Executive

Perspectives

Unlocking Potential from AI and

GenAI
Finance
October 2024
Introduction In this BCG
Executive Perspective,
AI a key topic top for CFOs today. After working with numerous clients
in the past year, we are sharing our most recent learning in a new series we articulate our
designed to help CFOs navigate AI. With AI at an inflection point, the focus vision for the future
in 2024 is on turning AI’s potential into real effectiveness and efficiency gains. of finance with AI
In this edition, we discuss the future of finance and the role AI will play
in unlocking the finance function’s full potential. We address key questions on
the minds of finance leaders:
• Where are the opportunities for AI in finance?
• How will the finance processes evolve as a result?

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

• What impact does AI have on the finance operating model and talent?
• How do I get started… and how do I get this right?

This document is a guide for CFOs and finance leaders to cut through
the hype around AI in finance and understand what creates value
now and in the future.

1
Executive summary | Unlocking impact from AI in the finance function

While finance functions have taken a bit longer to get started, CFOs are now starting to
The time to act explore GenAI and how it can help finance become proactive "value drivers" for business
on AI is now
~15% of companies1 are actively piloting or implementing GenAI in the finance function

AI is a key unlock for finance to deliver new capabilities (e.g., AI-based forecasting) and
higher-quality business support (e.g., GenAI-powered FP&A insight generation)
AI unlocks many
sources of value AI delivers both efficiency and effectiveness benefits for the finance function, e.g.:
• Efficiency: 20-40% capacity unlock, enabling more focus on business partnerships

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

• Effectiveness: 50%+ increase in forecast accuracy, 2x faster insight for decision making

CFOs start their journey by building high-value use cases and demonstrating early impact

Executing Yet to fully unlock value from AI, finance functions should move beyond use case exploration to
successfully reinventing end-to-end (E2E) processes, by:
• Adapting the operating model, talent, and ways of working to support new processes
• Modernizing the tech stack and rewiring data hierarchies to enable further value gains
1: BCG CFO Excellence Panel survey 2024, N = 204 2
Next-gen finance functions break the compromise between efficiency and
effectiveness, to become proactive value drivers for the business

Next-gen finance functions actively …by delivering on BOTH

integrate strategy and drive value… effectiveness and cost

“Excessive” “Truly
excellent”
Increasing contribution to
company’s strategic
decision making Value
driver

Effectiveness
Strategic Integrates
corporate and
partner financial

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Evaluates
Business performance strategy;
partners with
partner and partners re: BUs in
Platform strategy with
Supports and BUs; actively optimization
manager reports BU drives planning & manages
Traditional Shares best
operations; and resource
value driver
oversees target performance
function practices and setting
allocation
maximizes BU “Bureaucratic” “Minimalistic”
Serves as synergies
passive
“bookkeeper" Efficiency
Finance functions Top performers
Facilitator Partner Initiator
Source: BCG CFO Excellence Panel, N = 767 3
Inspiration | What next-gen finance functions look like

Common starting points Digitally mature functions AI-enabled functions

Planning, Fragmented data, heavy Excel use, Integrated cross-functional

Predictive AI models for planning,
budgeting, & manually intensive and siloed planning, modernized processes &
forecasting, and scenario analysis
forecasting budgets / forecasts tools with single-source data

Reporting Static tools / reports with limited Automated dashboards with Self-serve dashboards with GenAI-
and business visualization, built on manual standardized metrics, using based commentary, supported by
intelligence processing of fragmented data harmonized data & definitions conversational querying and search

Fragmented ERP; Excel-heavy, Lean GL, automated journal entries, Predictive close, AI reconciliations,
General
time-consuming reconciliations; system-triggered controls & GenAI entry recommendations, AI

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

accounting
manual statutory reporting exception monitoring pattern / error recognition

Lack of end-to-end systems, leading Integrated sub-ledgers, automated AI-based data extraction (e.g., OCR)
Finance
operations to off-tool calculations, tracking / invoice generation / processing, & validation, predictive collections,
monitoring via Excel auto-matching and data validation supplier analytics, AI controls

Integrated solutions for transaction AI-based cash forecasting, FX

Expert Reliance on Excel & manual inputs
functions processing and global monitoring, hedging models, working capital
limits visibility and future planning
with standard insights / alerts optimization, risk monitoring
4
A combination of levers is required to achieve finance excellence, and AI is
an important lever in the digital modernization journey
Levers to achieve finance excellence:

