Artificial Intelligence Still Needs Human Creativity for Guidance
In my book, Winning Marketing Strategies Using Generative AI. and in my blog posts, I caution that AI is a tool and must be reviewed by business strategists and leaders to insure the right objectives are met. AI is dependent on information that is available and may be biased by the prompts that is directing its actions.
To validate this, I presented AI with a problem that was faced by a grocery company for which I designed a cash control system. The company had retail stores throughout a region and wanted to build a milk plant. Unfortunately, interest rates were above 7% and would have made the project too expensive. The CEO thought there was cash in the system that could be freed from operations to fund the milk plant project without needing to borrow from financial institutions.
He had his CFO and team work on a process to free the cash from operations to fund growth projects. That team worked on a solution for quite a while but was unsuccessful in developing an acceptable solution. The problem was presented to me for a resolution.
Before I tell the solution that was successfully implemented, I want to show you the results AI gave.
Here is the prompt I gave AI:
Assume a retail business with 300 stores, each doing $1 million per year, Purchases are made at the stores in cash or check or credit card. About 30% of payments are in cash. Each day the store manager for each store orders cash from the bank for daily operations which include cashing checks buying certain inventory items etc. At the end of each day each store depoisits all checks cash and credit card receipts to the bank. The corporate management wants to use cash from the stores to build a milk plant because borrowing the money would be too expensive since interest rates are 7%.
Here is the AI response:
To explore how corporate management of this 300-store retail business can fund a milk plant using store-generated cash instead of borrowing at 7%, we need to analyze cash flow, cash availability, and working capital managementpractices. Here’s a structured approach:
1. Situation Overview
Total Annual Sales: 300 stores × $1M = $300 million/year
Daily Sales per Store (approx.): $1M ÷ 365 ≈ $2,740/day
Total Daily Sales (All Stores): 300 × $2,740 ≈ $822,000/day
Cash Sales: 30% of $822,000 ≈ $246,600/day in cash
Cash Handled Annually: $246,600/day × 365 ≈ $90 million/year
2. Current Cash Handling Practices
Store managers order cash from banks daily (incurring bank handling/transport fees).
At day’s end, all cash, checks, and credit card receipts are deposited.
This implies zero float — cash isn’t retained or pooled for corporate use.
3. Opportunity: Using Operating Cash More Efficiently
Key Insight:
You have $246,600/day in cash, circulating and being banked immediately, meaning zero interest-free use of this asset.
Options for Using Cash to Fund the Milk Plant:
Option A: Centralize Cash Management with a "Cash Concentration" Strategy
Steps:
Reduce daily cash orders to what is truly necessary.
Change banking protocols to allow delayed deposits (e.g., hold 1 day’s cash).
Pool idle store cash balances centrally for investment in the milk plant.
Create a corporate cash sweep account to pull excess cash into a central treasury.
Result:
Even a 1-day delay in depositing $246,600/day = $246,600 in usable float.
If you hold 2 days of float: ~$500K.
Across 300 stores, small holding thresholds (e.g., $1,000/store) = $300,000 in total.
Over time (e.g., a year), you could accumulate millions in internal funding.
Option B: Modify Daily Operations to Minimize Bank Fees and Maximize Cash Use
Stop unnecessary daily bank orders — assess actual vs. requested cash needs.
Implement cash recycling at stores (use received cash for operations).
Adopt "just-in-time" banking — deposit only when cash exceeds safe thresholds.
Example:
Assume you can retain $500/day per store safely and legally.
$500 × 300 = $150,000/day in retained cash.
Over 30 days: $4.5 million available.
This becomes your internal funding stream for capital projects.
Option C: Use a Corporate Cash Pooling Arrangement (Zero-Balance Account Structure)
Set up zero-balance accounts for each store.
Each day, surplus cash in store accounts sweeps into a central account.
Use that cash to fund short-term notes or directly pay for plant development.
Results:
Avoids external borrowing at 7%.
Preserves cash liquidity.
Keeps capital within the company.
4. Milk Plant Cost and Cash Requirement Estimate
Let’s assume the milk plant will cost $10 million.
If you can generate $150K–$250K/day in internal cash from better float, retained balances, and cash pool savings:
In 40–60 business days, you can fund the project without any debt.
Or fund it in phases, aligning construction spend with available cash.
5. Risks and Controls Needed
Security: More cash at store = more theft risk → need armored safes, monitoring.
Regulatory: Ensure compliance with local laws on cash handling and deposits.
Controls: Automated tracking of cash levels; managerial approvals for holding cash.
Cash Forecasting: Robust daily cash forecast model to manage funding timing.
6. Recommendation
Corporate should:
Conduct a cash flow audit of all 300 stores.
Design a cash optimization and pooling strategy.
Implement a rolling funding schedule for the milk plant using daily operational cash.
