王峰 AI × 经营 约一次诊断

总目 / 文章 / WF-0418

文章 · 存档

业财一体最容易走偏的,恰恰是最努力的财务人

业财一体、业财融合几乎成了财务圈的集体信仰。系统要不要上、数据能不能通、BI要不要做、AI能不能接,成了每个财务负责人绕不开的问题。也正因为大家都很努力,踩坑反而踩得更深、更痛。

财务人其实是带着一种专业自信进入业财一体这件事的。我们懂核算、懂逻辑、懂规则,也懂系统,直觉上会认为只要把业务数据和财务数据打通,事情自然就顺了。现实很快给人泼一大盆冷水,系统买了、接口打了、数据通了,领导却傻眼了:然后呢?管理有什么进步吗?隐患发现了什么吗?问题排除了多少呢?风险被挖掘了几个呢?都没有?难道我花了这么多钱就是为了让财务减少工作量吗?

好不容易安抚住了领导,业务又跳了。系统依赖于业务数据的及时准确完整,于是让业务及时提交各种数据,业务说忙,没时间,催一下录一录,不催就过两天再说,催急了就说财务在折腾他们,还给领导打小报告。我去,领导以为我们工作量大幅度减少了,可以偷懒了。岂不知我们更忙了,加班次数更多了。

财务人总以为业财融合就是把业务数据和财务数据打通就可以了,但是当数据好不容易打通了又发现领导根本不买账。于是又开始做业务数据的各种各样的财务口径分析,发现分析完了业务也根本不买账。其实财务人都忽略了很重要的一点:业财融合合的是数据,融的是逻辑。逻辑在先数据在后。业务逻辑没打通之前就强制把业务数据财务化,那么即便是数据打通了也只是财务实现了自动记账而已,距离业财融合还远得很。

真正容易踩的第一个坑,就是把业财一体误以为是一个技术问题而不是一个认知问题。

从全球范围看,不管是顶级大厂还是麦肯锡这类咨询大佬,还是近几年AI管理领域最前沿的研究,一个共识越来越清晰:数据不是起点,甚至不是核心,逻辑才是。AI时代当然要数据,大量的数据,这既是起点又是基础,但其实底层的逻辑如果错了,那么结果就会谬之千里。AI最擅长的是放大逻辑而不是替你修补逻辑。如果业务逻辑本身是混乱的、模糊的、甚至彼此冲突的,那么系统只会把混乱放大得更快、更“漂亮”。

很多企业的业财项目本质上是在做一件看起来很高级但对经营几乎无感的事:把原本分散在各个系统里的数据,统一变成一套可以自动出凭证、自动出报表的流水。对财务来说这是效率革命;但对经营者来说这依然只是算账更快了而不是决策更清楚了。

第二个最容易踩的坑,是财务视角对业务的过早殖民。

当领导不买账业务不响应,财务人最自然的反应是什么?是继续加专业。开始做更复杂的口径调整、更精细的拆分分析、更漂亮的模型,甚至引入AI预测、算法分析,希望用高级感换取话语权。但问题恰恰出在这里。

业务并不天然排斥数据,它排斥的是被不理解的方式解释自己。如果一个业务动作在它自己的世界里,本来是为了抢时间、保交付、稳客户,却被财务直接映射成毛利下降、费用异常、效率偏低,那这种分析哪怕再准确也很难被真正接受。

国外一些走在AI前列的企业反而在早期刻意压制财务标签的输出。他们先用AI做的是还原业务决策当下的真实约束条件:交期压力、资源瓶颈、客户优先级、风险敞口,而不是直接生成财务评价。等业务逻辑被充分理解、被尊重之后,财务指标才被自然引入,成为共同语言而不是外来审判。

第三个隐藏得更深的坑,是把业财一体当成一次性项目而不是持续演化的过程。

很多企业在推进业财融合时都会不自觉地用项目制思维:立项、上线、验收、复盘。好像只要系统跑起来业财一体就完成了。但从科研和实践角度看,真正成熟的业财体系更像一个不断学习的系统而不是一个固定结构。

AI在这里提供了一个非常好的镜子。现在最先进的AI管理模型已经不再追求一次性建模准确,而是强调持续反馈、持续校正。它们假设业务逻辑本身就是会变的,市场在变、策略在变、人也在变。如果你的业财体系不能跟着变,那么再漂亮的数据,都会迅速过期。

所以,当你回头再看业财融合合的是数据融的是逻辑这句话,会发现它其实还有一层更深的含义:逻辑不是一次打通的,而是反复对齐的。不是财务把逻辑教给业务,也不是业务迁就财务,而是在真实经营中不断逼近同一个真相。

其实我很想替很多认真做业财的财务人说一句公道话:踩这些坑,并不说明你不专业,恰恰说明你太专业、太用力、也太着急了。真正成熟的业财一体反而往往开始得很慢,看起来也不炫。它先解决我们是不是在讨论同一件事,再解决这件事值不值得算,最后才是怎么算得更好。

如果一定要给一个不那么鸡汤、也不那么说教的结论,那可能是:业财一体不是让财务变成业务,而是让财务真正走进业务的世界,同时又不丢掉自己的判断力。这条路走对了,你就会发现数据突然开始自己说话;走偏了,再多AI再多系统也只是在加速原地踏步。

Business-Finance Integration Are Often Go Off Track

The People Who Work the Hardest in Business-Finance Integration Are Often the Ones Who Go Off Track Most Easily

Business-finance integration has almost become a shared belief in the finance world.

