The core optimization of Adspert is based on the market curve theory, meaning the modeling of the market by all available observations. This contains especially the estimation of market curves to the market volume (clicks) and competitive pressure (effective CPC) dependent on the maximum CPC. Further market curves are for instance the function of CPCs to available, relevant KPIs (i.a. ROI, CRR, profit).
In addition, Adspert ascertains the conversion rate, and in case of dynamic shopping basket values also the conversion value, by information inheritance. With the help of scheduling bid adjustments, Adspert optimizes seasonal fluctuations (weekday and time effects). Furthermore Adspert models a possible conversion delay.
Adspert also supports all possible Google bid adjustments. That means in detail bid adjustments for device, demographics, location, ad scheduling, remarketing lists for Search and Display as well as the Display targeting methods interest and topic.
The Adspert algorithm is topped off with several heuristics of the portfolio theory, i.a. inheritance algorithms.
Period under consideration and evaluation
Adspert considers a historical period of up to one year. At it, there is no predefined evaluation of certain periods but Adspert decides dynamically on the basis of the particular database. For accounts with a low amount of conversions, Adspert considers a large period. By contrast, if there are greater conversion numbers, Adspert refers to the more current part of the history.
Consideration of Google Analytics data
Adspert does not consider directly data from Google Analytics for the optimization but takes account of possible imported goals from Analytics to AdWords that occur as conversions in AdWords. Adspert does this once a day at night. Generally, Adspert gathers all relevant data for the bidding process from AdWords, meaning impressions, clicks, conversions, conversion rate, maximum and effective CPC.
Bids for new keywords, ad groups and campaigns
Adspert is characterized by the fact that even for a small amount of data accurate forecasts for the long tail optimization can be made. For new keywords, ad groups and campaigns or for those already existing but having few data, Adspert starts with the current set bids. Those bids will be even increased by Adspert to generate and evaluate traffic. For a wide range of new keywords, Adspert is very careful and applies an information inheritance mechanism to limit learning costs. For bid increases, Adspert prefers keywords for which it expects higher clickthrough rates or higher conversion rates.
Frequency of goal adjustments by the user
When setting a new goal, one must consider the time Adspert needs to learn. Usually Adspert needs a few days to adjust itself to a new goal. Moreover, one must take factors like conversion delay into account. From those two components arise the recommended temporal minimum distance between goal adjustments. Considering the mentioned factors one can adjust the goal setting at any time according to requirements.