How HR Can Contribute to Business Growth: The Chain to a CFO Number

Your CEO asks what HR will contribute to growth next year. The deck is ready. Engagement is up. Attrition is down on last year. Two leadership programs shipped, and the manager scores moved in the right direction.

Then the CFO asks which of those shows up in the plan, and the room goes quiet for a second too long.

The question is fair. It’s the same one the CFO asks marketing and sales, and those functions answer easily because their work lands on a line somebody already reports. HR’s work doesn’t. Under the accounting standards most companies report on, money spent on people is an expense in the period you spend it, and internally generated capability never becomes an asset on the balance sheet. Nothing in your accounts gets bigger when your people get better.

So HR has to borrow. Every growth claim has to start on a lever you control and end on a number somebody outside HR already reports and would defend upstairs.

Below is the chain we build with clients, and where AI moves it.

The three answers that lose the room

The engagement score is the most common one. It’s a real signal, and as the headline it fails. Your CFO can’t price a point of engagement, and has no reason to accept that the causation runs from your survey into their plan. You’re asking the room to trust a number only HR produces and only HR can move.

Culture is the second, and the hardest thing to hand to anyone in a finance meeting. It shows up as behaviors that change what work costs and how long people stay, and those effects have names and owners. Say those. The word itself buys you nothing at that table.

Retention framed as a virtue is the third, and it gets closest before it slips. Attrition down sounds unarguable until someone asks which leavers. People who were going anyway, from roles you were about to reshape, cost you very little. Four experienced engineers off the team the plan depends on cost a specific, computable amount. A blended number hides the difference.

None of these three is wrong. Each stops a link short of something finance can use.

Growth from HR is two numbers: how well people perform, and how long they stay

Employee lifetime value is the answer that holds. How much a person contributes across their time with you, against what they cost, and how long that time runs. Performance and tenure. It’s the number Culturro has worked on since 2017, when we started as a people consultancy.

The useful part is what it does to your list. Every activity HR funds raises performance, extends tenure, or does neither. That third bucket is mostly downside protection: employee relations, compliance, pay equity, safety. Those don’t raise eLTV. They stop it collapsing. A CFO understands insurance perfectly well when it’s called insurance, and gets suspicious when it’s dressed up as growth.

Sort your list into those buckets before the meeting. The sorting alone changes what you walk in with.

The chain, one lever at a time

Four columns. The lever you control, the mechanism, the operating number that moves, and where the money shows up. If you can’t fill the fourth column, you don’t have a growth argument yet.

Levers that move performance:

HR lever	What it changes	The operating number	Where it lands
Selection at hire	The level people start at and how fast they climb	First-year performance spread by role family, from your own review data	Output per head, supervision and rework cost
Onboarding and ramp	Weeks of payroll before a new hire carries a full load	Time to productivity, by role family	Payroll spent on non-productive weeks. For quota roles, revenue timing in the plan
Manager quality	How much of a team’s capacity turns into finished work	Goal attainment and cycle time in the team’s own operating report	Output per head, overtime, contractor cover
Skills held inside the company	Share of the work delivered in-house	Contractor, agency and outsourced spend by function	Operating expense, directly, no modeling required

Levers that move tenure:

HR lever	What it changes	The operating number	Where it lands
Stay conversations in the segments the plan depends on	Whether experienced people are still here in eighteen months	Regretted attrition by segment, with a replacement cost range	Recruiting spend, agency fees, overtime cover, re-ramp payroll
Pay position inside band for specific segments	When a resignation lands, and whether it’s a surprise	Pay-band position against market, by segment	Payroll, and unplanned backfill cost
Career paths that exist in practice	Whether the next step sits inside the company or outside	Internal fill rate, tenure at level	Cost per hire, agency fees, know-how walking out
Manager quality	The reason given most often when good people leave	Regretted attrition by manager, at cohort grain	The same lines as above

Manager quality sits in both tables. That’s why it’s the highest-value lever HR owns, and the hardest one to buy. It doesn’t arrive as a system. It arrives as a few hundred people getting better at conversations they currently avoid.

Two things about taking these tables into a meeting. Pick one row. Eight rows reads as a wish list, and your CFO will test the weakest. Put a range on the fourth column with the assumptions visible, since a range you can defend beats a point estimate you can’t.

The last link has to be somebody else’s number

Contractor spend belongs to the function head. Time-to-fill belongs to your TA lead. Replacement cost assumptions belong to Finance, and use theirs even when you think yours are better. You want the argument to be about the people work. A last column HR built alone hands your CFO the easiest way to say no.

Raise the causation problem before someone else does. You can’t prove a coaching program caused a margin movement, since too much else moved in the same quarter. What you can do is name the mechanism, agree the operating number in advance, and say what result would make you stop funding it. Saying that third part first is what makes the first two credible.

One more thing if you operate in Europe. Under the EU sustainability reporting standards, in-scope companies report on their own workforce, and those disclosures get read by people who have never opened your engagement dashboard. Numbers that leave the building are numbers you’ll be asked about.

Where AI moves this chain, and where it doesn’t

In HR, trust sets the sequence. Return doesn’t. A wrong answer here lands on a named person, and there’s no rerun that fixes it. So we score each idea on what it does to eLTV, what trust it has to earn before anyone will use it, and whether the data and the ownership are there. Trust gets written in words, Little, Some, Real, A lot, The most, so a strong impact number can never cancel out a trust problem. The method sits on the framework page, and how to prioritize HR AI use cases works through the order it produces.

