#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <stdbool.h>

typedef int STATUS;
#define ERROR -1
#define OKAY 0

struct version

 unsigned char major;
 unsigned char minor;
 unsigned char build;
 unsigned char patch;
;
STATUS is_less_than(struct version * original, struct version *compared, bool *result)
 compared == NULL 

Is there a cleaner way to do this?

Yes, there is a cleaner way:

if (a.major != b.major) 
 *result = a.major < b.major;
 else if (a.minor != b.minor) 
 *result = a.minor < b.minor;
 else if (a.patch != b.patch) 
 *result = a.patch < b.patch;
 else 
 *result = a.build < b.build;

return OKAY;

I reordered patch to come before build since that's how it is usually done. If your version scheme is different from this, good luck.

Instead of unsigned char I would choose unsigned int so that your code can handle versions like 1.0.20190415.

Return status

You create this:

typedef int STATUS;
#define ERROR -1
#define OKAY 0

which is basically a boolean status. Personally, I'd return a straight bool.

Bug/Not what you mean

Doing a

result = NULL;

is changing the local variable (parameter) result. It's not setting the result to NULL. In fact the caller won't probably have a pointer at all, but just a bool, which cannot properly be NULL.

Shorter version

I'm not sure this is cleaner, but here I go:

bool is_less_than(struct version * original, struct version *compared, bool *result)

Next time, add a driver/test suite to your question, to ease the life of people answering. This can be one:

int main(void) 

 struct version ref = 1, 2, 21, 8 ;
 struct version lower1 = 0, 2, 21, 8 ;
 struct version lower2 = 1, 1, 21, 8 ;
 struct version lower3 = 1, 2, 20, 8 ;
 struct version lower4 = 1, 2, 21, 7 ;
 struct version equal = 1, 2, 21, 8 ;
 struct version higher1 = 2, 2, 21, 8 ;
 struct version higher2 = 1, 3, 21, 8 ;
 struct version higher3 = 1, 2, 22, 8 ;
 struct version higher4 = 1, 2, 21, 9 ;

#define TEST(a,b,expect1,expect2)
 do 
 bool result1, result2;
 is_less_than((a), (b), &result1);
 is_less_than((b), (a), &result2);
 puts(result1==(expect1) && result2==(expect2)?"ok":"failed");
 while(0)
#define TESTL(a,b) TEST(a,b,true,false)
#define TESTE(a,b) TEST(a,b,false,false)
#define TESTH(a,b) TEST(a,b,false,true)

 TESTL(&lower1, &ref);
 TESTL(&lower2, &ref);
 TESTL(&lower3, &ref);
 TESTL(&lower4, &ref);
 TESTE(&equal, &ref);
 TESTH(&higher1, &ref);
 TESTH(&higher2, &ref);
 TESTH(&higher3, &ref);
 TESTH(&higher4, &ref);

 return 0;

I don't see any advantage to having the function to take three pointers (two for input and one for output) and return a status code. As a result of that unnecessarily error-prone design, the function has to handle the possibility of null pointers, and the caller is expected to handle a status code. But why should such a simple comparison have these failure modes at all?

The danger is further complicated by the fact that neither of the in-parameters is declared const.

Just pass the two versions by value, and you would eliminate all of that complication! On any modern 32-bit or 64-bit processor, passing a four-byte struct by value should actually be more efficient than passing it by reference — especially since you don't have to dereference the pointers to access each field.

With all of the potential errors out of the way, taking @RolandIllig's suggestion, you could then reduce it down to one chained conditional expression:

bool is_less_than(struct version a, struct version b) 
 return a.major != b.major ? a.major < b.major :
 a.minor != b.minor ? a.minor < b.minor :
 a.patch != b.patch ? a.patch < b.patch :
 a.build < b.build;

I'd go further and recommend using unsigned short instead of unsigned char for the fields. Using unsigned char for numeric values is awkward, since you would have to cast them when using printf(). On a 64-bit architecture, a struct with four 2-byte fields would occupy 64 bits, so you wouldn't be saving anything by using unsigned char instead of unsigned short.