Launched in June 2022, Knytify already has a fully operational MVP and has predicted more than 30k cases of fraud propensity. CEO and co-founder Aurélien Savart tells us more about his new startup. The solution is freely available until 2023.
Your mission is to “build a fraud-free world for digital advertising”. Why is advertising so full of fraud and how does Knytify expect to change this?
Serious evaluations like the one made by straits research estimate that one-third of the overall media budget spent is involuntarily dedicated to fraud. As protections against payment fraud have evolved, fraudsters are seeking new and easier ways to commit fraud, including digital advertising fraud. Digital advertising is a more vulnerable target in this context.
The weak point in the fraud detection solutions offered by the market is that these solutions are reactive. They state fraud after it happens, whereas Knytify is estimating the propensity of fraud in real-time (under 50 ms). In this way, we empower advertisers to take decisions in real-time to prevent fraud before it happens.
How much money do companies lose due to fraudulent ad spending?
Estimations vary, but €100 billion a year is a common estimate for 2023. The same sources estimate that one-third of the digital advertising investment is spent on fraud for every advertiser. As our solution is able to detect 95% of digital ad frauds, with simple computations considering the previously mentioned number we are able to leverage the average digital advertising ROI by 350%.
How do you distinguish yourselves from your competition?
The two main points that distinguish us are the prediction aspect and the fact that we are addressing every ad channel (from displays to big networks like Google or Facebook) for every kind of product or service (web, mobile apps, etc.)
What mechanism allows your solution to prevent fraud from taking place in the first place?
Our solution consists of a proprietary machine learning and AI algorithm that is able to predict the propensity of a user to be a fraudster. We use more than 60 indicators and thousands of value combinations to detect if any user is what he is pretending to be.
The whole process, including server communications, is completed in under 50 milliseconds. Then the users of Knytify solutions decide by themselves which propensity of fraud is acceptable. As our solution is available through API and integrated in most of the ecommerce CMS, our users are able to decide the kind of reaction they want and the “hardness” of their fraud shield.
Let’s take two use cases to illustrate this. Case 1: You are a small ebusiness, you have integrated Knytify through the prestashop plugin. You can configure it in less than 5 minutes so that you are asking for additional confirmations for users that have a 20-40% probability of being fraudsters. From 40-60% you are automating some tests (like captchas for example). On the other side, you are not allowing users with a propensity above 60% to even reach your conversion funnel.
Case 2: You’re working for a very well-established company and are trying to acquire new leads and concentrate efforts on the most valuable one. In this case, you will for example simply and purely reject leads which have a risk above 33% of being a fraudster. From 10-33% you direct them to your marketing automation, and between 0 and 10 you pass them on to dedicated business specialists.
Which businesses are most in need of your solution?
Basically every advertiser on the web, regardless of the ad networks they are using and their maturity level. Of course, we do have a freemium strategy that aims to demonstrate the power of our product rather than simply claiming its benefits. With our focus on the respect of GDPR and the general inclusion in the European ecosystem, our natural market is European advertisers of every size and technical maturity.
What are your goals for the rest of the year?
As we have finished our MVP, we are now 100% dedicated to increasing traction. We want to demonstrate the market fit of our product and we are actively seeking users, even free ones.