Digital modernization AI tools bring cognitive and predictive capabilities to augment

including AI existing tech solutions (e.g., ERP, specialized tools, RPA)

End-to-end process

Potential for value creation

reimagination
• Reinvent the finance offering with Generative AI

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

new AI-powered capabilities Artificial
• Adopt AI across E2E processes to intelligence
deliver exponential value Robotic and machine
process learning
Best-of-breed automation
applications
Legacy ERP / (e.g., treasury)
New operating model accounting
systems
and talent profiles
• Changes in roles, accountabilities, and
skills/expectations driving need for Recency of application
new capabilities and ways of working 5
 While GenAI is still nascent, companies are actively exploring its potential, in
addition to expanding the use of traditional AI

GenAI in finance
still nascent, in GenAI in finance 10% 64% 11% 7% 8%
exploration stage

FP&A AI/AA-based forecasting 4% 12% 23% 61%

Traditional AI
already prevalent

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

AI-based report generation 6% 21% 35% 38%
or being planned
across multiple
areas of finance AI-based account
6% 19% 47% 28%
Accounting

reconciliation

AI-based invoice recognition 4% 31% 42% 23%

Not explored Plan to implement Being implemented

Source: Select extracts from BCG CFO Excellence Panel survey 2024, N = 204 Exploring Pilot started 6
 Early estimates – potential to change as tech evolves

AI combined with other levers will be critical to unlock transformative

value across the finance excellence journey

Transformation of FP&A Transformation of Accounting & FinOps How AI accelerates

(~20-30% efficiency gains applying all levers) (~25-35% efficiency gains applying all levers) this transformation:

• Reduction of manual work,,

From To From To e.g., data extraction, rule /
Compliance tracking;
pattern-based validation,
Business data support calculation
system controls

Scenario models, Customer/supplier

Preparing plans, contracting and comms.
budgets, & forecasts market analysis, AI- Policy writing, workflow • Task reinvention through
based optimization improvement, etc. AI-created first drafts, smart
Close & consol.,

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

reconciliation and Predictive risk mgmt. reviews/alerts, analytics,
Data analysis and Self-service and statutory reporting and controls; AI pattern etc.
creation of static proactive insight / error recognition
management reports
Transaction tracking, GenAI-enabled • New finance capabilities
Stress-testing AI-built review and approval negotiation & analytics and insights enabled by AI
budgets & forecasts AI-enabled close,
recons, & statutory
Data collection, reporting
processing, & validation Insight generation from Manual transaction
dashboards; ad hoc processing Exception-based review
of AI-generated invoices, Capacity unlock for more
analytics
journal entries, etc. strategic advisory and
business partnering
Note: Estimated 3- to 5-year impact 7
Size = Illustrative share of workflow; pace of change will vary greatly by the starting point of your function
Use cases | Key opportunities exist for AI across finance processes

Planning, Business strategy: Analysis AI-drafted plans and budgets, GenAI-created variance AI-based forecast and
budgeting, & of market/demand, leveraging KPI driver trees and investigation and scenario modeling, AI
forecasting competitive landscape, etc. automated data feeds commentary generation refinement of driver trees

Reporting Standard dashboards/reports: GenAI- Ad hoc reporting: GenAI-based data Proactive monitoring (e.g., overspend,
and business drafted commentary, performance search and visualization, performance project delays) and optimization (e.g.,
intelligence analysis, creation of leadership decks analysis and insight generation working capital)

General Subledger close: AI-recommended Consolidation & filings: AI-based Compliance & policies: AI-based controls
journal entries, proactive/predictive close balance sheet reconciliations, preparation and error detection, GenAI policy and
accounting monitoring of statutory reports guidebook writing

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Finance Procure to pay: AI/OCR invoice processing and Order to cash: AI-based prediction models for credit Others
matching, payment terms monitoring, supplier scoring, early warnings for DSO/aging, predictive (payroll, fixed
operations spend/risk analytics + optimization suggestions collections with AI, GenAI-tailored customer comms. assets, etc.)

