Adaeze is the head of talent acquisition at a 4,000-person technology company that hires roughly 300 people a year, and she has a compelling AI opportunity: a resume-screening tool that will save recruiting about 40 hours a week. In her first telling, that was the business case, the word "efficiency" and a single number. Then she rehearsed it. Her CFO would ask about quality impact and she would not know. Her Chief Diversity Officer would ask about fairness implications and she would not have analyzed them. Her CEO would ask about risk exposure if the tool introduced bias and she would have no answer. And nobody in the room had yet asked the question she cared about most, which is what an algorithmic screen does to what candidates say about her company. A one-dimensional case does not survive contact with an executive table. This lesson builds the five-dimensional case Adaeze needs, quantifying efficiency, quality, fairness, brand, and risk into a single defensible proposal.

Dimension One: Efficiency Value

Start with what you know, but be precise. Adaeze's baseline: screening 200 resumes per role at 3 minutes each is 10 hours per role, and with 15 open roles at any time that is 150 hours a week, or 3.75 full-time equivalents at 40 hours each. She discounts the vendor's claimed 70 percent time reduction by 20 percent to a planning figure of 56 percent, but models conservatively at a 50 percent saving: 75 hours freed per week. At a fully loaded $60 per hour, that is $3,900 a week, or about $202,800 a year. Establishing the baseline first is what makes the rest credible, because a saving is only meaningful relative to a number someone can check.

The critical move is what she does with the freed time. Eliminating 3.75 people is a cost-cut case that carries severance costs and loses expertise the function cannot quickly rebuild; instead she frames it as redeployment. The tool frees 75 hours a week that recruiting reinvests in relationship-building with hiring managers, in candidate experience, and in interviewing and assessment, all of which are higher-value than mechanical resume review. That framing is more honest and more defensible than pure savings, and it also pays forward into the other four dimensions. Recruiting speed improves because fewer bottlenecks sit in the funnel, quality improves because more attention goes into assessment, and candidate experience improves because communication becomes more personalized. Those are not separate benefits bolted on; they are what the redeployed hours actually buy.

Dimension Two: Quality Value

Quality means better hiring decisions. It is harder to quantify than efficiency and typically worth more, and the first question is uncomfortable: do you track it at all? Adaeze asks whether her function has data on retention by hire, on performance ratings by hire, and on time-to-productivity, and like many organizations she finds the answer is partial at best. You cannot improve what you do not measure, and you cannot claim a quality benefit you have no baseline for. So she starts collecting those three series immediately, before deployment, because a baseline gathered after launch is not a baseline, it is a guess with a date on it.

The arithmetic, once the baseline exists, is straightforward. Her average hire stays 2.5 years, and replacement cost runs 50 to 200 percent of annual salary depending on role complexity; for a $70,000 hire that is $35,000 to $140,000. If better screening improves retention by 3 months per hire across 300 hires, that is 75 person-years of retained talent. Separately, if a new hire needs 6 months to reach full productivity at $70,000, the ramp cost is about $35,000, and trimming ramp from 6 months to 5 saves roughly $5,800 per hire, which across 300 hires is about 25,000 hours of productivity recovered. The honest caveat is that quality value needs longer measurement timelines and better data than most functions have, so Adaeze presents it as the dimension with the widest error bars and the largest upside rather than letting the CFO discover that.

Dimension Three: Fairness Value

Fairness is often framed as compliance, as in "we need this tool to avoid legal risk," and Adaeze considers that framing backwards. Fairness is a positive value stream, not a burden. Blind resume screening removes demographic signals and lets the model evaluate job fit, which surfaces strong candidates the current process would never see and widens the pipeline. Broader pipelines mean better selection, and better selection is a quality argument as much as an equity one. Her hiring is roughly 70 percent male in a tech market where the benchmark is 65 percent, a 5-point gap that costs her access to about 5 percent of the available candidate pool. Is that pool lower quality? Not necessarily, and there is no reason to assume so. It is simply talent she is currently not reaching.

She quantifies the diversity effect carefully, because this is the dimension where business cases most often overreach. Research associates gender-diverse teams with higher innovation and ethnically diverse teams with better decision-making, and the temptation is to convert that into a large headline number. Adaeze refuses. She states it precisely and conservatively: "if our hiring were demographically representative we would have about 20 percent more women; research shows gender-diverse teams have around 15 percent higher innovation; modeling a conservative 5-point diversity improvement, we could unlock roughly 0.75 percent performance uplift across the organization." That is concrete, bounded, and defensible under a follow-up question, which is the only kind of number worth putting in front of an executive. She then carries $0 for fairness into the totals and describes the value qualitatively, letting it show up in the risk and brand columns where it can be defended arithmetically. These figures illustrate her own modeling, not an external statistic.

