they visit. By using this information, the advertiser can promote the products to the most relatable customers. There is likewise the option to make custom conversions in a similar way to making custom audiences. A custom conversion is made by choosing a goal page and naming the conversion. This enables the advertiser to make custom conversions, separately from the ads, and decide when to start using them. There is also the option of inputting a monetary value. Newly introduced feature were the standard events – while custom conversions are fixed to a URL these doesn’t necessarily need to be. Rather, conversions can be followed by including an extra line of the code to the tracking page. There are different categories that can be used with Standard Events, each with its own line of code. This code will reveal to the platform what to track, and will be embedded into the new pixel code—however just on the page the advertiser needs to track changes on.
Building successful advertising campaigns on any platform is tied in with testing diverse procedures, measuring what works best, and launching more promotions like the profitable ones. The advertisers should proceed with this cycle until they’re content with the cost per activity (CPA). By monitoring the results from the ad campaigns, advertisers can learn which ad is giving more conversions at the least value (CPA). The next step is investing more finance to these effective campaigns and trying to create new similar advertisements.

1.2 How in advertising we match offline sales data via e-mail address matching?
Considerable number of the present advertisers are endeavoring to see how both online and offline customer behavior can be coordinated with their multichannel reaching strategies. Todays’ high competitive environment has forced them to integrate brand communications with clients and prospects over various channels. Clients are omnipresent, regardless of whether it’s email or social networking, and if advertisers don’t take the chance to improve the customer relationships by connecting with them over several mediums, they will suffer a great loss.
Offline data is data collected and stored in “offline” systems, like CRM platforms, POS systems and email-marketing platforms. Customers spend a lot of time offline and by tracking their activities, a detailed profile of the client can be acquired which can enhance the advertising success. As the number of worldwide email users top 3.7 billion (The Radicati Group, 2017), it’s comes with no surprise that marketers are profoundly interested in using e-mail addresses as a data onboarding platform.
It is believed that many customers may conduct product research online, or may be influenced by online advertisements or emails, but make purchases offline in response to these online
activities (US Patent No. 20040024632A1, 2004). In this case, in order to match the offline purchasing records to relevant online data a common matching tool is needed – such as an email address. At that point the matched profile – the customer’s offline purchase record, linked with his/hers online patterns – is anonymized and given an ID. This specific identifier is then incorporated into an advanced digital marketing tool (like cookie or profile store), on pages where the customer is logged on and have a database record. When this one of a kind link between the offline and online identifiers has been made, the offline purchasing activity of the customer can be related with past web searching activities and can be translated into messages like internet advertising.

1.3 ISSUES WITH COOKIES AND CROSS-DEVICE CONVERSION TRACKING
Cookies are small records which are stored on a user’s PC. They are intended to hold a limited amount of information particular to a specific user and site, and can be accessed either by the web server or the user’s PC. This enables the server to present a page custom fitted to a specific client, or the page itself can contain some content which is related to the cookie data. These cookies can, in fact, make the online shopping more convenient.
Gaining more credible information for advertisers is becoming an issue as users are continuously switching devices. This is when cross-device attribution becomes relevant. Cross-device attribution show not only when customers interact with multiple ads before completing a conversion, but also when they do so on multiple devices. This gives the advertisers valuable insight into how the customers use different devices on their path to conversion. Cross-device conversions start as a search or display ad click on one device and end as a conversion on another browser, device or app. Google, Facebook and Twitter all use their user log-in data to track cross-device conversions for those who log-in to a service on multiple devices. Cross-
device conversions start as a click on an ad from one device and end as a conversion on another device (or in a different web browser on the same device). In order to measure cross-device conversion statistics, the advertisers (or platforms) gather aggregated and anonymous data from users who have previously signed into the same service (application). Cross-device conversion data can provide additional insights into the value of the ads and campaigns. Cross-device conversion tracking system and method provides online marketers with information regarding advertisements that can later translate to purchases (Gould, 2014).
One of the biggest differences between cookies and cross-device tracking is that cookies are not capable of measurements across multiple devices. However, one thing is sure – they revolutionized the online marketing process. Parallel to their popularity growth raise the questions for privacy and security.
Privacy is a widely acknowledged as a user’s right, and consequently, online consumers have the right to prevent access to their personal and financial data. However, there are cases when third-party websites have access to users’ cookie information, without their knowledge or approval. Additionally, cookies can be obtained by these websites, with the purpose of improving the success of the advertising campaigns.
There are two types of tracking: probabilistic and deterministic (Gulay, 2016). Deterministic tracking involves recognizing personally identifiable information, like an email address, when it is used across multiple devices to log into apps and websites. The deterministic method of device ID tracking is typically seen as more accurate than the probabilistic method. As an alternative to deterministic device ID tracking, probabilistic cross-device tracking relies on algorithms. The probabilistic method tracks large number of anonymous data points from different elements tied to digital use – such as location, Wi-Fi network, screen resolution, OS, device type and so on. In the same manner as the privacy and security concerns for cookie usage, cross-device conversion tracking likewise imposes issues to misuse of personal data, as purchasers use multiple devices. Transparency, consumer choice, data sensitivity and security are integral to respecting a user’s rights to privacy and security. The Federal Trade Commission (FTC) of the United States of America has made efforts to start regulating the usage of cross device tracking (e360, 2015). However, privacy is still a concern. Users are not informed on how extensive cross-device tracking is, how their information is used for data matching and probabilistic tracking, and are not aware of information sharing with third party websites. In the case that they are aware of these actions, they are not offered the option of controlling the data usage. Most sites, however, haven’t established privacy policies and are not informing users on their data sharing practices. Users are most of the time unaware of cross-device tracking, as well as where the consequential information is going.
Internet users and buyers are gaining awareness of online protection and security issues. With the rising privacy and security issues over the utilization of cookies and cross-device tracking, a natural progression is the exposure of the tracking practices, granting the right to choose how the personal activity is tracked, establishing tracking boundaries on certain topics, and prohibition against unapproved utilization of information.

REFERENCES
Chapelle, O. (2014). Modeling Delayed Feedback in Display Advertising in KDD ’14 Proceedings of the 20th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (pp. 1097 – 1105). NY: ACM
e360 (2015). Cross-Device Tracking for Improved Decision Making. e360. Retrieved from: https://element-360.com/cross-device-tracking-for-improved-decision-making
Gordon, B. R., & Zettelmeyer, F. (2017). A Comparison of Approaches to Advertising Measurement: Evidence from Big Field Experiments at Facebook. US: Kellogg School of Management, Northwestern University
Gould, C. M., Lanser, E. W., Dahlby, J. J., & Nyhus, J. R. (2014). Cross Channel Conversion Tracking System and Method. United States Patent Application Publication. Retrieved from: https://docs.google.com/viewer?url=patentimages.storage.googleapis.com/pdfs/US20140207567.pdf
Gulay, O. R. (2016). A New Approach for Reaching the Customer of the Digital Age: Cross-Device Advertising. Journalism and Mass Communication, 19-25
Johnson, G. A., Lewis, R. A., & Nubbemeyer, E. I. (2016). The Online Display Ad Effectiveness Funnel & Carryover: A Meta-study of Predicted Ghost Ad Experiments. Simon Business School Working Paper No. FR 15-34
Perry, Morgan. (2004). US Patent No. 20040024632 A1. Washington, DC: U.S. Patent and Trademark Office
Roullier, T., Robbins, S. D., Young, L., Jones, S., & Zukerman, M. (2017). US Patent No. 9665883 B2. Washington, DC: U.S. Patent and Trademark Office