Numbers not fit for Serious Business Use

Question

I run a small busieness with 8 users all using Macs (10.6.8) with 10.6.8 server.  Nothign dramatic, lots of small and medoum sized word procesing dicuments and spreadsheets.  We are long-time Mac users, I stayed with Appleworks for quite some time as we were used to it.  However over the last few years I have grown to like Numbers, and think it is quick and easy to produce very nice-looking output.
So long as you don't do anything complicated or business critical - I have found out the hard way that it just isn't safe for grown-up use.  The reason I say this is baed on my experience with a spreadsheet I have written to record all jobs quoted out and analyse our performance.   The basic data table is about 400 rows x 50 columns.  There are a few dozen analysis sheets many with lots of instances of some quiet complex formulae,  Numbers is creaking at the seams, it takes 30 seconds just to open the file.  But my big problem is, it is very easy to get results that are just plain WRONG.  After many hours of debugging I finally discovered the reason.  As this spreadsheet has been developed there have been quite a lot of formatting changes, notable between "automatic", "text" and "date", and between "automatic", "text", "number" and currency", and (I assume) that asa result of this Numbers doesn't necessarily interpret data as being the format that you think it is.  Here's an example:
The basic data recorded is jobs quoted out, including engineer, quote date, and amount.
One of the tables shows an analysis of value quoted per engineer per month, using SUMIFS.  This matches data on engineer initials and date.
The quote date is in a cell whsoe column is formatted as a date.
The quote amount is in a cell whose column is formatted as currency.
One engineer did a vewry big quote - and I noticed that thsi value was not showing up in the analysis table.
I selected each column in turn and set the format of the colium, this did not make any diference.
I cheecked (without changing anything) that the values for date and quote vaue were sensible -they were.
Then I retyped the data for date and quote value.  I retyped exactly the same values that were there before.
Bingo - now the analysis shows the correct amnounts.
The conclusion is clearly that that the carrency and date values weren't matching the SUMIF as expected.  This is undeniable.
It is a fact that retyping the data fixed it, one can speculate that the formatting of th cell had somehow become borked for some hiostorical reason, when new data was entered the format as forced to whatever the colium format was set to.  (we know from experience that even if you set the format for a column, individual cells in the colum can still have a different format if Numbers thinks this is what you intended)
The problem is that, whatever the reason, Numbers gave me the wrong numerical result.  The raw data was displayed perfectly Ok with nothign at all to warn me of a problem.  The error was silent.  Therefore, I cannot trust what Numbers is telling me.  I cannot think of any way to check whether or not this sort of error is occuring...it was jut chance that I noticed on this occasion.
In how many other placesis Numbers silently giving me the wrong answer?
What about hundreds of other Numbers spreadsheets that I have used for quoting jobs - how do I know that these are numerically accurate?
It is quite scary to have built up a big intellectual investment in Numbers, and then to discover that is is unreliable.
Apple, of course, do not respond to feedback from me requesting a contact.
Clearly no-one should be using Nukber for mission-critical applications.
What are the alternatives?   I have avoided Excel for many tyears, ever since I lost a few days' work when it corrupted a file for me!  but these days Excel is so well established and in common use that I guess it can't be that bad any more.....can it?  And I suppose that it can;t have the same sort of errors that i find in Numbers.....peopple woudl have noticed by now, wouldn;t they??
the real answer is that no one shoudl be using spreadsheets for busioness critical applciations - theyare just too dangerous, to difficult to get demonstrably right, too easy to break by accident.  When I can find time I am going to replace all the above with a Filemaker database. 
Any thoughts or experiences from anyone else?
Any chance of this post not getting deleted?

Answer 1

Hi Luke,
Chances of the post surviving or being deleted depend mostly on an interpretation of whether it fits the terms of use for Apple Support communities.
I cheecked (without changing anything) that the values for date and quote vaue were sensible -they were.
Then I retyped the data for date and quote value.  I retyped exactly the same values that were there before.
Bingo - now the analysis shows the correct amnounts."
I suspect that the previous change you mentioned may have altered the Date and Time value (the 'date value' you name) into something Numbers did not recognize as a Date and Time value when the column was reformatted to Date and Time, show date only. Retyping "exactly the same value as was there" displayed would remove any extraneous bits, permitting recognition as (the date part of) a Date and Time value, and allowing Numbers to recognize the row as one that should be included in the SUMIF. The case might be similar for the quote value in that row, although I think this would be less likely. If you changed both before checking the result of that change, there's little chance of discovering what caused the amount to be disregarded in the calculations.
"There are a few dozen analysis sheets many with lots of instances of some quiet complex formulae,  Numbers is creaking at the seams, it takes 30 seconds just to open the file."
Numbers is not speedy, particularly with quite complex formulas (and many of them). Part of the issue in Numbers '08 and '09 is that every change in a table triggers a recalculation of the whole table. That issue has been at least partially addressed in Numbers 3, where the recalculation appears to not extend to the full tableeach time there's a change, but to only th cells affected by the change.
"...Excel is so well established and in common use that I guess it can't be that bad any more.....can it?"
Excel has been "well established and in common use" for many years now. That alone, though, can't be relied upon as definitive evidence that "it can't be bad;" only that many copies of it have been sold and installed, many people use it, and files saved in a format it can open can be opened and read by many people. Whether it's "good" or "bad" (or should be described in more emphatic terms toward either end of this range) depends on the user, her purpose in using it, and how well that user feels it fits that purpose. (My personal take on it is that Excel does the jobs I've used it for well, but that I have used it infrequently enough that I have not learned where things (text size, data formatting, etc.) are found.
But I'm not you, and my needs are not your needs. The standard advice applies: Use the tool that fits the job. That's fairly common advice in these discussions.
Regards,
Barry