Dimension Four: Brand and Candidate Experience Value

Candidates talk about hiring experiences, and those conversations are the employer brand. Executives most reliably discount this dimension as soft, and it is also the one where an AI screening decision is most visible to the outside world, because every applicant who is filtered has an experience of being filtered. Adaeze treats brand as a two-sided ledger rather than an upside story. On the negative side, organizations known for opaque AI hiring, or for hiring discrimination, struggle to attract top talent; candidates route around them and the company competes in a smaller pool for the same roles. On the positive side, organizations known for transparent, fair, and efficient hiring attract candidates who heard about the process from someone they trust. Positive word-of-mouth is the strongest recruitment channel there is, and the one channel a company cannot buy directly.

She quantifies through two levers she already measures. The first is referral. Employee referral is her strongest source of hires at roughly 25 percent of all hires, and referral volume tracks employee and candidate satisfaction with the process closely. A 2-point increase in referral rate compounds into substantial recruitment value over time because referred hires arrive cheaper, faster, and with better retention. The second is offer acceptance, which is currently 70 percent, meaning 30 percent of offers are declined, and each decline restarts recruiting and extends time-to-fill on a role that has already consumed months. If a smoother, faster, fairer process raises acceptance from 70 to 75 percent across 300 hires, that 5-point gain is worth roughly $200,000 conservatively in avoided re-recruiting. Adaeze banks only the acceptance number and describes the referral and speed effects qualitatively as upside she expects but will not put in the total, because a brand claim that cannot be defended arithmetically damages every other number in the deck.

She then ties the brand argument to two hard mechanisms rather than to sentiment. The first is speed. At 200 candidates per role, faster screening shortens time-to-first-response, and candidates who hear back within 48 hours are materially less likely to have accepted a competing offer by the time recruiting reaches them. If her current first-response time averages 9 days and the tool cuts it to 3, she models a 1 to 2 point recovery in candidates who would otherwise have dropped out of the funnel entirely. The second is consistency. A model applied uniformly removes the variance of fifteen recruiters each screening to a slightly different bar on a busy Friday, and candidates notice consistency even when they cannot name it, because it is the difference between a process that feels like a process and one that feels like luck.

Crucially, Adaeze names the downside risk in the same breath, because an AI experience can damage the brand it is supposed to improve. A candidate rejected in 90 seconds by an opaque system tells a worse story than one rejected after a visible human review, and that story travels further because it is more interesting. So she builds in a transparency commitment as part of the case rather than as a compliance footnote: candidates are told that AI assists screening, and a human reviews borderline cases. She treats that commitment as a brand investment with a return, since the alternative is a process that saves time by producing exactly the experience most likely to cost her referrals.

Making the Brand Claim Measurable

Brand arguments get discounted because they are usually asserted rather than measured, so Adaeze instruments hers before deployment. She adds employer-brand perception questions to the existing candidate survey and the employee engagement survey, capturing a pre-deployment reading of how candidates and employees view the hiring process, then reads them again at six and twelve months. Alongside that she tracks the four operational numbers that brand actually moves through: referral rate, offer acceptance rate, time-to-first-response, and candidate survey scores segmented by whether the candidate was advanced or rejected. That last segmentation is the one most functions skip and the one that matters most, because the brand risk of AI screening lives almost entirely in the rejected population, and an aggregate score dominated by successful candidates hides a deteriorating story until it surfaces somewhere public.

Dimension Five: Risk Mitigation Value

The final dimension is risk, framed as value rather than cost. Compliance risk: deploying hiring AI without fairness monitoring creates regulatory exposure under frameworks like the EEOC's adverse-impact standard and NYC Local Law 144's bias-audit mandate; if an EEOC complaint could cost $500,000 in legal and settlement, a fairness-monitored tool that reduces that risk 30 percent is worth $150,000. Operational risk: a failed deployment might waste 6 months across 3 to 5 FTEs, $360,000 to $600,000 at $60 an hour, and a rigorous evaluation and pilot process that halves that risk is worth $180,000 to $300,000. Reputational risk: discriminatory or opaque hiring damages the brand and the damage shows up in the referral channel first; a 10-point drop in referral rate at $5,000 cost per hire across 300 hires is $150,000. Summed at midpoints, risk mitigation value is roughly $150,000 plus $250,000 plus $150,000, about $550,000.

