Internet Search

In January 2021, MediaRadar established a partnership with Adthena, a leading global provider of search engine intelligence. With this partnership, Adthena's best-in-class coverage & methodology will serve as MediaRadar's source for ad spend & keywords.

In January 2021, MediaRadar established a partnership with Adthena, a leading global provider of search engine intelligence. With this partnership, Adthena's best-in-class coverage & methodology will serve as MediaRadar's source for ad spend & keywords.

MediaRadar receives a monthly summary report detailing estimated search expenditures by destination URL. All destination URLs are visited by MediaRadar specialists in order to more accurately map the site to a product in our MediaRadar Central taxonomy, enabling the integration of search & multimedia ad spending at a product level.

Accounting for Close Variants & Misspellings

Branded terms that have been identified by Google's search algorithm as misspellings or close variants to the canonical search term have been removed. Misspellings and variant terms often inflate estimates as their baseline volumes are not easily differentiated from the core term in Google Keyword Planner. While the removal of these terms results in the reduction of estimated spend attributable to text ads on both desktop and mobile devices, their removal provides a truer picture of advertiser activity.

Brands with Two or More Top-Level Domains (TLDs)

During the course of data collection, Adthena will routinely monitor multiple versions of the same brand domain on Google search results pages. For example, European brands will often register a country specific TLD (e.g. .FR and .UK) in addition to the more common.com TLD. In these instances, Adthena actively works to clean the data to aggregate domains to a single brand and eliminate duplication of reporting. Note that all TLD variations are collected and are based on the results from local Google queries generated within the country being reported.

If the URL only includes a domain name, we map to a product for the advertiser that best represents the domain name.

If the destination URL includes a domain & subdomain representing advertisers owned by the same company, we map to a product for the more specific member of the pair. If the two advertisers are owned by different companies, then both entities are reflected in the mapped-to product with the domain related advertiser listed first.

Some destination URLs contain a subdomain that does not represent an advertiser but instead refers to a sub-section of the site. Whether or not to include this detail when selecting the mapped-to product is determined based on the defined methodology for the category that the advertiser falls into.

Adthena uses a multi-step process to estimate search expenditures. Outlined below is an overview of the variables & resources informing Adthena's spending estimation.

Keyword Coverage

Adthena systematically monitors a broad list of keywords in each market at least three times per day, collecting information about the brands featured in the search results as well as their relative position. Please note that keywords are predominantly focused on high volume branded and generic category terms, while less emphasis is placed on collecting near variants and misspellings. This is to ensure efforts are focused on reporting of high value terms with the brand's active list of terms. Adthena's observational data provides both breadth and depth of a brand's search activity. In the U.S. Adthena monitors approximately 300,000 keywords daily.

Source Data for Key Variables

Adthena sources data from Google Keyword Planner to establish baseline volume estimates. Information collected from the Google API include impressions, national CPCs (local pricing is not an input), variant terms related to the primary keyword, and CTRs for each keyword. These data are augmented by various third-party reporting and other sources available in the public domain.

Estimations

Adthena cleans and enters key variables into its independent models to estimate impressions, click through rates (CTR) and cost per clicks (CPC). The outputs of these models are combined to produce search expenditure estimates for each advertiser / domain collected in its observational data of brands. Adthena's models takes into account the brand's frequency and position in the search results related to each unique keyword. All models are periodically tested, refined, and validated against known advertiser search spending. It is important to note that Adthena's estimates reflect a theoretical maximum number of clicks across all its observational data.

Additionally, expenditures reflect 'average national pricing.' As such, the estimated spend reported for any individual advertiser will differ from the actual amount paid. Spending estimates are based on the toр 250 terms supported by the brand within the monthly period, and as such may represent a subset of keywords the advertiser may be supporting. Each month, MediaRadar receives data detail for the Top 20,000 domains. The Top 20,000 are based on those domains active in that specific month of collection. In any given month, if a domain appears in the monthly rankings for the first time, historical data for that domain will bе backfilled for up to 12 months, as available.

Model Maintenance

Adthena uses proprietary models to estimate spend, clicks and impressions based on prior months and predictive trending. They run daily estimations for impressions, clicks and spend on every keyword + advertiser combination according to available monthly search volume, daily ad appearances and daily search term performance data. In doing so, their model automatically adjusts to any changes in the market, even some seasonality shifts.

Occasionally, Google will make algorithm changes or introduce updates to their ad formats. When this occurs, Adthena's model will absorb those changes by using an 'auto-refresh' methodology, i.e. new data is found, the model is retrained, updates are made to CPC/CTR, calculations for traffic and spend updates on a go-forward basis. This consistency in model adjustments is responsible for the reduction of massive restate efforts over time.

Finally, Adthena conducts reviews of their CPC data for accuracy and correlation to Google's averages. These reviews incorporate recent daily activity to reset estimations and bring the estimates in line with expected market fluctuations.

Keywords & Clicks

Adthena also provides modeled estimates (not actual empirical counts) of clicks per keyword for each domain. Keyword clicks are calculated by multiplying the output values from the CTR and ad impression models (Clicks = Impressions* CTR). For keywords with low impression volume, it is possible for the modeled click estimate to be less than 1.00. Overall, Adthena estimates the theoretical maximum number of clicks across all its observational data.

Historical Data

Reporting of Desktop and Mobile / Text and PLA spending commences with January 2021 data forward. Due to issues in device, format and keyword coverage, historical data prior to January 2021 is not available and will continue to reflect estimates based on Desktop Text only.

Aggregate Spending Perspective

Search estimations will reflect coverage across Desktop & Mobile / Text and PLAs reported in aggregate, as one number.

Impact on Spending

An analysis was conducted on the impact of spending due to the expanded coverage beyond Desktop Text only. When comparing total Desktop & Mobile / Text & PLA spending to Desktop Text only, Adthena's spending estimates trend between 2-3 times higher than Desktop Only. Trends will vary by category and brand.