Your business is very important to us, just not important for us to do it right.
November 4, 2009 7:01 PM   Subscribe

Is the company I work for unusual or is most corporate data this messed up? Every month I see hundred and sometimes thousands of the same errors.

Anonymous because I have mentioned the company I worked for in previous answers. I work in customer service for a Fortune 500 company and I mainly clean up codes and things on accounts that are wrong. I've been there for several years but I am planning to leave. If I do, I want to know if I can expect the same thing in most other big companies. So, is this typical of your experience in larger companies, that clear-cut errors go on for years with no real attempt to prevent them?

FWIW These errors do not violate the law, but they do call into question most of the numbers my department reports, sometimes in our favor and sometimes to our detriment. Anyone speaking up about the errors is told, "Yeah we wish that other department would fix them, but they don't."
posted by anonymous to Work & Money (15 answers total) 5 users marked this as a favorite
My experience is, yes, there's a lot of garbage in corporate systems.
posted by dfriedman at 7:09 PM on November 4, 2009 [2 favorites]

Yes, the same thing is true of all corporations in some department or other, usually most of them. The public or those not-inside don't realize this because there are a number of filters between the results/products they get and the actual inner workings. I have consulted for 100's of firms, and no matter how cool, streamlined and organized the public face, there's always some ugly chaos in the back rooms somewhere, usually involving computer systems or QA.

That is, there's a saying about making sausage that applies here.
posted by rokusan at 7:28 PM on November 4, 2009 [1 favorite]

Frankly, I've never met data generated by non-programmers that wasn't completely fucked up. Even data generated by programmers is often messy.
posted by Netzapper at 7:29 PM on November 4, 2009

Yes. I once worked on a project with a large company who applied a "multiplier" to their marketing figures. The multiplier was set by state. When I asked how it was derived, it became clear they had guessed long, long ago and had just continued to use these figures, even though they were significantly impacting their reports and thus their strategies. They were privately held though, so no audit would catch it.

A friend who works as an auditor for a big four runs into absolutely insane numbers that no human could ever hope to understand because they're simply made up most of the time. For publicly traded companies, the auditors get to go in and try to fix everything. And those numbers are just on the accounting side, with money, where it counts. On other things? Good luck.

Basically, if no system is in place to enforce proper data entry, and no system is in place to further verify the quality of the data initially entered, it's a given that a good chunk of your data will be pretty much crap. Most people just hope it's consistent enough in its crappiness (which it rarely is) or ignore the harsh reality of crap.
posted by disillusioned at 9:21 PM on November 4, 2009

Yeah, when I worked for a foreclosure law-firm it was basically my job to make sure that the folks we were serving with foreclosure papers were the right people. You'd be surprised (or maybe not) how often we had the wrong address and occasionally the entirely wrong person. I would say about 10-20 times a WEEK I would find a borrower with an incorrect address, more on the order of 50-80 lienholders who weren't being properly notified. I told the department before us in the line EVERY TIME it happened...did that fix the problem? Of course not. It was frustrating, but I'm thinking this is a common occurrence in corporate america. Pass the buck, and try not to be the guy who actually gives a shit about doing the right thing. They rewarded my hard work with forced telecommuting and contracting work, and eventually cancelled my contract...the only guy at the company who actually cared enough to fix the mistakes...ah well. /rant.
posted by ThaBombShelterSmith at 9:37 PM on November 4, 2009

Yes, there's usually some mess lying around in most big corporations. I worked for one too. A big namebrand Indian s/w MNC you all would've heard of. I can't even begin to tell you the dance steps we had to rehearse just before any audit or quality certification inspections (CMM, 6 sigma, ISO..blahblahblah). We were literally asked to go and backdate check-in dates in VSS for someone's shit piece of code lying around which none of us in my dept had even touched with a 10 feet long network cable. Up until this point in my career, I thought it's just us Indians and our 'way' of working, and maybe the west is more systematic and organised in these matters. I went on to work for two big US based MNC firms both in US and India and they have their own way of creating mess. There, on the surface there it might seem like there are a lot of inhouse utilities and tools to track various resources but the real situation in no reasonable way can be said to be reflecting what's shown in the tools. A lot of friends have shared similar stories about various big and small companies.

A lot of times, this happens because the guys who are tasked with enforcing these rules - not that it is their fault - IT, Admin etc are not in touch with the day to day running of the show where things keep changing whereas the processes and figures become stale and are never changed or revised because nobody has time or inclination to do some housekeeping in which nobody is interested and it creates a very unique situation where both the parties feel handcuffed.

Sometimes I just think it's really fascinating that these high-falutin companies continue to remain and even become higher-falutin despite all these internal inefficiencies, fucked up data, infighting between departments, the constant struggle between diff. layers of management etc. etc. It's almost a parallel world from the usual of creating products and selling them for profit - which is what most people think they do. There's so much going in that I'm sure that if you round up ten random employees and them some basic questions about what their company actually does and how do they make these profits you are going to get quite different answers because nobody really knows - yet the show continues to run.
posted by forwebsites at 9:42 PM on November 4, 2009 [2 favorites]

You want errors? Try the public sector.
posted by fixedgear at 4:57 AM on November 5, 2009

It never ceases to amaze me how dysfunctional big companies are. You're going to see the same thing in your next job.

