Easy

Launch Process

Brief on launching new campaigns with details on each step of the process.

Defining Campaign Goals and Targeting

Each campaign has a unique set of goals (pageviews, leads, purchases, upsell). Upscale Digital DSP is capable of optimizing towards multiple goals simultaneously.

On this stage we determine what goal steps are measurable and define target price for each of them. Usual set up measures up to 4 events:

  • Landing - user visits landing page
  • Clickout - user takes action on landing page
  • Lead - user submits his information
  • Conversion - user makes a purchase

Targeting settings usually are defined by GEO, regions, device type, OS, OS version, time schedule etc.

1 or several creative variations are uploaded to the DSP

Each creative consists of following information:

  • Landing URL
  • Display URL (end URL)
  • IAB category
  • Title, up to 60 symbols
  • Description, up to 150 symbols
  • 1 image with size over 800x600
  • Call to Action text, up to 15 symbols
  • Sponsored by text, up to 15 symbols
  • Language

Depending on placement requirements all of the above data can be shown on publisher site or only part of it. Image will be automatically adapted to exact size/ratio required by SSP and served from closest CDN server.

Unique pixels are generated for each campaign, client sets up postback (S2S or image pixel) for each event type.

Test conversions are made to generate pixel events and validate that everything works properly.

Campaign begins it's Explore phase when it tests out given creatives and different placements.

Data on each winning bid is stored in data warehouse to be further processed and analyzed. DSP begins to understand where it should bid and with what price.

Once there is enough data to make a prediction, Exploit phase comes in buying ad impressions that will bring conversions for given CPA goals.

Every additional dollar spent provide more data for Predict Optimization to learn from and the algorithm fortifies it's knowledge.

More campaign budget is allocated to Exploit bids and less spent on Explore bids. Ideal scenario for the campaign is to have 70/30 Exploit/Explore ratio, which will allow to scale campaign while hitting the target and keep exploring for new data.

Once campaign starts hitting the target CPA amd bringing in profit (this usually takes from 2 to 14 days depending on niche) it can move on to scalng stage

When Machine Learning model picks up on profitable traffic patterns, DSP will try to maximize the most profitable traffic slices within the current campaign targeting and can be immediately applied to new inventory or creative.

This stage also involves adding inventory from new SSPs, exploring new placements and testing creative variations.

Stay in the know

"Let me know when you are out of Beta
or have any substantial updates"