Expert Treasury: AI-based cash Tax: AI-based provisioning Investor relations: Earnings Risk management: Early
forecasting, FX hedging, calculation, proactive flags for call Q&A prep, investor warnings using pattern
functions balance sheet optimization deferred tax impacts sentiment analysis recognition; fraud detection

Cross-finance opportunities: Chatbots, guided workflows, co-pilot support, etc.

Analysis of opportunities based on additional value

High Medium Low
that AI can unlock (beyond other digital tools): 8
 Example 1

AI in action (I/III) | Using AI-based financial forecasting to drive impact

Context Solution overview Impact

Large manufacturing client AI-based demand forecasting engine leveraging
internal and external data:
Struggled to forecast market 50%+
accurately (17% increase in forecast • Trend-sensing engine to identify early shifts in market Improved
error)... sentiment, production technology, etc. forecasting
accuracy
..driving business challenges: • ML-based modeling to adjust for business trends such as
seasonality, competition, etc.
• Inability to anticipate and adjust to
market situation Driver tree models linking executive metrics 80%+
to operational variables (e.g., production units): Reduction

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

• Increased costs of labor & overtime, in forecast
transport, and increased inventory • More granular, data-derived forecasts for root cause variance
analysis and sensitivity assessment and bias
• Customer dissatisfaction from
decreased service levels and fewer • Ability to model various demand / production scenarios
on-time, in-full deliveries and enhance decision making
Forecasts
AI/ML center of excellence created, to enhance models and scenarios
and develop new AI offerings generated
rapidly

9
 Example 1

Deep dive | Structured and leading-edge AI modeling provides

step-change in operational driver forecasting capabilities

Illustrative
Operational driver for financial model

Monthly demand (in #)

Endogenous time-series forecast
Baseline Features: seasonality, long-term category trend,…

Machine learning time-series correction

Operational
Features: market data, competition, economy,

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

drivers COVID stringency,…

Planner override (micro / macro)

One-off
Features: extreme weather occurrences,
events regulatory changes, supply shortages,…

Impact: Variability better explained & anticipated, forecast error cut, bias reduced

10
 Example 1

Deep dive | Driver trees link operational drivers to financial results,

underpinned by AI algorithms
Illustrative

Level 2 metrics Operational drivers Data inputs

OTR shipping

Cost per mile Automatic inputs

Intra-network OTR supplies
Automatic data feed of actuals/
costs
forecasts directly from data systems
Miles
Miles per run

AI model outputs
Transportation Runs

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Outputs from relevant AI models
costs (e.g., demand forecasting model)
Units that serve as inputs into driver trees

Pallets Units per pallet

Outbound Manual inputs
transportation
costs Cost per pallet Assumptions and manual adjustments
(CPP) Historical CPP

Fuel price Δ
11
 Example 2

AI in action (II/III) | Using GenAI-based business intelligence to drive

impact

Context Solution overview Impact

Large US retailer struggles to get GenAI chatbot for conversational queries of data and
deep insights on performance, due generation of dynamic visualization:
to rapidly changing market / 2-4x
• Provides 2nd and 3rd order performance insight, with
demand environment, complexity of Faster report
suggested actions for improvement generation
product portfolio / categories, and
fluctuations across cost areas • Creates live charts that can be interrogated, adapted, and
exported into leadership decks
FP&A analyst faces several
challenges while investigating Driver tree engine enables FP&A analyst to build and
variances to advise business live-edit dynamic relationship models between key <1 day

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

partners on the drivers behind operational and financial variables: Turnaround
for ad hoc
performance: • Driver model enables deep investigation into root causes analytics
• Poor view of drivers’ under/ behind variances, to enable timely corrective action
overperformance
Scalable infrastructure that can be connected to any
• Significant time needed to analyze data source or system, providing visibility across BUs,
variances and identify root causes categories, and processes More
data--driven
• Slow turnaround for ad hoc data decision making
investigation requests

12
 Example 2
Deep dive | GenAI-based chatbot interface and driver tree relationships
allow for quick drilldown into root causes of performance

Dynamic charts generated

Conversation between analyst and tool throughout the conversation

FP&A analyst GenAI chatbot

What is driving the forecasted sales miss for [category A]
for November and December of 2024?

The forecasted sales miss is due primarily to a drop in sales velocity

(-25%), with a slight decrease in expected average price (-5%).

What are the drivers of the drop in sales velocity?

The forecasted sales velocity decline was primarily due to a large decrease
(-20%) in purchases per order and small decrease in order rate (-6%).

