Zorak

I fully agree with what I quoted above, and anonymous comments on an internet forum should be treated with healthy skepticism. At the same time, can you prove they're wrong?

I mean, it seems to me that there's a lot of room between "we should be wary because anyone can make stuff up", and the total hostility it seems to have provoked.

blitzen

the burden of proof is not on me. You have to proof that it is a valid/reputable source ....

unsubscribing from this waste of time post

Zorak

I guess you won't read this then, but I don't need to prove anything. I don't care who believes it, and I don't know if I even believe it myself. It's interesting to consider though, and I'll just see what happens next. I don't engage in any edgy stuff anyway so it's not going to matter much to me.

All I'm saying is, it's one thing to say "this is unsubstantiated so I don't see why we should believe this" and another thing entirely to feel the need to have personal attacks on the source, unproven or not. It's not like they questioned anyone's parentage.

So I have to agree with GundamWing01, the reactions are interesting.

mia

Please write about American Express products, policies and practices, but not about each other. Make your point once, and the let it go. We are here to exchange ideas and information, not to beat anyone else into submission.

hhdl

Speaking as someone who could fairly be described as a software engineer in an area adjacent to ML, this feeling isn't wrong.

uppereastsider

A quick LinkedIn search shows that Phoenix has a ton of ML and Data science engineers working at American Express. Not trying to validate the anonymous quote, but that’s detail an insider knows unless this person went to great lengths to make a false statement

GUWonder

It’s a detail known by others too, including those with no intention of making a false statement.

Personally, I am of the opinion that Amex would be perceived by some as derelict in its duty if not running its customer data more and more often in different ways to figure out how to generate more and more member fees, to get more diversified and reliable revenue streams and profits from its customer base and target markets, and to learn about customers that aren’t as good for it. And in the process there may be consequences in the form of fired customers. Mining databases, coming up with predictive models, and then using those to make decisions is to be expected more and more.

hhdl

In general, trying to exploit loyalty and reward programs is betting against the data scientists. It's almost certainly not a long term winner.

MaxVO

The anonymoous quote claimed there were just 2 such engineers, not a small army. Of course it makes little difference to us either way. The whole purpose of that elaborate "work" was exactly what it had accomplished -- created controversy out of thin air and safely diverted our attention to a subject of little relevance to FT.

akr1970akr

I read the AXP investor presentations and they make a big deal of how much better they are at data analysis than their peers. The DoC comment sounds credible when combined with what I know from following the company over the years. Of course, it could be deliberately written that way, but it lines up other sightlines I have.

TL;DR - Amex is getting tired of churners, even though they have seemed super friendly toward them in recent years. Adjust & Adapt !

synzero

The post sounds plausible from what I know of data science and tech. One thing that stands out as implausible however is the comment “don’t carry a balance”. It’s obvious that churning is negatively correlated with carrying a balance so that’s a bit of a red flag that whoever posted that is making stuff up. Also if these really are neural models it would be next to impossible to figure out exactly what the criteria are for shutdown, so the specific advice at the end is dubious.

Having said that I’ve been concerned myself about AmEx eventually concluding I’m not profitable enough — though I don’t do MS or self-referrals I have maximized the credits and I can’t justify using the Amex Platinum that often for organic spend because of the 1x on most purchases and the 5x travel doesn’t yet have trip protection and using it for hotels doesn’t get me loyalty points. For that reason alone I’m thinking of getting a Green card so I can justify doing more AmEx organic spend.

GundamWing01

agreed. regardless of the anonymous OP and its controversial contents, nobody should be surprised that companies like amex (who get exploited) are driven to significantly reduce risk and loss by developing an advanced AI neural-net processor, a learning computer, because this community has grown exponentially after the CSR n00b event back in 2016. its my belief that "judgement day" is coming in some form. in the past, the numbers where tolerable by amex. now, its at unacceptable levels.

i want to remind everyone that companies do indeed look at the Total Lifetime Profit Value of the customer. the details contained within the anonymous OP make a lot of sense. just take a look at the very old article about customer profit value posted by no other than amex themselves.
https://www.americanexpress.com/en-u...er-scott-allen

notquiteaff

I don’t think that is necessarily inconsistent. Two engineers are working on this project. A company like Amex probably has many unrelated data science and machine learning projects underway.

GUWonder

I remember when Amex was basically the only large financial intermediary to have a collection of people working Six Sigma projects for it. Back then there was some kind of connection with Arizona too when it came to the Six Sigma efforts being undertaken by Amex and with rather senior people for Amex in NYC too. That reminds me of something to look into.

Gig103

Churners bug me because they end up forcing airlines to higher redemption rates. But why does Amex hate them? Interchange fees pay for the rewards with money leftover, don't they? Does Amex even pay the airlines a penny per mile transferred?