Cracks in a media (thin film) while flowing the energy, creates situation of change of media or change in a characteristic impedance, in turn, situation of mismatch.

The crack also creates scinario of dispersal of energy (escape out through crack).

At the point of crack, the plane wave becomes spherical wave (acts like a secondary source)

These are the phenomena, leads to change the characteristic and reduction in energy (out of expected) results into net effect as decrease in ferromagnetism

For ferromagnetism, each crack reduces the exchange coupling and the neighboring parts across the crack would be coupled at most via the weaker dipolar interaction. If you have a whole network of cracks, then you may need to think of your thin films as a collection of finite-size thin pieces, with a corresponding decrease and smear in critical temperature, size of FM susceptibility etc. (One could probably model an extreme limit of such a network as a FM polycrystal.)

Additionally, cracks are likely to introduce considerable modification of elastic properties and thus would affect the FM properties via the modification of the magnetoelastic (magnetostrictive) coupling.

I completely agree with comments made by Amipara and Yonko. One shouldn' t forget ferromagnetism is a collective events, and any thing  that disturbs that connectivity  reduces the magnetic susceptibility.

of cause...but sometimes it can increase it, depending on the system and cracks

If the thin film surface is in unsealed condition (It may be requirement), moisture added to crack will also change the characteristics

The cumulative effect of crack infested thin film surface is to be noticed in the decrease of the susceptbility value. This is expected expected as discussed by Mr Yonko. 

If any micro-crack should occur on the stochastic rough  surfaces  -even by chance- any where at the surface of the thin film  that  micro-crack would be completely unstable under the compressive stress, which may arise due to misfit strain associated with the lattice mismatched at interface between  the film and substrate. The result would be the propagation of micro-crack through the film, and finally the catastrophic failure or the decimation of the film. Therefore situation would be more dramatic when one observed cracks on the surface then the reduction in susceptibility. 

But what do you exactly mean by "decrease of ferromagnetism"?  If you mean the remanent magnetization, then the answer is pretty obvious: the material with cracks will not be uniformly magnetized, so the magnetic moment of such a sample as a whole will be lower than that of perfect one.

On the other hand, cracks may be the place where chemical reactions may occur more easily, and the products of such reactions may be non-magnetic at all.

well let me elaborate a bit what exactly I have observed. I have made films on different substrates with In as-deposited condition the compressive strained film had higher magnetization than the tensile strained films. When annealed in oxygen the ferromagnetism was suppose to increase but  I observed a decrease in saturation magnetization of annealed compressive strained films. When the AFM of the two films is observed the compressive strained films seems to have develop too many cracks on annealing but no such cracks are observed in annealed tensile strained films. Though this cannot be the only reason but still I wanted to know whether cracks can play a crucial role in decreasing ferromagnetism of material.