How I Use AI Every Day Working in FP&A

Profit Playbook
12 followers
AI Generated·May contain errors
I've trained hundreds of finance professionals on how they can use AI and FP&A work faster
and add more value to their organizations. And in this video, I want to show you some
of the most practical examples of how I use AI on a daily basis for finance work.
From things like variance analysis to board prep and financial
modeling with the exact prompts you can copy.
I'm Christian Wartek. I run the FP&A program at Wharton
and spent years in senior finance roles at P&G, Unilever and Squarespace.
Today, I help finance teams move beyond spreadsheets as a corporate trainer.
Let's dive in.
My day usually starts with reviewing financial results.
Whether it's the first week of the month after close or just a regular Tuesday,
My first task is often variance analysis.
This is where I compare our actual financial results to what we forecasted.
And the goal is to answer three questions for each major variance.
What happened? Why did it happen?
And what should we do about it?
That last part, the so what, is what separates good analysis from great analysis.
But here's the challenge.
You need to drill down from those high-level numbers all the way to the root causes.
Let's say I see a $250,000 variance in professional fees expenses.
I can't just put that number on a slide and call it a day.
You need to understand whether it's driven by rate changes,
volume changes, scope changes or something else entirely.
Creating these driver trees manually takes time.
I'm usually under pressure to have answers ready for the variance analysis meeting.
So this is where AI comes in.
I use it to quickly map out a multi-level business driver tree.
It breaks down high-level drivers into more detailed sub-drivers.
And it keeps drilling until we reach root causes
that actually point to actions the company can take.
Here's the prompt I use.
I say, I need to create a business driver tree analysis for my company
to explain revenue variances or expense variances, whatever I'm looking at.
Ensure that it breaks down high-level drivers into multiple levels of more detailed drivers.
While you can't show all business drivers due to the complexity of the business,
ensure that you drill down multiple layers for a few of them
until you reach a root cause that points to actions the company can take.
For context, the objective is to use this analysis
to guide the monthly variance analysis process of the FP&A team.
Now, it's crucial here that if the AI doesn't know yet where you work
and what your company does, what industry you're in,
then you need to explain that at the beginning of the prompt.
And then what this gives you is a structured framework in minutes
that would normally take much longer to put together.
I then validate it against our actual data and customize it for our specific situation.
For example, I used this approach to trace a professional services variance
down to a rate increase from our auditor.
They had increased their hourly rate on us, and that led to a concrete recommendation,
which was that we should compare prices with other audit firms
to see if we can save some money with the next audit.
By the way, if you want frameworks like this that you can use in your day-to-day FP&A work,
I'm giving away my top 10 most popular one-pagers for free.
Just click the first link below to grab them.
Alright, let's keep going.
Now, most days include at least one conversation with a business partner.
Maybe it's the head of marketing to discuss their budget
or operations to talk about cost overruns
or product development about a new initiative they're planning.
This is where interpersonal skills become just as important as technical skills.
And here's the challenge with these conversations.
I want to come across as collaborative, not like I'm interrogating them.
I need to gather information that's actually going to lead to useful financial analysis.
And I have to ask the right questions without damaging the relationship.
But coming up with thoughtful questions that hit all of these marks takes preparation.
So before any business partnering meeting, I use AI to help me brainstorm.
It helps me think through angles I might not have considered on my own.
And it tailors the questions to the specific relationship dynamic.
Here's the prompt I use.
I say, I'm a finance manager at a software company.
I need to speak with my CMO and I would like to start my conversation with questions
on how they believe our company can improve ROI on marketing investments.
The objective of the conversation is to make our company more efficient with our marketing spend.
For each question, make sure to ask for data that can be used to run a financial analysis
and eventually make a concrete recommendation.
We want to ensure that our conversation leads to an effective presentation
that helps with brainstorming ideas to improve marketing ROI.
Return the top three to five questions you think will help me the most.
Note that I'm more junior than our CMO and I would like this conversation to be
as collaborative as possible and not to feel imposing.
You can customize all of that for your specific situation,
who you're meeting with, what the topic is, what the dynamic looks like.
The output gives me questions that position me as a partner, not as a police officer.
It helps me focus on gathering data that actually leads somewhere useful.
And it makes me walk into that meeting feeling prepared instead of just winging it.
At some point during the day, an unexpected request comes in.
This is what we call ad hoc analysis.
The unplanned stuff that breaks up your routine.
Maybe it's an email from a director saying,
we're thinking about launching a new product.