Adaeze sharpens the compliance section so it survives a question from her General Counsel, who will want specifics rather than the word "exposure." She names the two regimes precisely. The EEOC's adverse-impact analysis rests on the four-fifths rule, also called the 0.80 disparate-impact ratio: if the selection rate for any protected group is less than 80 percent of the rate for the group with the highest selection rate, that is the threshold the agency treats as evidence of adverse impact warranting scrutiny. Concretely, if her tool advances 50 percent of male applicants but only 35 percent of female applicants, the ratio is 0.70, below 0.80, and the case is in trouble before it starts. So the business case does not merely promise "fairness monitoring," it commits to computing the four-fifths ratio for every protected class at each stage of the funnel and to halting rollout if any ratio falls below 0.80 without a documented job-related justification.

The second regime is NYC Local Law 144, which since its July 2023 enforcement date requires employers using an automated employment decision tool on candidates for NYC roles to commission an independent bias audit within the prior year, publish a summary of the audit results, and notify candidates at least ten business days before use. Because Adaeze's company hires in New York, this is not optional, and she puts the audit cost, roughly $30,000 to $50,000 annually for an independent auditor, directly into the cost column rather than hiding it. That honesty is part of the risk argument, since a case that has already priced in the audit is far more credible than one that discovers the obligation after launch. The publication requirement is also a brand fact, not only a legal one: the audit summary is a public document candidates and journalists can read, which means the tool's fairness result will eventually be visible whether or not she chooses to share it. The regulatory floor is also rising, with similar requirements emerging in other jurisdictions, so building the monitoring capability now is cheaper than retrofitting it under deadline.

Worked Example: Structuring Adaeze's Full Case

Adaeze assembles the five dimensions into one statement. Efficiency: $203,000 in annual redeployment value. Quality: $35,000 retention plus $6,000 productivity, about $41,000. Fairness and diversity: $0 quantified conservatively, but significant qualitatively and feeding both the brand and risk numbers. Brand and experience: $200,000 from improved offer acceptance. Risk mitigation: $550,000. Total annual value is about $994,000. Against that, costs: $150,000 tool, $80,000 implementation and training, $50,000 ongoing monitoring and governance, totaling $280,000. Net value is $994,000 minus $280,000, about $714,000 a year, a payback period of roughly 5 weeks. The numbers are illustrative of her own modeling rather than external benchmarks, but the architecture is the lesson.

How she structures the document matters as much as the arithmetic. She opens with a one-paragraph executive summary that states the net figure and its sources in a single breath: the tool delivers roughly $714,000 in net annual value through redeployed recruiting efficiency, improved hiring quality, expanded diversity, stronger employer brand, and reduced compliance risk, against $280,000 of implementation cost, with payback in about five weeks. Then a detail section showing the calculation for each dimension with its assumptions exposed and its sensitivity noted, including the dependency that if efficiency comes in lower, quality and brand have to be higher for the case to hold. Then a risk-mitigation section whose purpose is to demonstrate that this is not reckless deployment: a pilot plan, a governance structure, ongoing monitoring, and clear escalation procedures with named owners. Then a timeline showing the phased approach: pilot for 8 weeks, measure baseline fairness, train the team, then deploy broadly.

The Sensitivity Analysis That Saves the Case

A single net number invites a single objection: "your assumptions are too optimistic." Adaeze disarms that objection by bringing the sensitivity analysis to the table before anyone asks for it. The discipline is simple. She lists the assumptions the case depends on, flexes each one to a pessimistic value, and shows whether the case still clears its hurdle. Three assumptions drive almost all the variance. The first is adoption: she modeled the efficiency value at a 50 percent time saving, but that only materializes if recruiters actually trust the tool. If adoption lands at 60 percent of recruiters with a 20 percent override rate, effective time saved falls to roughly 30 percent rather than 50, cutting efficiency value from $203,000 to about $122,000, an $81,000 reduction. The second is the offer-acceptance lift: if acceptance rises only 2 points instead of 5, the brand value drops from $200,000 to about $80,000. The third is the compliance probability, the single largest line and the softest, so she models it both ways and shows the case at the low end.

