FICO Falcon Platform Reviews

I use FICO Falcon app manager. I think it was from 2019, but I'm not quite sure. Maybe it was 2017. I'm a customer of the product, and I use it for bottom-line and operational risk.

For the granular referral, we use it to predict whether a transaction is fraudulent or not. So, it has the machine learning capability that FICO has in its Falcon platform.

We're looking for tools that can process transaction data, like the transaction amount, credit card number, or card number, to predict whether a transaction is fraudulent. This is what we use in our day-to-day cases.

The first feature is its real-time capability. For example, if a transaction is deemed fraudulent, our analyst and rules can run automatically to prevent the transaction from happening in the first place. In other words, it can make real-time decisions or real-time declines.

The second feature is the consortium data model. Not many profilers have this because it requires a lot of data, especially transaction data from customers. I'm specifically looking for products from companies with a significant number of customers, particularly banks. 

The functionality is fine, but the price is getting more expensive, so I need to find one or two alternatives to FICO.

The consortium model seems AI enough, even though it's just machine learning. , but AI is an umbrella term.  Maybe my imagination is limited, but I don't see significant improvements they could make beyond what they're already doing.

FICO users are business users, not developers. When setting up rules or parameters, some coding is required, and the learning curve is quite steep. This is a disadvantage of FICO. 

Additionally, the offline analysis of transactions is not flexible. Transactions need to be flagged as fraud to show up in the analysis, making it difficult to identify false negatives.

The Identity Resolution Engine (IRE), an extension to FICO Falcon Platform, is helpful. However, integration with FICO Falcon Fraud Manager can be challenging due to its rigid template and data format requirements. This can cause delays in development and integration, potentially postponing projects by a month or two.

AI is another thing they could improve is how they create the data model for my bank's detection data. From what I understand, they already have the capability to do that. But the data being processed in my bank, the ones they can process, is not really much. Maybe it's just a mistake on my part. But in terms of AI other than machine learning that they've already implemented, I have not seen any significant use case.

I have experience with this tool. 

It's not exactly stable. I would rate the stability a six out of ten.

The problem with FICO is that the system sometimes has elevated response times, which can be quite a nuisance, especially during events like discounts on our banking application or during the onboarding process. 

Our system is very critical because transaction execution needs to happen through FICO before it can go to core banking, and that's the problem. The impact it has is quite critical. So, if FICO could increase response time maybe I could give it a nine. 

We are an enterprise business. For retail transactions, FICO is really good because it's designed to tackle all transactions in real-time. So far, it's quite satisfactory. 

But, my company is trying to expand its transaction channel to wholesale customers with very different behavior. 

So scalability across segments, it's not as relevant, but it still does the job. For retail customers, I'd rate it an eight, but for wholesale customers, maybe a five. I'm quite satisfied because it does the job.

I'm actually quite satisfied. The customer service and support are quite responsive and solid, so there's no complaint regarding support. 

It's just annoying to have a technical problem that our team could have tackled if we had access/insight. But because we don't know what's inside, we need to call their customer support.

Positive

In terms of the system, it's a bit complicated, but I'm not quite sure whether it can be regarded as very complicated or mildly complicated because it's all technical, and I'm not really from the technical team. 

But in terms of the parameter deployment, the one we use to capture fraud, it's actually quite straightforward. So, no complaints there.

Deployment time:

To deploy the solution is not entirely on FICO because there are some internal regulations and governance that we need to take, which can also impact the deployment process. But for the migration process, the one we use to move from the deployment environment to the production environment, it can take around one night or two nights. We always do it in the middle of the night.

Integration challenges:

FICO's Python management has a rigid template and data format that needs to be followed. The development and integration process can take a while because different teams manage it, and not everyone is accustomed to the data specifications required. This can postpone the project by up to a month or two from the target date.

Maintenance:

For maintenance, the main critical point that needs to be fixed with FICO is how we can monitor its product. 

For our technical team, FICO is like a black box. When something goes wrong, it's very hard for us to pinpoint the problem because it usually happens within the rule engine, which is confidential and only FICO's support team can access. 

Every time there's a problem, we need to call up FICO for technical support. I don't think the same problem applies to other fraud detection systems we have. If there's another system that can be more transparent than FICO, that would be ideal.

