Bidding and Algorithms
  • All auctions are real time bidding auctions (except for Yahoo campaigns which are API based). Third party exchange supply sources are generally second price auctions. All impressions are purchased on a CPM. For CPC campaigns, we model the CTR, take on the CTR risk, bid on a CPM, but charge the CPC. The eCPM for your CTR campaigns give a good indication of your bidding CPMs. 
  • Fees including usage, optional behavioral and brand safety, and agency margin are taken out of the max bid price. For example, if a $10 mCPM is set, a $2 Behavioral audience is selected, and the max usage fee is $2 if the most expensive inventory is purchased and is optimized to spend the max, then the effective max bid is now $6 ($10 - $2 - $2) but you will be charged the original $10 if you win the auction and this is the price that will show up as Spend in reporting.
  • Algorithmic optimization, which is set by the Goal type, starts at and requires a statistically robust sample sample size at the most granular creative-placement-device level. In the absence enough data at this lowest level, information at higher, aggregated levels is used to inform bidding optimization. This hierarchy explains how, with a CTR goal selected, you can have a creative with a relatively higher CTR yet fewer overall impressions as there might not be enough information at the most granular creative-placement-device level to inform the bidding strategy for this creative. The exact nature of the bid curve and the impressions required depends on the goal type selected. For Conversion campaigns, a multiple of the input target CPA is required per placement to inform the optimization. 
  • In general, bids are reduced in price if performance is relatively weak on a given placement and bids are maxed out on relatively strong placements.
  • Max bidding: if auctions are won at a price below the max bid due to bidding optimization and you are bidding a CPM, then these cost savings are automatically passed onto you.

DR optimization strategy

  • Generally speaking, we recommend a two prong approach: first, set up a broadly targeted campaign, ideally targeting all categories, all devices, and all products. This campaign will serve as lead or learning generation campaign where you will find the sites, types of sites, devices, creative, and products that convert well. Even with this being an exploratory campaign, it should be manually optimized as well first in terms of blocking sites/placements that are spending a lot of money relative to the budget and CPA goal. Second, set up one or more finely targeted campaigns that take the learnings from the broad, lead generation campaign and hone in on one or more of its successful elements like proven white-listed sites, creative, devices, and ad types. Bids for this more targeting campaign can generally be more aggressive than the exploratory campaign since the parameters chosen have been proven to be effective. Information for targeted campaigns can also be culled from campaigns run with other partners: for instance, sites, categories or creative that have worked well.
    • Enter in a target eCPA when generating the Conversion Pixel and a target CTR when using a CTR goal. This will help focus the algos on sites that are in your goal range and help stop spend per site outside of the goal range
    • Performance over Delivery should be chosen in the General targeting section, as this will more aggressively focus on the CPA goal independent of the budget and end date of the campaign ('Delivery' will take into account the budget, end date, and performance)
    • If our conversion pixel cannot be placed on the landing page action, then we encourage the use of our dynamic passback parameters for optimization purposes to tie in our site and creative ids to the conversions as measured in your analytics software and then used for manual optimization purposes. We also recommend using the Site Tier optimization Goal Type since the sites that the campaign will initially deliver on have been historically high performers. 
    • Our conversion pixel will measure clickthrough, not viewthrough conversions. The look back window is 30 days
    • In-Feed tends to be a high performing product in terms of overall conversions
    • Consider use of frequency caps and/or relevant product purchase related day parting
    • Consider retargeting campaigns at various levels of the customer journey on site to reach pre-engaged consumers

Branding optimization strategy

  • Generally speaking, for branding campaigns that have numerical KPIs (e.g. CTR, TOS, $/PV, etc) we suggest a strategy similar to DR where the campaign is continually manually optimized by first the site/placement list according to the KPI performance per site followed by creative, device, and product blocks. We encourage the use of the extensive audience, including first and third party behavioral targeting, and campaign targeting tools available to reach your target audience using the platform. However we caution against the use of too many targeting settings crossed together or layered on top of each other in a single campaign or else volume can be a concern: for instance, the use of behavioral targeting along with frequency caps and narrow category targets can create an extremely nice audience where budget pacing can be impacted. There is always a proper balance to be sought between performance optimization and budget delivery; the pacing preference type of Delivery or Performance can help shift the balance in either way and can be changed as the campaign progresses. For instance, if under-delivering after a few weeks of Performance pacing you can select Delivery to spend more towards the end of the campaign. You could also open up targeting or raise the bids if you want to keep Performance selected. As in DR optimization, we encourage the use of our dynamic passback parameters for optimization purposes to tie in our site and creative ids to the branding KPIs as measured in your analytics software.
    • Behavioral audiences are available through Lotame and their DMP partnerships
    • First party audience uploads are available through Liveramp, Oracle/BlueKai, Lotame, and Adobe DMPs
    • White and black listing of sites is permitted to reach pre-qualified audiences
    • Campaigns broken out by different verticals/categories in the Supply tab allows for different audiences to be targeted with different creative
    • In-Feed native video is a great branding tool where the video plays on-page and overlaid in the publisher content feed. VAST/VPAID, MP4, and YouTube links can be served In-Feed. This campaign must be set at a CPM price
    • Oustream auto play video is a great tool for getting cost efficient plays and completed views. The campaigns must be set at a CPM price.
    • Viewability can be selected as a Goal type and pricing metric to drive viewable branding impressions
    • Pre-bid brand safety technology is available through our integration with AdmantX
    • Pre-bid page level contextual targeting is available through our integration with AdmantX
    • Pre-bid fraud protection is available through our integration with IAS
    • Brand safety, viewability, and brand studies impression tags are accepted allowing a full independent analysis of your branding campaign

Creative Tips:


Reach

  • Arts & Entertainment, News, Business, Tech, and Sports are the largest categories respectively. If you are having trouble delivering, please consider the use of one or more of the categories if relevant.
  • Site Tier 3 should be selected to start and then moved down based on delivery to hit the most premium supply if on pace
  • If using Behavioral Targeting, we recommend targeting all categories since the audience is being pre-defined at the user level. If contextual targeting must also be used, please use one or more of the above categories. 
  • Retargeting reach can be limited since user ids must be set and synced with cookies in third party exchanges. 
  • Countries with the most supply are the US, UK, CA, UK, and AUS

General tips

  • If bidding a mCPC, ~$1.50 tends to be the floor for sufficient volume (CTR dependent)
  • If bidding a mCPM, ~$4 tends to be the floor for sufficient volume 
  • For In-Feed and Outstream video, ~$8 floor CPM are suggested, as these are premium, competitively sought, highly viewable placements
  • In general, your bid price is a combination of your performance metric (CTR, CPA, or Engagement) and your input max CPC/CPM

  • All fees (e.g. data and usage) get taken out of the bid price, so the max price should be adjusted upward 

  • If wanting to spend less, reducing the campaign budget is generally preferred over reducing the bid price since the latter can impact the quality of sites the campaign can show up on. 

  • Mobile and Tablet traffic tends to be cheaper and have higher CTRs
  • When pacing behind, consider raising the bid price or opening up the targeting parameters (e.g. adding more categories, removing a frequency cap, adding more day part hours, site tier, etc)