The output is a small table she can defend line by line. In the base case, net value is about $714,000. In the pessimistic case, with low adoption, a 2-point acceptance lift, and the compliance benefit cut in half from $150,000 to $75,000, the value streams fall to roughly $122,000 efficiency, $41,000 quality, $80,000 brand, and $475,000 risk, about $718,000 gross, and after the unchanged $280,000 of cost the net is still positive at roughly $438,000. That is the sentence that wins the room: even if the three biggest assumptions all come in materially worse than planned, the case is still strongly net-positive, and the pilot exists precisely to replace these estimates with measured numbers. Adaeze also identifies the break-even point. Holding costs at $280,000, the case breaks even when total annual value falls below that figure, which would require efficiency, quality, brand, and risk to collapse together, a scenario the pilot would catch long before broad deployment. Showing the break-even, rather than only the expected value, signals to the CFO that she understands the limits of her own model.

Validating Assumptions in the Pilot

The pilot is not a formality before the real launch; it is the instrument that converts Adaeze's assumptions into evidence. She designs it around the three numbers the sensitivity analysis flagged as load-bearing. For adoption, she instruments the tool to log how often recruiters accept, edit, or override its rankings, and she sets a pre-registered threshold: if the override rate exceeds 25 percent after four weeks, that is a signal the model is not earning trust and the efficiency number must be revised down before any rollout. For fairness, she computes the four-fifths ratio on real pilot data across every protected class at every funnel stage, comparing it against the human-only baseline she started collecting before deployment, because a tool that merely matches a biased human baseline is not a defense; the goal is a 0.80-or-better ratio and a measurable improvement over the prior process.

Quality is the hardest to pilot honestly, since two-year retention cannot be measured in eight weeks. Rather than pretend otherwise, Adaeze uses leading indicators and labels them as proxies: hiring-manager satisfaction with the slate, early-stage interview pass-through rates, and first-90-day performance ratings. For brand she runs the candidate survey on the pilot population specifically, segmented by outcome, so she has an early read on whether rejected candidates experience the AI-assisted process worse than the manual one. She runs the pilot on two or three representative role families rather than the easiest ones, so the results generalize to the roles that will carry the volume. At the end she rewrites the case with measured values in place of estimates and re-presents it, having committed in advance to killing or redesigning the project if the data moves net value below the pessimistic-case floor. That commitment is what makes the original numbers credible: they are hypotheses with a falsification plan attached, not advocacy.

Three Anti-Patterns

The efficiency-only case. "This tool saves us $200,000 annually" is a complete business case in some organizations, and it is fragile. If adoption is slower than modeled, if the tool needs more override than expected, or if implementation runs over, the case collapses because there is nothing else holding it up. The failure has a predictable shape: six months in, fairness issues surface, remediation work begins, and the CFO says some version of "I thought this was going to save us money, and now we are spending money fixing it, so this was a waste." Adaeze loses credibility not because the tool failed but because she promised one thing and that one thing slipped. Always include quality, fairness, brand, and risk alongside efficiency, so the case survives a disappointment in any single dimension.

Unvalidated assumptions. A case that reads "the vendor says the tool saves 40 hours a week and we will assume 80 percent adoption, so 32 hours saved" is not a forecast, it is a stack of guesses presented as arithmetic. The year-one reality is often adoption at 60 percent rather than 80, a 20 percent override rate, and actual savings of 12 hours a week rather than 32, at which point the case has not materialized and every number in it is retroactively suspect. The fix is to validate the critical assumptions in a pilot, measuring adoption, override, and time savings on real usage, then update the full case before broad deployment rather than after.

Ignoring mitigation costs. Some cases calculate benefits carefully and treat the costs of responsible deployment as rounding error. Fairness monitoring, governance structure, training, and the independent audit are all real line items with real invoices. The failure mode is discovering mid-implementation that true costs are $300,000 rather than the $50,000 assumed, which turns a strong net figure into a marginal one and gets projects delayed or cancelled at exactly the moment the team has already spent the political capital. Include everything upfront: tool cost, implementation, training, ongoing governance, monitoring, and audit. Better to over-estimate and be pleasantly surprised than to under-estimate and face sticker shock in front of the people who approved the original number.

Practice

Each of these produces a document you could hand to a CFO, which is the standard. A business case that exists only as a conviction is not a business case.