We have engineers and architects for this work. I haven't encountered any issues. Maybe around 15 people, give or take. 

But I need to make it clear that those 15 people are not only from my bank but also from the consultant, the implementation partner, and FICO themselves.

It is expensive. 

I'm quite interested to know more about Fiserv and FastFire, and also about the new product that comes out from GBG, the next-gen product.

I'm still in the process of finding out more about them. It's still preliminary research.

Overall, I am satisfied with the product. I would rate it a seven out of ten. 

Falcon identifies anomalies in bank customer behavior to detect fraud. For example, say a person in Sao Paulo consistently spends $200 a week for 10 years, and we see a $2,000 transaction one week. Falcon can set rules to identify suspicious activity at each point of sale during every transaction. 

Let's say we see seven million transactions on a given day. In all those transactions, we might detect 20 people potentially committing fraud. For me, 20 people are too many because that could mean 140,000 transactions that we need to refer to a call center for follow-up with the customer. 

We need a customer service rep to call and say, "Oh, hi, ma'am. Did you make these transactions? Yes or no?" It's so much work to validate a transaction. Without Falcon, we would need much more staff to call customers and verify transactions, which means we would spend considerably more to make this work. 

I previously worked at Santander Bank FICO and Oracle. It was only about six months. FICO is working on a local model specifically for Brazil, whereas tools from Oracle and other vendors are only developed for general use. I haven't seen any other vendors who think like that until now.

Our solution here in Brazil doesn't work for Europe, Africa, the USA, Canada, or Mexico. It's unique to Brazil and designed based on the transaction patterns of people living here. 

FICO has all the information on our transactions, and the logs of old transactions are on the central server. It's the same for other banks like Fordesco, Santandar, CitiBank, etc. All these banks send their transactions to the server, and FICO Analytics uses all the information about transactions in Brazil to build a model.

Fraud will always be a problem here in Brazil, and I think blockchain technology is the future because it increases transparency in transactions. Even with Falcon, we need to spend time verifying transactions. I think a better way is to put the transactions in a blockchain. 

You can put these people on a block list if you know who made the fraudulent transactions. There is no way to do that because nobody has a blockchain solution, but I think that's where the industry is headed. Everything will be blockchain in the future. 

I worked with FICO for three years in Brazil and later as a consultant configuring Falcon for a bank. 

Falcon is stable.  The only problem we had was caused by a lost connection. To solve that problem, we set up a second account with another telecom provider, so if we lose connection on one, we can switch. 

Falcon is scalable. At first, we had 5 million daily transactions, but it increased to 7 million as more people began using credit cards. We needed to implement a solution to handle more transactions in the same amount of time. In AWS, it is easy to add servers and VPNs if you need more resources, so it was no problem for us. 

I've had bad experiences with FICO support. At the bank, we've faced critical problems where we only had minutes to find the solution, and it's hard to explain quickly. We contacted FICO, but they told us it was beyond their scope of support. They try to triage the call by asking us how much money we're losing, but I can't go into that. They have people responding to but not solving the problem. It was a frustrating experience.

Setting up Falcon is complex, and it takes a long time because you have to work with several people to configure the rules and tune them for all the account holders based on user behavior and the markets. 

It's hard to say how long your deployment will take. Once you establish a model, it takes some time to know if it works. It's based on long-term patterns in behavior. One day of transactions isn't enough to establish a pattern. I need at least 12 months of data to set a behavioral rule for a customer. However, most banks won't wait a year. They will say it's good enough after six months and give the go-ahead to implement the solution. This is one of the reasons why our model isn't working as well as it should. 

If you ask how long it takes to deploy Falcon, I would say it depends on what you want from me. If you don't have historical transaction data that you can use to build a model, it will take at least one year. If you have 20 years of data, we can implement it now with no problem.

When I was working at another bank, people often complained that FICO was expensive. I would tell them that it depends on the volume—for example, the old bank million 7 million transactions. If you don't have an adequate system, you will lose more money than you would spend on Falcon. FICO can pay for itself in one or two days if you compare it to the money you would lose otherwise. 

FICO is one of the best fraud detection platforms. However,  a small bank might find it too expensive. In that case, I would agree it's probably not worth it. However, it's a no-brainer if your bank has 50 million accounts and 7 million credit card transactions daily. 

I rate FICO Falcon Platform 10 out of 10.