Posted by Andrew on 02/23/06 01:37

"Erich93063" <erich93063@gmail.com> wrote in message
news:1140647157.967448.102260@g47g2000cwa.googlegroups.com...
> Ok so I'm gonna attempt to explain my problem as easy as I can. I have
> a database of contacts (they are actually vendors). I am writing a
> procedure that will query the vendors and generate a tab delimited file
> that will be used an accounting program (MYOB). In the vendors database
> I am exporting from, I have a field for "First Name", a field for "Last
> Name" and a field (looked up by ID) for "Company Name". Now here is the
> problem. In the usual case there will be a company name, a first name
> and a last name for example:
>
> Company Name: ACME Corporation
> First Name: John
> Last Name: Smith
>
> HOWEVER, sometimes the vendor is not part of a company and the users of
> the database have just been entering in the vendors name as the name of
> the company. So it would end up looking like this:
>
> Company Name: John Smith
> First Name: John
> Last Name: Smith
>

Unless I've misundestood what you're describing - the thing you absolutely
have to address is the data issues. It's a hoary old truism - but, GIGO.
In situations like this - I have always found that you will get a
significant number of entries which are duplicates. Eg - If you have John
Smith, Dave Jones and Min Patel from the same company - you may well have at
least 3 entries. If you have John Smith from several - perhaps competing
companies, or as individuals, you have a serious mess.
If you are trying to report on this data - you're feeding your decision
makers crap.
In the first instance - you have to try and get high level buy in to put
significant pressure on those entering the data to clean it and keep it
clean. This may involve a change to the interface at the front end - for
example, to force the user to enter a postcode/zip code for any new entry,
then next option is to choose from an entry already in the database - and
then - if it's not there - to allow them to select an option from a
commercial address lookup such as the PAF software available in the UK -
where you have to do a search linked to zip/post code.

All of this - obviously - has recource issues. You have to make it clear
that no matter how clever you try to be - you can't address shite data with
increasingly complex workrounds. A chaotic contacts database loses the
company money in a load of ways. If you claim that you can guesstimate what
people "should" have put in - the first time you fail on a juicy contract -
you're in the crap.

Your only professional and sensible option is to go to the people who are
living by this information to feed their families - get buy in and get the
people putting crap in to put it right, get agreement to implement a
strategy for ensuring that the data is useful, getting agreement that this
is stuck to.

Contacts are silver - leads are gold - you have to make sure they're usable
and tracable. Getting in touch with the wrong Mr Smith or Mr Singh is not
acceptable. Therefore, trying to address the issue by any other means than
a data cleaning exercise involving those putting in crap is not a good
option. Thwe ymay be putting in rubbish - but I bet they know who 90% + of
the people are. End users think computers can make heuristic decisions
based on their phone contacts with the customer - ot that yo can program
them. If you persist this myth - it'll get worse.

I'll be honest - an attempt at a workaround is easier to sell - but in this
case you really have to take the hard road and sell other people who know
who is who putting it right and knowing they have to keep it right. It's
not going to make you popular and by the time your company is making money
on it - you'll be forgotten. You'll be entitled to sit back and feel smug
though (;->

Summary;
There is *no* work around for unusable data. The better the data quality -
the more money your company makes. Mistakes on your on your contacts
database can be "very" embarassing and lose money - if you sell a
workaround, *you* are in the firing line. Get the data right. This helps
people using the data make money. The more they make - the more chance of
keeping your job - or getting the pay rise you deserve. This is a
management issue - *not* a technical one.

IMHO

All the best
Andrew

[Back to original message]