Two agents sit on the chain above, one on each side.

A coaching agent for managers, on the performance side

It lives in Slack or Teams and coaches a manager through whatever’s in front of them. A report who’s gone quiet. A promotion request they can’t grant. A peer conflict they’ve been avoiding. It asks what a good coach would ask, offers a few approaches, and helps the manager land on a plan and the words to use. The manager decides what to do with it; L&D sets the scope and the privacy rules.

The chain is short here. Better manager conversations move goal attainment and cycle time in the team’s own operating report, and regretted attrition on that team. Both sides of eLTV, one lever.

Trust to earn: Some. The privacy model is the product. Conversations stay private to the manager, never shown to their own manager or to HR, never retained as HR records, and only aggregate capability trends leave the system. It reads nothing from your HRIS or your performance records, knows only what the manager types, and points at HR, legal or your ethics line when a situation needs a formal process. Announce it as something HR rolled out to keep an eye on managers and nobody will touch it. The agent page carries the guardrails and the proof design.

We’ve run this one. Merlin, the coach inside Risely, has worked with 5,000+ people from 40+ companies, in 40 languages, across 83 workplace skills. What we see is consistent: managers come back unprompted when the privacy line is real, and stop the week they suspect it isn’t.

A retention investment review agent, on the tenure side

This one is the chain in software. It turns your cohort attrition drivers into a segment-level watchlist, say engineers between eighteen and thirty months of tenure sitting below pay-band midpoint, with an estimated cost of losing against retaining that segment. HRBPs and the CHRO decide which segments get funded: stay-interview programs, band corrections, career-path work. That’s the fourth column of the tenure table, at a grain the CFO can act on.

Trust to earn: A lot. It sits one step from scoring individuals and one step from pay decisions, so the constraints matter more than the model. Segments only, above a minimum group size, no names, no individual scores. Any individual action, a stay conversation or a retention adjustment, stays with the manager and HR with the person in front of them, and never gets triggered by an output. Cost ranges come from Finance, and the agent doesn’t set them. Distribution stops at HRBPs and the CHRO, each recipient logged, and your works council gets consulted where a retention pay adjustment might follow. The detail is on the agent page, and the other 120 sit on the map.

The parts AI can’t do

It can’t create a link that doesn’t exist. If nobody outside HR owns your fourth column, an agent hands you a better-formatted version of the same unowned claim, faster.

It can’t decide about a person, and we don’t build systems that try. AI may inform, surface, rank, summarize or draft. A person makes every call about a person, and the system gets built so it can’t do otherwise. “A person decides” is also a different design from “a person can override”, and only the first one holds by week three.

And a flight-risk score on every employee raises nothing while spending the most trust on the list. It hands a manager a verdict about a named person before they’ve had a thought of their own, and nobody in the demo can say what that manager does on Tuesday when a name turns amber.

What to bring to the next growth conversation

Four things. None of them is a survey result.

1. One borrowed number. The single operating number in next year’s plan that people work moves. Contractor spend in engineering. Ramp time for the sales hires the plan assumes. Name whose report it comes from.
2. One lever, at the size you’d actually fund it. “Coaching for the forty managers in the two functions that are hiring” carries weight in that room. “Investment in leadership” doesn’t.
3. One owner outside HR. The person who reports that number and will say in the meeting that they expect it to move. If nobody will say it, you found that out before spending anything.
4. One date, and what failure looks like. What you expect to see by when, and what result would make you stop.

That’s a growth conversation, and it takes an afternoon to prepare.

Start with the lever you’re already funding

Take a program that’s live right now and write out its four columns. Most teams stall on the fourth, and the stall is the finding. It tells you which parts of your spend have a defensible growth story and which have been running on goodwill.

Then take the AI idea on your desk and score it on the 10X HR-AI Framework. Eight questions, a couple of minutes, an honest tier at the end, including the tier that tells you to wait.

Doing that for every lever in your function, working out which chains your data can carry this year, and putting a defensible number on what the list does to eLTV is the strategy month. Here’s how it runs.

Frequently asked questions

My CFO wants causal proof and I can’t give it. What now?

Say that first, then say what you can give: the mechanism, the operating number, the owner, and the date you’d both look at it. Finance funds plenty of things on that basis. What loses credibility is a causal claim that collapses under one follow-up question.

Which number should HR own outright?

Regretted attrition at segment grain, with the definition written down and agreed with the function heads. It’s the one people number that’s unambiguously HR’s, unambiguously about growth, and defensible without anyone else’s model. Everything else in the chain, borrow.

So engagement surveys are pointless?

No. They’re a leading indicator, and they belong in the middle of the chain between the lever and the operating number, telling you months early whether something’s working. They just can’t carry the headline in a growth conversation, because nobody can price them.

How long before a chain like this shows anything?

Tenure levers take about a year to read properly, since you need enough movement to see a pattern at segment grain. Performance levers show earlier. Ramp time and contractor spend move within a quarter or two when the lever was real. Pick one of each.

What if the straight answer is that HR can’t move the number the plan depends on?

Then say so, and say which part of the downside HR is protecting. A CHRO who walks in and rules out two of their own programs gets listened to on the third one. That’s the trade.