Consider a pilot in 10 stores, measure float potential, then scale.
This seems like a very good approach but I wondered about the definition of excessive cash, so I asked the following:
Cash holdings that exceed the amount needed for day-to-day operations, short-term obligations, and planned capital expenditures.
Detailed Breakdown
To operationalize this, excess cash is:
✅ NOT required for:
Paying suppliers, employees, taxes, or rent in the short term.
Maintaining minimum operating balances (e.g., daily till money).
Managing cash volatility (e.g., seasonal demand swings).
Planned capital spending or debt repayments.
✅ IS available for:
Investment in new projects (like your milk plant).
Debt reduction or interest expense avoidance.
Shareholder returns (dividends, buybacks).
Strategic reserves or acquisitions.
In Retail Store Context
For each store:
Operating cash might be $500–$1,000/day (for change, cashing checks, etc.).
Excess cash is anything above that — especially if it's being deposited and earning zero return.
Formula You Can Use
Why It Matters
Holding too little hurts operations (e.g., can't make change).
Holding too much hurts returns (opportunity cost vs. 7% debt).
Properly identifying and centralizing excess cash creates an internal funding source.
Why the Corporate Office Should Manage Cash Efficiently Across Stores
✅ 1. Centralized Visibility and Control
Corporate has access to consolidated cash flow data from all 300 stores.
This allows for standardized policies and aggregate decision-making.
Individual stores lack the financial sophistication or strategic view needed to manage excess cash efficiently.
✅ 2. Optimization of Idle Cash
Corporate can:
Implement cash pooling or sweep accounts.
Invest idle funds or use them to fund capital projects (like the milk plant).
Reduce or eliminate expensive borrowing (e.g., 7% interest rate).
✅ 3. Risk Management
Corporate can set:
Limits on how much cash can be held in-store.
Security and audit procedures to minimize theft or fraud.
Policies on deposit timing and vault cash requirements.
✅ 4. Working Capital Efficiency
Corporate can smooth cash flows across stores with different sales cycles, cash needs, or seasonal patterns.
Efficient management can reduce bank fees, improve liquidity, and enhance return on capital.
What the Corporate Office Should Do
Develop Cash Handling Policies
Set thresholds for “cash on hand.”
Determine when to deposit, how much to hold, and how to order cash.
Implement Treasury Management Tools
Daily cash sweeps to a central account.
Central forecasting of expected cash needs vs. available balances.
Monitor and Enforce Compliance
Use dashboards to flag deviations by stores.
Audit cash ordering, depositing, and handling practices.
Communicate the Strategic Purpose
Let store managers know that better cash use = company investment (like the milk plant).
This builds alignment and accountability.
Summary
Yes, the central corporate office must take responsibility for managing and using store-generated cash efficiently. It’s the only level with the tools, visibility, and authority to align daily cash operations with long-term strategic funding needs.
The system I developed was managed by the store managers because each store had different needs, different communities to serve and different operational challenges. So I asked AI about an alternative solution:
What about this as an alternative. Store managers are told to manage cash at their discretion but that if more cash is deposited at the end of the day than is withdrawn from the bank at the beginning of the day then the amount deposited over the amount withdrawn would be excessive cash, this cash would be considered a loan to the store that would affect the store's profit and the managers bonus.
Interestingly, AI agreed there was another approach possible and pointed out some caveats.. Here is its response:
That’s a clever decentralized incentive-based alternative, and it could work if designed and managed carefully. Let’s break it down:
Your Proposed Model (Decentralized Cash Accountability)
Rule:
Each store manager controls their own cash use. If the store deposits more at day’s end than it withdrew that morning, the “excess” is deemed unnecessary and becomes a “loan” from corporate. This “loan” is charged to the store’s P&L and affects the manager’s bonus.
Pros of This Model
✅ 1. Decentralized Ownership
Empowers store managers to think like business owners.
Encourages just-in-time cash management and reduced waste.
✅ 2. Performance-Based Discipline
Managers are motivated to avoid unnecessary bank orders and reduce idle cash.
Bonus impact makes the incentive real and measurable.
✅ 3. Identifies True Excess
The difference between deposited cash vs. bank-ordered cash is a simple, objective measure of unused cash.
Makes excess cash tangible and accountable.
But Also Consider the Challenges
⚠️ 1. Oversimplification Risk
This method assumes that all excess cash = inefficiency, but:
Some “excess” may be due to unexpectedly strong sales or conservative forecasting.
Penalizing this could discourage caution or even encourage underreporting.
⚠️ 2. Incentive Distortion
Managers may start under-ordering cash to avoid penalties — leading to:
Inability to make change.
Customer service issues.
Staff stress.
⚠️ 3. Security Concerns
If managers start holding more cash in-store (vs. depositing it), theft or loss risks increase unless proper safes and insurance are in place.
