Estimation question

Estimate the savings a mid-size company gets from Ramp's procurement agents.

Practice this question out loud. An AI interviewer asks it, follows up like a real interviewer would, and scores your answer. Type or speak.

What this question tests

Tests structured estimation of savings from an AI driven procurement automation feature for a defined company size.

Define the company: assume a mid size company with 500 employees and roughly 20 million dollars a year in vendor and procurement spend.
Identify the savings levers: reduced maverick spend outside approved vendors, better negotiated pricing through automated benchmarking, and faster approval cycles reducing delays and rush costs.
Estimate maverick spend reduction: assume 10 percent of spend today is maverick or off contract, and procurement agents cut that in half, recovering 2 percent of total spend, or 400,000 dollars.
Estimate pricing improvement: assume automated benchmarking against market rates saves an additional 3 percent on negotiated contracts, or 600,000 dollars on the remaining spend base.
Sum the levers: roughly 1 million dollars a year in combined savings, before accounting for Ramp's own fee, which would be netted out separately.
State clearly that spend base, maverick spend percentage, and savings rates are assumptions you would confirm against the company's real procurement data.

Breaks the estimate into distinct savings levers, maverick spend reduction and pricing improvement, rather than guessing one blended savings percentage.
Anchors each lever to a specific, labeled assumption, like 10 percent maverick spend, making the logic transparent and adjustable.
Nets out Ramp's own cost conceptually, recognizing gross savings and net savings to the customer are different numbers worth distinguishing.
Sanity checks the total against the company's overall spend base, showing the estimate as a reasonable single digit percentage of total procurement spend.

Giving one estimate without breaking it into the specific levers driving the savings.
Not accounting for Ramp's own fee when discussing net savings to the customer.
Presenting assumed figures like the 20 million dollar spend base as if they were confirmed facts.

How would this estimate change for a company with much lower existing maverick spend?
What would make the pricing improvement assumption too optimistic?
How would you validate this estimate with a real pilot before making the claim broadly?