Frequently Asked Questions
How much work is it to integrate?
All that's required on your side is integrating our tag on your pages. It is fully asynchronous, non-blocking, and extremely light. Integrating it requires zero engineering work if done via a tag manager such as Google Tag Manager or Tealium, or a couple of hours a work to drop our JS directly on your page. Once our tag is on your page, we do the rest with no additional work required on your side: we build a custom model, deliver the predictions back to your site in real time, and send the predictions into the appropriate systems and workflows for activation.
Will your tag slow down my site?
The Black Crow AI tag is fully asynchronous and non-blocking. It’s extremely light (less than 10KB zipped) and optimized for minimal thread usage. Finally, it's a one-time download, thereafter it is cached client-side in the browser -- minimizing additional requests.
What types of data do you use? Do you use 3rd party data?
We only use your first-party data — from your users and how they behave on your site. We don't use any third-party or network data.
We do this for three reasons:
1. It's sustainable. In a world of GDPR, CCPA, iOS14, and cookie-less Chrome, solutions and products built on top of first-party data are sustainable.
2. It's customer safe. We don't require tracking users as they browse the web in order to make high quality predictions.
3. It's powerful. The strongest signals come from how a particular user behaves on your site, not on what he or she may do elsewhere on the web.
Do you comply with the California Consumer Privacy Act (CCPA)?
Yes.
Is Black Crow AI GDPR compliant?
Yes.
Will Apple's iOS14 / IDFA changes impact Black Crow's ability to predict future user value?
No.
Black Crow AI predictions are made entirely using our partners' first-party data. We don't use any third-party or network data. We do to this for three reasons:
1. It's sustainable. In a world of GDPR, CCPA, iOS14, and cookie-less Chrome, solutions and products built on top of first-party data are sustainable.
2. It's customer safe. We don't require tracking users as they browse the web in order to make high quality predictions.
3. It's powerful. The strongest signals come from how a particular user behaves on your site, not on what he or she may do elsewhere on the web.
How else can I use the scores other than in paid marketing?
These are your predictions, you can use them however you want, we make them available everywhere and anywhere you'd like to use them. We like to start by proving their value in paid marketing becuase (i) it is such a large expense for most of our partners, (ii) our predictions are proven to drive material, P&L impacting upside in paid marketing performace, and (iii) we can prove and unlock that value within 30 days of completing the model build. From there, we can funnel the predictions into any other workflows or systems to drive similar improvements.
Don't Google and Facebook already use ML to optimize our campaigns? How is this different from Smart Bidding?
Our product is complementary to, not competitive with, Smart Bidding and other ML-driven optimization features offered by Google and Facebook. Our predictions supercharge Smart Bidding's performance. Here's how:
1. Google and Facebook use the conversion rate of any given audience to inform their smart bidding algorithms
2. We create custom, ML-learned, predictive audiences for our partners. These audiences have significantly different conversion rates from the average. We 'de-average' your users.
3. When our custom, predictive audiences get fed into the Google and Facebook optimization algorithms, it lets them work better -- to bid up on audiences with higher predicted value, and bid down on audiences with lower predicted value, driving big gains in overall profit delivery from your search and social spend.
Two other ways in which our scores are unique:
1. They're portable: you can integrate them into any system or workflow across your business, from paid marketing, to email, to CRO, and more.
2. They are built off of a custom model made just for your site and visitors instead of aggregated network data.
How is Black Crow different from Amazon Sagemaker, Dataiku, Data Robot, or Data Bricks?
If SageMaker, Dataiku, and Data Bricks are hammers, Black Crow AI is an entire house.
Those products are suites of tools that allow organizations to build their own, in-house, ML solutions. They enable data scientists, data engineers, and analysts to experiment, build and deploy models, and generate predictions. Making effective use of these platforms requires material investment: staffing up a team of experts to make use of the tools, structuring and normalizing data, determining what predictions to make and what business problem to solve with them, and then iterating and experimenting. They are the hammer.
Black Crow AI takes all of that work away, and replaces it with a plug-and-play platform that delivers custom predictions in real time, on every page load, in milliseconds. We handle data normalization and structuring, model training and performance, daily retraining, and more. We provide guidance on how to deploy those predictions to drive real-world benefit, based on our years of experience with over 300 ecommerce partners. Our predictions are untethered to any single system or platform, and can be integrated anywhere. It's the full house!
How much does this cost? What's included in the price?
The price you'll pay is driven by the volume of your visitor traffic. For all but our very largest partners, our annual pricing is substantially less than the cost of a single data scientist.