Can you help us figure out if it makes financial sense?
Or maybe it's a CFO asking why margins on a certain product
have been declining over the past quarter.
These kinds of requests usually mean I need to work with multiple teams.
Sales for revenue projections.
Operations for cost of goods sold.
Marketing for the launch budget.
Then I build a model that projects potential outcomes.
The model produces a number that helps decide whether it's a go or a no-go.
Maybe a return on investment figure
or the number of months it takes to pay back the initial investment.
Now, sometimes the ad hoc work isn't about building something new.
Sometimes it's about reviewing someone else's work.
A business partner submits the budget and I need to go through it.
I have to identify risks and opportunities in their assumptions.
And I also need to do this without damaging the relationship.
So I use AI to help me analyze budget submissions
and suggest probing questions.
It helps me approach the conversation from multiple angles.
And it makes sure I don't miss critical assumptions that need challenging.
Here's the prompt I use.
I say, I have received the attached budget submission from my business partner,
the director of the R&D department at my company.
Analyze and suggest questions so that I can ask my business partner
to challenge the assumptions and understand possible risks and opportunities.
Output 10 questions covering a wide range of angles.
Note that I'm an FP&A manager
and my goal is to evaluate the risk involved in the submission
while protecting the relationship with my business partner.
Now, one important note here.
Always check with your manager and your IT department
before you upload any confidential information to AI tools.
Every company has different policies around data privacy.
Some AI tools actually store your inputs for training purposes,
which could expose sensitive company data.
When in doubt, you can also anonymize your data,
replace company names, specific numbers,
and identifying details with placeholders.
You can still get valuable outputs without risking sensitive information.
Just make sure to run this approach by your manager as well to be on the safe side.
The day often ends with communication work.
Maybe I'm finalizing a weekly pacing report
or preparing slides for an upcoming leadership meeting
or getting ready for a board presentation.
And here's the trap most people fall into with reports.
The numbers might be accurate, but the report isn't engaging.
Or we just share the data and let people come to their own conclusions.
What adds value is analyzing the numbers and telling the story behind them.
Which metrics need immediate attention?
What are the proposed next steps?
That's what leadership actually wants to know.
So, I use AI to review my reports and suggest improvements.
It looks at both the commentary and the data visualization.
And it helps me make sure the story is clear for my specific audience.
Here's the prompt I use.
I say, help me improve the attached report by applying FP&A best practices.
I would put five ways I can improve the report,
both in terms of commentary and data visualization.
Include short explanations of why your recommendations make sense.
Specifically, I want to ensure that the story behind the numbers is clear
and that the report is tailored to an audience of senior leaders
across different functions.
Now, if you want to go deeper and actually master FP&A,
not just AI, but actually to get a repeatable system
you can use to get from close to insights to having a real impact,
I created the Insight FP&A Academy.
It's a complete program where I teach you the frameworks
specifically adapted for finance work.
You can check it out at insightfp&a.com slash academy.
The link is in the description.
All right, lastly, let's talk about presenting.
Now for presenting, there's another challenge.
Reading slides word for word gets people to disengage.
The best presenters share insights that aren't actually on the slides themselves.
But coming up with those additional talking points
takes time and preparation.
So I give AI my slides and ask it for additional insights I can share verbally.
These voiceovers make a big difference in keeping the audience engaged.
Here's the prompt I use.
I say, I have a financial presentation coming up.
I want to make sure I don't just read what's on the slide
because it wouldn't be very engaging for my audience of senior leaders.
Based on the draft slides I have uploaded,
recommend three ways I can add additional insights for each slide.
Your suggestions should serve a starting point for my analysis.
Consider context, action, result, also known as CAR.
Drill down and drill up.
Comparisons, reservations, and recommended next steps.
And for board meetings specifically, there's even more at stake.
Board members ask tough questions and there's limited time to prepare.
So I use AI to help me anticipate what they are likely to ask
based on the materials and their backgrounds.
Here's the prompt I use.
I type this into AI.
Brainstorm which questions I can expect from the members of our board
when I present the financials.
I've uploaded the draft slides for the upcoming meeting,
the biographies of each board member,
and the slides for the last three board meetings
along with the list of questions I was asked for each one.
Based on this, I'll put 15 likely questions
and include two to three bullets for each to help me start preparing for them.
I hope these prompts will help you when you need to do these tasks next time.
If you want to see how I think about one of the most important parts of FP&A,
which is choosing the right forecasting technique for the right situation,
then watch this video next where I rank every forecasting method from best to worst.
