4 Mistakes to Avoid with Your Farm Data

Yep.  I just went there.
For those of you who aren't familiar, this is a screenshot (obviously edited) from a scene in the show South Park. Yes, I know, it's often vulgar and offensive. Unfortunately, it's also probably the most intelligent cultural commentary on television.

In this particular episode called "Gnomes," one of the boys discovers that the Underpants Gnomes are stealing his underpants.  The rest of the boys get together to stake out the situation.  They follow the Underpants Gnomes back to their hideout and the Gnomes explain that "Stealing underpants is big business!"  This proves to be a timely revelation as the boys have to write a school report on business, so they decide to interview the Gnomes to see what they know about business.  As the Underpants Gnomes reveal their business plan (see the picture above), it becomes pretty clear that they have literally no idea what they're doing... which brings us to our application to agriculture...


Perhaps you've heard this hilariously sad quotation from Dan Ariely:

I cannot tell you how many times a new client has come to me with reams of paper filled with pretty maps full of "precision" data that tell me... (DRUMROLL PLEASE...)

basically nothing.  

Seriously.  I hate to be the bearer of bad news, guys, but I'm terribly afraid that Ariely's quote is a far too accurate description of what we do in commercial agriculture.  We're like the Underpants Gnomes in the picture.
Phase 1: Data

Phase 2: ?
Phase 3: Profit!

Do you see the problem here?  What are we doing with that data?  How do we take that data and use it to actually make money.  

I suppose that the easiest and most obvious answer is "grid sampling for variable rate broadcast fertilization."

And now I'm REALLY going to upset some of you.  

In case you didn't know this, Iowa State University researchers came forward 15 years ago and pointed out that grid sampling doesn't pay.  Don't believe me? Better make sure you're sitting down when you click on that link then.  Or this one, where they recommend sampling in management zones instead of grid sampling.

I mean, don't get me wrong.  I think that grid sampling has a place-- specifically for variable rate lime application.  But I think I've made it pretty clear in the past that broadcast commercial fertilizer is a joke.  And to me, both the practice, and the data that we use to arrive at that "solution" are flawed.  

Ok... now that that's all out of my system I can FINALLY get around to the actual point of this article (sorry that took so long).

In this article I'm going to talk about 4 Mistakes to Avoid with Your Farm Data (and yes, I only called it that because it's one of those stupid catchy article titles that piques your curiosity just enough to make you click the link, and BAM! suddenly you're this far into the article and you're thinking, "Well, hell.  Now I might as well finish it").

#1. Not having any.
Despite what I said above, the only thing worse than having almost unusable data, is having none at all.  How does this happen?  Again, I cannot fathom that this would apply to anyone actually taking the time to read this blog, but to me, it would mean that you haven't kept track of anything-- not what seed you planted, or what fertilizer trials you ran, or your soil types, or your yields, etc.  This is the guy that just dumps the seed in, plants, ignores the crop until harvest, harvests the crop, and then probably takes it and sells it for cash price out of the field.  If you're not keeping track of anything then I think it probably goes without saying that that's a problem.  

#2. Confounding variables.
Ok, so whereas I kind of think that none of you are guilty of Mistake Numero Uno, I also get the vibe that almost ALL of you are guilty of this one.  A confounding variable simply means that you didn't isolate the one single thing that you were trying to test and that some other factor may actually be responsible for the results you see (or don't see).  The most common mistake I see here is when a client comes to me and says, "Well, I tried starter, and seed treatment, and fungicide... and actually my check is a different number... but I feel like it was the fungicide that made the difference."  


That's usually about the time I try to hide this look -->
Seriously... I just want to politely remind you that while I did study Interpersonal Communication in school, I'm an agronomist not a psychologist.  In short, I don't care what you "feel" worked.  What does the science say?  

Oh and of course, you can play the, "Well, the other fields around had disease and mine didn't" card.  And you should if it's true.  But you also cannot forget that starter also often leads to a healthier plant that is also better equipped to fight disease.  Similarly, the right seed treatments help set the plant up for long-term success.  The only way to know for sure what's making the difference, is to test these things independent of one another... on the same hybrids (can't forget that little caveat).

The point is this guys: If you're going to take the time, spend the money, and put the effort in, and do science, then please don't half-ass it.  Forgive the language, but I really need to drive home this point.  If you have more than one variable in a test then you cannot know with 100% certainty what made the difference.  I mean this very sincerely.  If you're not going to do it correctly, then seriously, just don't waste your time doing it at all.  You didn't learn anything from the "tests" you ran anyway, so why mess around with it?

