Person
1 episodes · 22 quoted claims
Ciprian Porutiu says his role at Marsh has covered finance transformation, CRM work and setting up new products.
“Fortunately, the the role has been very kind to me and allowed me to run in all corners of the organization, from finance transformation to CRMs”
Porutiu says his work includes deploying AI tools for both colleague productivity and client- and partner-facing products.
“deployment of A I tools, both for colleague productivity, but also for the product, so client-facing and partner-facing applications.”
Porutiu describes financial results such as earnings per share discussed in quarterly reviews and earnings calls as lagging indicators.
“it's basically the the lagging indicators, the ones that everyone discusses in the quarterly reviews, in the E P S and every earning earnings calls.”
Porutiu says leading indicators beneath the surface create the conditions for reliable lagging results.
“But what is underneath it's the leading indicators and what creates the per the the the context and creates the situation for the lagging indicators”
Porutiu says different teams pull finance data from different sources, including fiduciary systems, master data, invoicing, Excel and emails, even at large firms.
“Someone is taking the data from fiduciary, someone is taking the data from some master, someone is taking the data from, I don't know, invoicing or anything else, or or even Excel and emails”
Porutiu uses a traffic analogy to argue that governance slows you down briefly so you can accelerate afterwards.
“the laws of um, circulation of traffic, it tells you go down to twenty miles an hour on that curve because you're gonna go off the road.”
Porutiu says organizations should establish a source of truth and standardize systems before automating.
“So you have to have a proper sequencing in which you establish the source of truth. You standardize the the systems”
Porutiu argues the solution is aligning on language, taxonomy, dictionaries and federated data domains rather than building one database.
“It's not a question of one database, it's more around aligning on language, aligning on standardization of taxonomy, dictionaries, and even federated data domains, you know?”
Porutiu says that, like errors in large language models, errors between manually bridged systems compound, and proper governance prevents rework.
“the errors compound, they don't stay with the stay at the same level, and it it prevents, if done properly, a lot of rework.”
Porutiu recommends phased rollouts that start with low-hanging fruit and scale up, to build trust.
“So this is where you go phase rollout, you know, you, you apply all the principle, the known principle of laws hanging fruit and take small and and scale.”
Porutiu says involving users in design and testing creates a sense of ownership when a system goes live.
“No, you bring them in the testing, you bring them in the design, and then by the time goes live, they're gonna be, oh, but this is mine.”
Porutiu says organizations should decommission old systems as a guardrail, paired with showing users what is in it for them.
“you definitely need to have both the the guardrails, you know, which is the stick, which is a certain point you actually decommission and prevent people from going there”
Porutiu describes the finance value chain as subledger, transaction processing, reconciliation and then reporting.
“So when you go from sub ledger put in order, you go into the transactioning processing in order, and then you go to reconciliation, and then you go into reporting.”
Porutiu gives the example that revenue can mean gross written premium to one team and net earned premium to another, so teams speak different languages.
“if for one revenue means gross written premium, for another one is the the the the net earned premium, or for one is the financial definition, for the other one is the sales definition”
Porutiu notes that revenue can follow US GAAP in one team and international accounting standards in another.
“for one is the gap definition, the other one is for the is the international standards definition.”
Porutiu says presenting data foundation work as internal cleanup makes it a sunk cost that executives won't fund.
“If you put it only as an internal cleanup, well, this is why I'm, you're a sunk cost.”
Porutiu says technology does not fix finance operations but exposes what is already broken.
“Yeah, I think, again, the technology doesn't fix finance operations, it simply exposes what is broken.”
Porutiu argues that the larger the organization, the more counterproductive a single database becomes, and data should be organized in domains owned close to the ground.
“Actually, the bigger the organization, the single database is counterproductive because if we organize our data in data domains in which the ownership stays closer to the ground”
Porutiu gives claims, revenue and servicing data as examples of separate data domains.
“if it's like a claims data, or if it's like a revenue data, or if it's like servicing data, it's very important”
Porutiu says a shared taxonomy should define basic terms such as account, active account, cost center, products and transactions.
“What's an account? What's an active account? What is a a cost center?”
Porutiu says the ability of data domains to connect through APIs is almost more important than the data itself.
“it's almost more important, the A P I than the than the data itself.”
Porutiu says that with a shared language, data can flow between finance, claims and underwriting domains.
“Then finance can take claims data, the claims will take, uh, you know, underwriting data, and everything else will flow.”