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

What were the drivers of decreased purchases per order?

Purchases per order forecast model attributes 50% of the decline to

anticipated increase in inflation (+1%), to predicted CDP reduction (-2%), and
the remaining 35% due to a drop in expected competitor prices for similar
systems ([CLIENT] 15% higher than the expected competitor average price).

What actions can we take to increase purchases per order?

The economic outlook cannot be adjusted by any levers. However, planning

for a steeper discount rate on [category A] heading into Q4 2024 will help
offset some of the expected losses and improve the outlook for CY 2024.
13
 Example 3

AI in action (III/III) | Using GenAI-based annual report creation to drive

impact

Context Solution overview Impact

Controllership team at a large GenAI-based creation of draft 10K/annual reports
company looking to reduce the time based on past filings and latest internal data
spent on generating investor reports 40-60%
• Tool auto-refreshes data, shifting human focus from Automation
by having Gen AI write the text- of statutory
data collection to review / refinement
heavy section: reports
GenAI-drafted MD&A commentary, using external
• Manual data aggregation and
data to synthesize market trends and implications
consolidation for 10K-style reporting
• Tool adapts comments to highlight key business
• Significant time spent on analyzing
conditions, e.g., demand, industry fluctuations, 2K hours
peers and market trends to write

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

economic landscape Saved in the
MD&A commentary controllership
Performance benchmarking across peers, based on organization
• More time spent by consolidation
publicly reported financial metrics, strategic
team on drafting reports and less
announcements, and other news
time on review / insights
• GenAI provides quick answers on competitive 40+
performance and peer outlook, replacing time- Benchmark
intensive manual analysis companies
compared via
GenAI tool

14
 Example 3
Deep dive | GenAI-enabled tool supports finance function
in generating the annual report and inquiring about peer companies

Annual report generation First-draft commentary generated by tool leveraging past

• Creation of draft of annual report annual reports, news articles, competitor publications, etc.
commentary based on prior reports
• Benchmarks and sentiment-based
text editing, including translation
and proofreading
• Conversational interactions to refine
and improve on GenAI-created drafts

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Peer reports and
press release inquiries
• Query peer reports and press releases
to refine commentary around market
sentiment, industry outlook, etc.
• Generate answers on peer financial
performance based on public data
15
AI implications: processes | AI enables finance team members to shift from
manual data tasks to strategic insights and business collaboration

Data collection and Performance Insights and Management Illustrative time

variance calculation analysis recommendations reporting allocation1

80%
FP&A analyst accesses Analyst looks into high Analyst writes Variance calculations
multiple systems to variance items by variance comments and comments are
pull and validate data reviewing source explaining findings and consolidated into
Current data, cross-referencing flagging key cost items PPT report and sent 20%
Excel used to calculate
process operational metrics, to be resolved and to leadership
Example of variance vs. budget and
past actuals and sending remaining budget
variance analysis Report Analysis
questions to business available
generation & insight
teams

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

70%
FP&A analyst uploads GenAI queries run to Analyst builds options GenAI used to build
prompt to GenAI tool, investigate driver trees to optimize cost and leadership report
to quickly collate data and dimensions (BU, simulates P&L with recommendation
30%
and run calculations region, GL item, etc.) impact of options with and P&L forecast
Future ML tool
process GenAI builds charts, GenAI provides Analyst reviews and
With GenAI
comments for review reasons for the higher- ML tool enables live socializes report with
Data Insight &
than-trend cost review and refinement stakeholders collection & decision
variances with business team calculation support
1Based on client experiences of typical breakdown of FP&A time spent 16
 • Shift of CFO role toward "chief performance officer"
AI implications: New mandates
driving strategic direction and decision making
• Increased focus on data stewardship, to build new
operating model | for the CFO insights and analytics for evolving business needs
AI drives changes organization
• Custodian of value, providing investment funding and
in roles, mandates, monitoring benefits realization from org-wide GenAI efforts

and ways of working

across finance • Evolution of finance service catalogue, with new
New offerings for the business (e.g., AI optimization engines)
engagement • Increased push toward AI-powered self-service,
models with driving leaner "finance business partner" teams
the business • Greater cross-functional collaboration, to fully leverage
internal financial, sales and operational data for insight