Solution: only work for companies with a maximum size of 10 to 12 people.
posted by DreamerFi at 5:07 AM on November 5, 2009

nthing the presence of major data fubars in major corporations. i just came across something a few weeks ago where the company i'm temping at is basically ripping off another company by overinvoicing about $500. per month. as much as i could tell, it has been going on for about 2 years.

what really is interesting is that once i brought it to the attention of my supervisors--they are allowing it to continue--because they don't want to deal with the hassle of getting it right.

dreamerfi's suggestion isn't good, either. more often then not it's even more messed up in smaller companies.
posted by lester at 6:32 AM on November 5, 2009

Whenever there is, or has been, a human touchpoint in a data flow, there are bound to be tons of errors. If databases are not normalized / put in 3NF, etc. there is going to be a lot of jacked up information. Whenever users starts going rogue and building Access databases and Excel 'databases', data is going to get messy.

At my previous employer, it was a known fact that our core business data was inaccurate. It was also a known fact that no one wanted to tackle the job of cleaning that data up. It is a public relations nightmare to go out to your clients and admit that you have no idea whether their key biographical and statutory information is correct in your system.
posted by jasondigitized at 6:53 AM on November 5, 2009

Policies change, data do not. Usage of codes change, data does not. Adherence to policy changes is only as good as enforcement and generally an analyst is going to segment out the cleanest sample they can. While you may see egregious errors, in aggregate they may represent a small percentage of error - even if you see multipe all day. Moreso, % of error-1 may have no impact on % of error-2. Consider the number of records in your system. Now consider the amount of verification that would need to be completed in order to ensure all records were accurate and up to date.

Data analysts mine data, try to screen out the junk, and (hopefully) advise policy changes based on trending. But, overall - a single mistake is inconsequential. Its when everyone makes a mistake or the numbers can't be made sense of that analysts drive policy change.

Quality, on the other hand, does try to understand how good the data is - although usually they don't actually know this. For any CS organization quality listens to a percentage of calls and reviews the records associated. They listen to enough to get the pulse of the department, and to be able to understand roughly what that is.

Now here's a tidbit for you the next time you are reviewed by quality: Say you have 100 people, each making 20 recordss a day. That means you have 2000 modified records a day, and assuming a 5 day work week 10,000 records a week, or roughly 40,000 records a month. Each representatitve has only entered 400 records that month. To understand the quality of the documentation/call in the department within the 95% confidence interval a quick calc for this would look like (40000/(1+40000*((1-0.95)^2)) resulting in 397 calls/records per month needed to be examined to know what the departmental quality is (within the 95% confidence interval). This means, the department reviews 4 random calls per person. - for a Qality Assurance team, that is a reasonable feat to get department level health. Now if the department scores individual performance on those 4 calls - they're doing it wrong... roughly it brings you to a 50% confidence interval for only listening to those 4 calls - or the flip of a coin as to whether a good call or a bad call was captured. (Note: your 'bad' may not be 'bad' just not representative of your individual quality. To score an individual within a 95% confidence interval they would need to listen to 200 calls per person - not a realistic feat.

Now, what I love is when Quality tries to do analytics. Say they listen to 20 calls of type X and they say process X takes this long on this call, then they apply that logic to every call for that product... which has, we'll say 15,000 records regarding process X... not to mention, they have only listened to 20 calls which were between 10-15 minutes (because they don't have the time or resources to do more), and not the 30-50 minute calls which are more prevalent (pareto this out biyatches!) for process X calls.... anyway... getting worked up here...

Well... I think you get a rough idea of how I bang my head against a wall for roughly 4 of my 8 hours of work every day... anyway... gotta go see the wall...
posted by Nanukthedog at 7:18 AM on November 5, 2009

dreamerfi's suggestion isn't good, either

It's worked for me. Of course, you have to be careful in selecting the company, there are indeed plenty of messed up small companies.
posted by DreamerFi at 8:53 AM on November 5, 2009

Yep. The data at my former mega-employer was all ahoo. All the time. There were many projects to try to clean it up, but those were either shelved or done in such a half-asses way that there was very little impact.
posted by bluejayway at 9:09 AM on November 5, 2009

Not always, but a lot of the time.

In my experience, the flatter the corporate structure, the less likely you'll have crap data, because there are less steps between that person and someone that can and will fire their ass. The deeper into corporate hierarchies you get, the less personal responsibility that's engendered; the more that's not my job you'll encounter.

Where I work, we are data Nazis. But that's because we actually give a damn, and are given the freedom to do things correctly. It's funny, but you need more permission to do things right than you do to do them poorly, generally because what's right is a constantly moving target, and only the most lithe ("agile") companies can keep up without getting bogged down in asking permission to fix shit. Most middle-managers would rather play it safe (No-you-can't) than do things right (Yes-you-can).
posted by Civil_Disobedient at 4:34 PM on November 5, 2009

I work for a company where accuracy is very important with our data and analysis, because we are involved in litigation and we know the other side will be looking over everything we do to try to find errors.

The only way we have results that are accurate is that every analysis and every data-entry must be done in duplicate (by different people) and then compared. And all of our numbers/facts need to be checked against multiple sources if possible.

Doing this means we often see how inaccurate company's records often are. Data sets you pay for are generally pretty accurate as long as you are careful about knowing what each value really means, but once you start looking at internal data for companies you start wading through all sorts of crap.
posted by vegetableagony at 9:44 PM on November 9, 2009

« Older Uncertainy is a long time friend.   |   Does anyone have a Toronto bank that they love? Newer »
This thread is closed to new comments.