• Build the five-dimension case. For your priority AI use case, calculate value across efficiency, quality, fairness and diversity, brand, and risk. Show your assumptions and your sensitivity for each. Subtract total cost from total value to get a net figure. That document is the full business case.
• Write an assumption validation plan. Identify the five most critical assumptions in your case, for example adoption rate, override rate, time savings, acceptance-rate lift, and fairness impact. For each, design a way to validate it during the pilot: what metric you will measure, and what specific result would make you confident or force a revision.
• Estimate costs honestly. List every cost of responsible deployment, including vendor fees, implementation, training, governance, monitoring, and independent audit. Give each a high and a low scenario, then compute net value in the best and worst cases and identify your break-even point.
• Draft the executive summary. Write a single page containing the top-line value proposition, value by dimension, costs, net value, payback period, and the risks with their mitigations. This is the artifact your stakeholders will actually read, so write it last and make it survive on its own.
• Run the sensitivity analysis. Ask whether the case still works if efficiency comes in 30 percent below projection. Ask what happens to ROI if fairness monitoring reveals a problem requiring tool modification. Identify which assumptions matter most, because those are the ones your pilot has to measure.

Reflection

Work through these before the meeting, since each one corresponds to a question somebody in the room is going to ask.

• What are the top three value drivers for your priority AI use case? Which dimension provides the most value, and which is hardest for you to quantify?
• What critical assumptions underlie your case, and if they disappoint, does the case still work? What is your break-even scenario?
• How will you communicate this case to different stakeholders? What matters most to your CFO, your CHRO, and your General Counsel, and does your case address each of them directly?
• What metrics will you track post-deployment to verify the case actually materialized across efficiency, quality, fairness, brand, and risk?
• If results disappoint in one dimension, what is your contingency? How much flexibility do you genuinely have to adjust the approach without restarting the approval process?

Glossary

• Employer brand. The reputation of an organization as a place to work. Strong employer brands attract better candidates, improve retention, and lower recruiting cost, and the quality of the hiring process is one of the main things that shapes them.
• Referral rate. The share of hires who come from employee referrals. High referral rates indicate a strong employer brand and satisfied employees, and organizations with good hiring experiences reliably see higher ones.
• Offer acceptance rate. The percentage of offers candidates accept. A 70 percent rate means 30 percent decline, and each decline extends time-to-fill because recruiting has to restart on a role it thought was closed.
• Time-to-productivity. The period between hire date and the point at which a new employee reaches full productivity, which for complex roles can run 6 to 12 months. Reducing it through better screening reduces ramp cost per hire.
• Retention rate. The share of hires who remain after one year, two years, and so on. High retention reduces replacement costs, and better hiring decisions are what improve it.
• Replacement cost. The total cost of replacing an employee who leaves, including recruiting, training, lost productivity, and management time. Typically 50 to 200 percent of annual salary depending on role complexity.
• Four-fifths rule. The EEOC's practical screen for adverse impact. If any group's selection rate is below 80 percent of the highest group's rate, the disparity is treated as evidence warranting investigation and justification.

Related Lessons

A five-dimensional case draws its inputs from work done elsewhere in the program, and it is worth knowing which lesson supplies which number.

• Building the Business Case: Efficiency, Quality, Fairness, and Risk covers the four-dimension version of this argument and how to sequence it for a specific room, which is the natural companion to the brand-inclusive case built here.
• Understanding Candidate Experience: What Matters in Hiring is where the brand dimension gets its substance, because a referral or acceptance-rate projection is only as good as your understanding of what actually shapes a candidate's impression.
• Metrics and Monitoring: Tracking Efficiency, Quality, and Fairness supplies the baselines this case depends on, and it is what lets you replace estimates with measured values after the pilot.
• Transparency and Disclosure: Telling Candidates About AI Use operationalizes the transparency commitment that protects the brand argument, including the candidate-notice obligation that Local Law 144 attaches to automated tools.
• Piloting and Iteration: Testing Workflows and Gathering Feedback is the mechanism behind the validation plan, including how to instrument override rates and collect feedback that is usable as evidence.
• Vendor and Tool Selection -- Evaluating AI Solutions turns the compliance and audit requirements in the risk dimension into shortlist criteria, so a vendor who cannot support an independent audit never reaches your cost column.

Closing

A multi-dimensional business case transforms an AI deployment from a cost-reduction initiative into a strategic investment. Instead of defending a narrow efficiency claim, Adaeze presents a comprehensive value proposition with several independent paths to value, which is why it appeals to different stakeholders and why it survives when one dimension underperforms. That structure also builds alignment as a byproduct: the CFO, the CHRO, and the General Counsel are each looking at a section written for them.