⚠️ 4. Fragmentation of Treasury Management
Without centralized oversight, you lose the ability to pool cash, negotiate better banking terms, or allocate working capital efficiently across the business.
How to Make It Work: Hybrid Model Suggestion
You can take the best of both worlds:
✅ Corporate sets a daily “Cash Efficiency Index” per store:
If deposits consistently exceed withdrawals by more than a target margin, it’s flagged.
Only persistent or extreme deviations result in a “loan” chargeback to the P&L.
✅ Establish Limits
Set a reasonable buffer range (e.g., ±$300/day).
Excess outside the range is reviewed or penalized.
Maintain manager bonuses tied to net operating performance including cash handling efficiency.
✅ Keep Central Oversight
Corporate treasury still receives data daily and manages cash pooling.
You preserve store-level responsibility without losing corporate control.
Conclusion
Your model is smart — it pushes accountability to where cash decisions are made. But to avoid operational and behavioral pitfalls, it works best with:
Clear thresholds,
Manager education,
Guardrails against underperformance,
And centralized monitoring.
I then gave further clarification that I had included in my consulting recommendation to see if it would be helpful.
What about setting a cash guideline set for each store based on an analysis of the past year's deposits. No penalty is charged regardless of the amount withdrawn if the amount deposited is within the guidelines. If a store's volume grows or shrinks the guidelines are adjusted based on a running average. Holding cash more than one day is frowned upon.
This was the AI response:
This is a mature, balanced, and operationally sound approach — combining data-driven guidance, flexibility, and accountability without heavy-handed penalties.
Your Proposed Model Summary
Set store-specific daily cash deposit guidelines based on historical data (e.g., last year’s average daily deposits).
Allow stores to withdraw as needed, with no penalty as long as daily deposits are within the expected guideline range.
Guidelines adjust automatically using a running average to reflect changes in sales volume (growth/shrinkage).
Holding cash overnight is discouraged to manage risk and liquidity.
Why This Approach Works Well
✅ 1. Data-Driven and Fair
Guidelines are based on each store’s own operating rhythm, not arbitrary limits.
Stores with higher cash needs aren't punished for being busy.
As business changes, the system self-corrects via the rolling average.
✅ 2. Encourages Natural Efficiency
Managers aren’t punished but are gently steered toward optimal behavior.
Outliers can be reviewed — not automatically penalized — keeping morale and trust intact.
✅ 3. Reinforces Financial Discipline Without Micromanagement
Managers understand there’s a “norm,” and exceeding it too often will invite scrutiny, but not automatic punishment.
That creates a culture of ownership, not fear.
✅ 4. Reduces Risk
The emphasis on not holding cash overnight reinforces good security and treasury practices.
Implementation Tips
🛠️ 1. Set Initial Guidelines Using Percentiles
Use the median or 75th percentile of last year’s daily deposit data to set a flexible but firm range.
Example:
If Store A averaged $2,400/day in deposits last year, set its guideline range at $2,200–$2,600.
🔄 2. Rolling Adjustment Formula
Update monthly or weekly using a 30–90 day moving average of daily sales and cash percentage.
This captures seasonality and growth without constant recalibration.
📊 3. Daily Monitoring Dashboard
Highlight:
Stores depositing significantly above or below guidelines.
Stores that consistently withdraw more than they deposit.
Only follow up with human review or coaching, not automatic penalties.
✅ 4. Align with Bonus or Evaluation Metrics (Optional)
Rather than penalize, make “cash efficiency” one of several performance indicators (alongside shrinkage, sales, labor cost).
Bottom Line
This approach blends strategic treasury control with operational flexibility and store-level trust. It's a best-practice model for a chain of this size.
My cash control system was adopted and the milk plant was built. Unfortunately, this project took place many years ago and AI was not available. That cash control system is still in place today.
Obviously, the CEO's only objective was not to free cash for a milk plant. his objective was freeing the cash while allowing the stores and the company to continue growing and being successful. Having the central office manage one of the key resources for stores would have hampered store managers effectiveness in growing local business.
The point here is that AI did give actionable advice in the beginning but might not have been optimal. For optimal results, the individuals that will be impacted by AI driven decisions must be considered and the objective viewed at the macro level. AI can help give options at the macro and micro level.
In Clayton Christensen's book The Innovator's Dilemma, he describes the decision of two retailers entering the discount retailing business. One, Woolworth, decided to leverage existing distribution and management systems to launch Woolco discount stores. The other, Kresge, decided on a decentralized approach and launched K-Mart. Woolco failed over time and K-Mart thrived. I am not certain but I think that AI would have found suggestions that would have supported both approaches.
I have found that optimal solutions normally come from people who work with an operation and understand the problems. Often they have suggested solutions. I think AI is best utilized in business when individual creativity gets the ball rolling and AI is used as a tool to fully develop the solutions and point out the potential shortfalls.
Comments