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

• New roles and profiles (e.g., solution architects)
to identify AI opportunities and build use cases
• Need for enhanced digital skill sets within finance,
Reinvention of
such as data analytics and AI capabilities
finance talent (e.g., ability to create scripts)
and skill sets
• Increased need for strong business acumen within
finance, resulting from new AI-based offerings and
reduction of transactional work

17
AI implications: technology | Three types of vendors
for AI in finance are suitable for different needs and use cases
Nonexhaustive

Enterprise tech solutions Point solutions Foundation builders

Augments their existing ERP / EPM Leverages AI to offer specific solutions Provides infrastructure and out-of-box
offerings with AI capabilities tailored for use cases, with focus on models to support a broad set of use cases
building new analytics (including finance)
Example offerings:
• GenAI invoice creation and AR Example offerings: Clients can leverage a mix of open- and
Approach management • AI / ML budgeting & forecasting closed-source models / engines to create
and use solutions tailored to their use case
• Controls / transaction review • GenAI management reporting & business
cases intelligence
• GenAI customer / vendor comms
• Fraud / risk detection • Predictive collections

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

ERP, Ariba, Tableau
Sample GPT
Concur AI Assist
vendors
Ability to customize
Standard, scalable offerings

Suitable to augment existing tools Suitable for companies looking for Suitable for companies looking to
with limited build effort off-the-shelf models with flexible build highly tailored models with
customization ability in-house resources

Source: Expert interviews 18

AI implications: technology | Target state tech landscape for finance will
significantly evolve, driven by AI requirements

Illustrative platform design incorporating AI Immediate priorities for AI execution

Smart business layer • Define stakeholder needs: AI use cases need to be defined
Interrogable dashboards/ Text generation based on current needs, vision for finance capabilities
Chatbot/copilot/ Conversation-based
knowledge search
models/interfaces for
decision support
(document creation,
emails, etc.)
code building • Create roadmap: Companies are increasingly building
quick AI pilots to prove effectiveness/efficiency impact

AI layer • Build models: AI tools can be developed today on top of

Model gardens for Machine learning / Knowledge graphs /
existing stack, prioritizing areas with higher data fidelity
LLMs/generative text predictive models relationship models
RAI guardrails • Set RAI guardrails: Standards are defined for data /
models, based on responsible AI frameworks

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Data layer
Data products • Augment using AI: While AI models are evolving, ML
Repository & storage Operational data services capabilities can be used to accelerate data clean-up
Distribution & integration
and improve quality / governance

Core transaction layer • Explore out-of-box capabilities: AI solutions are

Infrastructure and cloud increasingly being embedded into transactional
On-prem Cloud Hybrid TPU/GPU solutions (e.g., AI controls within ERP)

New layer Transformed layers 19

Getting started | 6 critical success factors for CFOs driving AI

Systematic
1 transformation
Use AI as a catalyst to accelerate end-to-end finance transformation, including processes and
operating model

Value-focused
2 build
Act as the value guardian, driving the highest-impact use cases and monitoring early benefits
realization

Technology
3 flexibility
Leverage off-the-shelf tools when possible and selectively build use cases
in-house when existing offerings do not fully address the requirement

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Data
4 foundations
Be the "chief data officer," continuously exploring opportunities to better leverage big data
for finance and business

Quality
5 governance
Establish safeguards against hallucinations and ensure reliability / security of results (e.g.,
human-in-the-loop review, GenAI testing and evaluation)

Leadership
6 buy-in
Get finance leaders and key business stakeholders onboard; drive change management/
culture toward supporting AI efforts

Source: Learning from BCG case experiences 20

 NAMR CFO EXCELLENCE (CFOx)
BCG experts | Michael Hardik
Key contacts James Tucker Jody Foldesy
Demyttenaere Sheth

for AI in finance Laurin

Aissa Matt Malavika
Henderson
transformation Menton Boudadi Harris Vishwanath

EMESA CFOx APAC CFOx

Alexander Marc Sebastian Anand

Roos Rodt Stange Veeraraghavan

Anne
Anna Hendrik

Copyright © 2024 by Boston Consult ing Group. All right s reserved.

Oberauer Ruellan
Schnelle
du Créhu

Andreas Norbert
Patrick Weber
Toth Wünsche

BCG X

Shervin Aaron
Khodabandeh Mike Beyer Nick Tanaka
Arnoldsen
21