Whether to buy new systems, whether data can be connected, whether to build BI, whether to add AI—these have become questions no finance leader can avoid.

And because everyone works so hard, the mistakes they fall into often become deeper and more painful.

Finance people actually enter business-finance integration with strong professional confidence.

We understand accounting, logic, rules, and systems, so we naturally believe that once business data and finance data are connected, everything will work smoothly.

Reality quickly pours a big bucket of cold water on us.

The system is bought, the connections are built, and the data is connected, but leaders are confused:

So what now?

What management progress did we make?

What hidden problems did we find?

How many issues did we solve?

How many risks did we uncover?

None?

Did I spend all this money just to reduce finance workload?

After finally calming the leaders down, the business teams get upset.

The system depends on business data being timely, accurate, and complete, so finance asks business teams to submit all kinds of data on time.

Business says they are busy and have no time.

If you push them, they enter some data.

If you don't push, they say they will do it in a few days.

If you push too hard, they say finance is bothering them and even complain to the leaders.

Wow—leaders then think our workload has dropped a lot and that we can slack off.

They don't know that we are actually busier than before and working overtime even more.

Finance people often think business-finance integration just means connecting business data and finance data.

But when the data is finally connected, they find that leaders still don't care.

So they start doing all kinds of finance-style analysis on business data.

After all that analysis, they find that business teams still don't care.

In fact, finance people ignore one very important point:

Business-finance integration connects data, but it integrates logic.

Logic comes first, data comes later.

If business logic is not clear, and business data is forced to become “financial” too early,

then even if data is connected, it only means finance has automatic bookkeeping.

It is still very far from real business-finance integration.

The first and easiest mistake is thinking business-finance integration is a technical problem, not a thinking problem.

Around the world, whether it is top tech companies, big consulting firms like McKinsey, or the most advanced AI management research in recent years, one idea is becoming very clear:

Data is not the starting point, and it is not the core—logic is.

In the AI age, of course we need data, and lots of data.

Data is both the starting point and the base.

But if the logic underneath is wrong, the result will be completely wrong.

AI is best at making logic bigger, not fixing broken logic for you.

If business logic is messy, unclear, or even conflicting, the system will only make the mess grow faster and look more “beautiful.”

Many companies' business-finance projects are really doing something that looks advanced but has little effect on management:

they take data spread across many systems and turn it into one flow that can automatically make entries and reports.

For finance, this is an efficiency upgrade.

But for managers, it only means faster accounting, not clearer decisions.

The second easy mistake is using the finance view to take over business too early.

When leaders don't support it and business teams don't respond, what is the most natural reaction for finance people?

They add more “professional” work.

They do more complex adjustments, more detailed breakdowns, nicer models, and even add AI forecasts and algorithms, hoping advanced tools will bring more influence.

But the problem is exactly here.

Business does not naturally hate data.

What it hates is being explained in a way that does not understand it.

If a business action is meant to save time, ensure delivery, and keep customers steady,

but finance directly turns it into “lower margin,” “abnormal costs,” or “low efficiency,”

then even if the analysis is correct, it is hard for business to accept it.

Some companies leading in AI actually hold back financial labels at the early stage.

They first use AI to restore real limits at the moment of business decisions:

delivery pressure, resource limits, customer priority, and risk exposure,

instead of directly giving financial judgments.

Only after business logic is fully understood and respected do financial indicators come in naturally,

becoming a shared language instead of an outside judgment.

The third, deeper hidden mistake is treating business-finance integration as a one-time project instead of a process that keeps changing.

Many companies push business-finance integration with project thinking:

approval, launch, acceptance, and review.

It feels like once the system runs, integration is done.

But from research and real practice, a mature business-finance system is more like a learning system, not a fixed structure.

AI gives us a very good mirror here.

The most advanced AI management models today no longer chase one-time perfect results.

They focus on ongoing feedback and ongoing correction.

They assume business logic will change—markets change, strategies change, and people change.

If your business-finance system cannot change with them, even the best-looking data will become old very fast.

So when you look again at the idea that business-finance integration connects data and integrates logic,

you will see a deeper meaning:

Logic is not connected once—it is aligned again and again.

It is not finance teaching logic to business,

and it is not business giving in to finance,

but both sides getting closer to the same truth through real operations.

I really want to say something fair for many finance people who take business-finance work seriously:

falling into these mistakes does not mean you are unprofessional.

It often means you are too professional, trying too hard, and moving too fast.

Truly mature business-finance integration usually starts very slowly and does not look flashy.

First, it solves whether we are talking about the same thing.

Then it solves whether the thing is worth calculating.

Only in the end does it focus on how to calculate it better.

If we must give a conclusion that is not too inspirational and not too preachy, it might be this:

Business-finance integration is not about turning finance into business.

It is about letting finance truly enter the business world, while still keeping its own judgment.

If you walk this path the right way, you will see data start to speak on its own.

If you go the wrong way, no matter how much AI or how many systems you add,

you are just speeding up while standing still.