It’s a ques­tion, usu­al­ly asked by an IT per­son, that For­rester Research’s Boris Evel­son says he gets often: Boris, what do I put on the CFO’s dashboard?

“The CFO knows he needs one because every­one has one,” Evel­son says of these con­ver­sa­tions, ​“but he doesn’t know what to put on it.” 

That peren­ni­al busi­ness-IT ques­tion came up when Evel­son and col­league Rob Karel recent­ly looked for trends in the BI orga­ni­za­tion­al struc­tures of 197 com­pa­nies. The sur­vey is part of a larg­er report set to be pub­lished in the next few weeks on best prac­tices for form­ing BI orga­ni­za­tion­al structures.

The sur­vey of ​“BI-savvy” IT pro­fes­sion­als came up with one con­sis­tent, and famil­iar, theme: The busi­ness doesn’t take enough own­er­ship of BI projects.

BI orga­ni­za­tion­al struc­tures tend to fol­low two extremes. They’re either com­plete­ly siloed, and as Evel­son puts it, the left arm doesn’t know what the right arm is doing. Or they are the oth­er extreme — com­plete­ly and total­ly cen­tral­ized, and projects move slow­ly and noth­ing ever gets done.

Obvi­ous­ly, nei­ther is the ide­al. The best BI orga­ni­za­tion­al struc­ture lies some­where in the mid­dle. It’s a cru­cial point: All of BI’s best prac­tices rest on devel­op­ing the right BI orga­ni­za­tion­al struc­ture, Evel­son says: The right struc­ture for agili­ty, flex­i­bil­i­ty and to be reactive.

And, pre­dictably, these are also orga­ni­za­tions where busi­ness takes prop­er own­er­ship of BI.

A char­ac­ter­is­tic of a good BI orga­ni­za­tion­al struc­ture is one that sep­a­rates data prepa­ra­tion and data usage. For­rester asked BI pro­fes­sion­als who took the sur­vey: ​“How close­ly cou­pled are your data prepa­ra­tion and data usage orga­ni­za­tion­al struc­tures?” A total of 35 per­cent answered that some are tight­ly cou­pled, oth­ers not; 24 per­cent said they are loose­ly inte­grat­ed and coor­di­nat­ed; and 22 per­cent said that they are one and the same.

Busi­ness peo­ple shouldn’t be spend­ing their time run­ning batch jobs or putting the data into a data ware­house, Evel­son says. IT needs to be in charge of those tasks.

But once the data is in one place, there is lit­tle rea­son why IT should be involved, he says. With all of the mod­ern BI tools, any busi­nessper­son who knows how to use Excel can define what they want on their report, query and dashboard.

Anoth­er char­ac­ter­is­tic of the best orga­ni­za­tion­al BI struc­ture is one that sep­a­rates the data needs of the front office and the back office. For exam­ple, accu­ra­cy is more impor­tant than speed to a CFO — a CFO can’t have incor­rect num­bers. But for the front office, it may be more impor­tant to be able to give a cus­tomer a time­ly, approx­i­mate answer. Know­ing how to dif­fer­en­ti­ate between the busi­ness require­ments for accu­ra­cy and risk tol­er­ance for laten­cy is cru­cial, Evel­son says.

“When IT starts feed­ing front office peo­ple the same approach as back office, that’s where the break­down occurs,” he says.

Oth­er inter­est­ing find­ings from Forrester’s sur­vey include:

• Cross-enter­prise data usage is the main dri­ver for BI sup­port cen­tral­iza­tion. When asked how they assigned respon­si­bil­i­ties of the centralized/​shared BI sup­port orga­ni­za­tion ver­sus indi­vid­ual busi­ness unit sup­port orga­ni­za­tions, 45 per­cent said it depends on cross-data usage, fol­lowed by how mis­sion crit­i­cal the sup­port­ing busi­ness process­es are, and how mis­sion crit­i­cal the data is.
• Too few orga­ni­za­tions take a quan­ti­ta­tive approach for BI mea­sure­ment. A total of 54 per­cent said they engaged in infor­mal, qual­i­ta­tive mea­sure­ments, while 33 per­cent said they didn’t have any real mea­sure­ments at all.
• The ratio of users to BI sup­port staff varies wide­ly. The largest per­cent­age (16 per­cent) answered that there were 50 – 99 users for one BI ana­lyst, archi­tect and devel­op­er. But answers var­ied wide­ly; rang­ing from few­er than 10 to 1, to more than 500 to 1.