The discipline of building the complete case is worth something on its own, independent of whether the tool gets funded. You cannot stay vague about quality impact when you have to name the metric. You cannot hand-wave fairness when you have committed to a 0.80 threshold at every funnel stage. You cannot describe brand as important without deciding which number would move and by how much. And you cannot skip the costs when your credibility depends on the net figure holding up twelve months later. That rigor makes for better decisions and, over time, for a leader whose numbers people believe before they check them. The strongest cases are comprehensive, honest about costs, conservative in assumptions, and backed by pilot data.

Key Takeaways

• Multi-dimensional cases are more persuasive and more robust. Include quality, fairness, brand, and risk alongside efficiency so the case holds even when one dimension disappoints.
• Brand is a measurable dimension, not a sentiment. It moves through referral rate, offer acceptance, time-to-first-response, and candidate survey scores, and the rejected-candidate segment is where AI screening's brand risk actually lives.
• Name the brand downside as well as the upside. A candidate rejected in 90 seconds by an opaque system tells a worse story than one rejected after visible human review, which is why the transparency commitment belongs in the case as an investment.
• Quantify conservatively and validate in a pilot. Discount vendor claims, measure actual adoption and override rates, and update the full case with measured values before scaling.
• Include every cost of responsible deployment. Tool, implementation, training, governance, monitoring, and the independent bias audit are real; naming them upfront builds credibility and avoids late sticker shock.
• Frame fairness as value, not burden. Broader pipelines, better selection, reduced legal exposure under the four-fifths rule and Local Law 144, and stronger brand are positive value streams.
• Run a sensitivity analysis and find the break-even. Know which assumptions drive the case, show it still clears the bar when the three biggest all come in low, and state the point at which it stops working.
• Use the case to align the organization. A complete case speaks to the CFO on ROI, the CHRO on fairness and brand, and the General Counsel on risk, all at the same time.

Frequently Asked Questions

Why not just lead with the efficiency number, since it is the easiest to prove? Because the easiest number to prove is also the easiest to attack. The moment a skeptical executive asks about quality, fairness, or legal exposure, an efficiency-only case has no answer and looks naive. The four other dimensions are not padding; they are the answers to questions Adaeze knows are coming, and presenting them unprompted shifts her from defending a tool to leading a strategy.

How do I make the brand dimension credible to executives who treat it as soft? Convert it into operational metrics they already trust and only bank the ones you can defend arithmetically. Adaeze puts $200,000 in the total from the offer-acceptance improvement, which is a number her recruiting operations team already produces, and she leaves the referral and speed effects out of the total while describing them as expected upside. She also measures employer-brand perception before deployment so the twelve-month claim is a comparison rather than an assertion. Executives discount brand when it arrives as adjectives and engage with it when it arrives as an acceptance rate.

How do I justify the monitoring and audit costs to a CFO who sees them as overhead? Frame them as the price of the risk-mitigation value, not as standalone cost. The roughly $50,000 in ongoing monitoring and the $30,000 to $50,000 Local Law 144 bias audit are what convert a $550,000 risk exposure into a managed line item. A CFO who would never leave a $1 million asset uninsured understands paying a small, known premium to retire a large, uncertain liability.

What if the pilot data comes back worse than my business case assumed? Then the pilot did its job. Adaeze builds the case with a pessimistic-case floor and commits in advance to revising or stopping if measured values fall below it. Walking back a number with evidence costs far less credibility than defending an inflated number that reality has already contradicted, and it protects her standing for the next proposal she brings.

Do I really need a sensitivity analysis for an internal proposal? Yes, and bringing it unasked is a credibility multiplier. Executives assume any single number is optimistic. Showing that the case stays net-positive even when the three biggest assumptions all come in low tells the room you have stress-tested your own thinking, which is precisely the signal that earns the budget.

We do not track retention or time-to-productivity today. Can I still claim quality value? Not credibly, and you should say so rather than estimate around it. You cannot improve what you do not measure, and a quality claim without a baseline is the first thing a CFO will pull on. Adaeze instead starts collecting retention by hire, performance ratings, and time-to-productivity immediately, before the tool deploys, then presents quality as the dimension with the widest error bars and the largest potential upside. That is a stronger position than a confident number nobody can verify.