#3: Overgeneralizing.
Here's a question: when you avoid Mistakes #1 and #2 and you get really good data-- I mean like true side by sides that you've replicated across hybrids and locations-- and it's a weather year like 2016 (whatever that was like for you), what do those results tell you about the product performance in 2017?  

Unfortunately... not a lot.  Remember, that weather can be a confounding variable.  I think it's pretty obvious that no two years are going to be exactly the same.  And it also follows that what works in a droughty year probably won't work in a monsoon year, right?  So we have to be careful to identify trends without getting overly excited (or discouraged... see Mistake #4 below) about a given product's performance.  The thing that we have to do is identify the limiting factor each and every year and figure out how to best address it on a year by year, field by field, case by case basis.  

#4. Throwing the baby out with the bathwater.
First and foremost, if you're like me and inclined to veer off on a bunny trail, here's a link to the very disturbing origins of that phrase-- and it's not what you think.  Anyway, as the phrase is commonly used, "Throwing the baby out with the bathwater" means throwing out the good stuff along with the bad.  Sometimes when you run tests and trials you don't get the results that you were hoping for.  In fact, one of the things that I love and appreciate about 360 Yield Center (yes, one of the companies that I represent), is that in last year's Yield Book you'll find the following excerpt.  Pay close attention to the portion I outlined in red:




Yes, that's right.  They had a 73% success rate.  I know.  I know.  Some of you are thinking, "Why on earth would they broadcast that?!"  
Because it's honest.  It's the truth.  And if you understand the way the world actually works, you also know that those were incredible results.

So what does this have to do with throwing the baby out with the bathwater?  Well, imagine if you were one of the guys in the 27% that did not see a yield response...  Now what?  Well, you might be tempted to say, "I tried farming practice X (in this case, Y-Drop) once, and it didn't work!" and then you might be tempted to make a mistake that even seems like a logical next step-- never trying it again.  Just because something didn't work one year doesn't mean that it won't pay 7 years out of 10.  It's a REALLY bad practice to base a long-term product decision on the results that you see on one location in one year.  

Similarly, the opposite is true.  Just because something worked on one location in one year, that doesn't mean that you should go hog wild with it and put it on every acre the next year.  Don't get me wrong-- I've been very vocal in my stance that I don't think any company out there is trying to make a bad product.  But I do think that at some point there are quality differences.  You do often get what you pay for.  Some chemistries work better than others.  Some starters work better than others.  Some seed treatments work better than others.  They just do.  Yes, they may be more expensive per gallon but boys, unless you're making moonshine, you sell your corn by the bushel, not the gallon.  Therefore, the cost per gallon of anything isn't really relevant now is it?  The only number that matters is "What is your cost per bushel produced?"

To best answer that question we need data.  And we need good data.  And we need to interpret that good data. And we need to avoid these common mistakes when interpreting that data. 

So let's bring it home now... in real world, totally practical terms, what does all of this look like in a real on-farm scenario.  For me, it means aggregating your data.  And I don't think you necessarily need some fancy computer program or overpaid consultant to do it.  What I mean is simply sitting down and overlaying your yield maps on any other data you might have with your monitor and starting to ask some pretty basic questions:
1. What strip trials did you run?  
2. What was the performance of each hybrid that you selected?  Did they work?  Why or why not?  
3. If a certain number or fertilizer or biological or whatever didn't work this year, is there a chance that it was weather related?  
4. How "normal" were your weather patterns this year?
5. What are the possible confounding variables that could have contributed to the response (or lack of response) that you saw?
6. What was your limiting factor this growing season?
7. Based on what you've seen, what tests should you run next year?
Theoretically at least, the things that you tried that worked, you should do on a few more acres.  The ones that didn't still deserve another shot, but I might back down the number of acres I do those on.

Ok, so I realize that data analysis and interpretation isn't everyone's cup of tea.  Not everyone nerds out over numbers like I do.  I get it. However, I also know that at the end of the day, this can make you money if you'll work at it and avoid these 4 data mistakes.

If you'd like help going over your data or coming up with some new trials to boost your bottom line in 2017, give me a call: 641-919-5574 or check www.DynamiteAg.com You can upload files there for me to look at as well.  I've also got our Fall Discount programs available upon request.  Thanks for reading guys!

Comments