A.I. & Optimization

Advanced Machine Learning, Data Mining, and Online Advertising Services

Programmatic Remarketing Optimization - Know-How



Background

Qudos is a B2B marketplace providing a matching engine to search business services. A company who is looking for a business service provider can use Qudos vertical search engine to find for example the right Public Relations firm who understand their business helping them with their press release.

Qudos B2B Marketplace
Fig 1. - Qudos B2B Matching Engine (source: www.qudos.com)

Setting UP Remarketing Campaign

Qudos takes advantage of different advertising channels to target people who are searching to hire a firm for their company. A prospect will be taken to one of Qudos landing pages where they can start searching for services they are looking for. However, the B2B hiring cycle usually takes in order of weeks or months. This is where a Remarketing strategy can play a crucial role.

AI Optify has worked with Qudos to setup their remarketing campaign. The goal of remarketing strategies was to increase conversion rate which consequently minimizes the eCPA (i.e. cost per acquisition).

The Challenge

After starting the remarketing campaign, there were three ad buying variables that we needed to optimize: (I) CPM bid value, (II) Recency parameter, and (III) Frequency Cap.

The Solution

(I) Optimizing The CPM Bid

For calculating the optimal CPM bid you can start by applying the following linear formula:

CPM Bid = CVR * Goal Value * 1000

You need to set your "Goal Value" (i.e. how much a conversion worth to you) in $ and initilize the "CVR" (i.e. conversion rate). For CVR you can start with your prospecting campaigns' CVR and optimize it from there as you spend money on the campaign and collect stats from your campaign.

Roughly speaking you should be able to achieve higher CVR for your REM campaign than your prospecting campaigns. This is because you're re-targeting users who have already expressed interest in your offer.

(II) Optimizing the Recency

How long do we need to re-target a user after the first exposure to your ad? This is the recency parameter. It's clear that longer we re-target a user less likely they become to convert. This is because they lose their interest in your ad over time. So we need to find out the optimal recency for remarketing strategies.

If we pick a long recency value (e.g. 90 days), we will waste a lot of money. On the other hand, if we pick a short recency value (e.g. one hour), we might lose many potential prospects. The optimal recency should be set based on how long it takes for a prospect to proceed and convert.

(III) Optimizing the Frquencey Cap

Lastly, we also need to optimize the campaign's Frequency Cap. We suggest to apply a small freq cap at the strategy level and a higher value at the campaign leve. You can use your CTR (or CVR) to make sure that your Freq Cap is not too high in which case you will have a high eCPA or too low in which case you will lose the opportunities to retarget the potential prospects which decreases your CVR.

The Results

We have measured the performance of Qudos campaigns pre-optimization and compared it with post-optimization. Our performance results show that our remarketing optimization effort has decreased Qudos campaign's CPC (i.e. cost per click) by 55%, increase the CTR (i.e. clickthrough rate) by 120%. And finally the campaign's eCPA has been decreased by 26% while the CVR has been increased by 34%.


So clearly our optimization effort has paid off by moving Qudos campaign in the right direction. At A.I. Optify we design customized machine learning algorithms for sophisticated media buyers.

If you're interested in learning about our machine learning services or the AI-powered bidder, please contact us at:
info@aioptify.com