Every vendor's ROI slide skips straight to the number that closes the deal. The math behind it rarely makes the presentation.
That's a problem. The real return on an AI voice agent depends on your firm's numbers, not a vendor's average customer.
Case volume, current staffing cost, and how many calls you're already missing all move the outcome. This guide walks through the calculation step by step, with a worked example you can adapt to your own firm.
It ends with the questions worth asking before you take any vendor's ROI claim at face value. Client expectations are part of the backdrop here too. Over half of legal clients now say they turn to AI tools first when they need help.
Step 1: Establish your real call volume
Start with how many calls your firm actually handles, not how many your intake team logs. A single PI case generates roughly 150 calls over its lifetime.
That volume includes client updates, medical provider follow-ups, insurer exchanges, and lien holder status checks. A case manager carrying 100 active cases is therefore fielding somewhere around 15,000 calls a year.
Multiply that figure across every case manager at your firm to get a firm-wide number. Most of that volume is routine — status checks, balance verifications, and scheduling calls rarely require legal judgment.
They still consume a case manager's day, one call at a time. Before you can calculate any savings, you need an honest count of how many of those calls exist.
It helps to break that volume into rough categories before you go further. New-client intake is one bucket, existing-client status updates are another. Medical-provider, vendor, and insurer or lien-holder follow-up calls round out the rest.
Each category automates differently, and vendors price and score their tools differently against each one. Some platforms handle intake well but struggle with operational call types; others do the reverse.
Pull two or three weeks of real call logs if your phone system keeps them. A short sample is usually enough to see which category actually dominates your firm's volume, rather than guessing from memory. That breakdown also tells you which vendor's strengths actually match your call mix.
Step 2: Price the case-manager time you'd get back
Not every call needs automating. The goal is pricing out the hours spent on the ones that do.
Firms running AI voice agents for PI operational calls report recovering 50+ hours of case-manager time weekly. That's the starting figure for this step, but your firm's actual number will depend on current call volume and staffing.
To turn hours into a dollar figure, multiply reclaimed hours by your fully-loaded case-manager cost per hour. That means salary, benefits, and overhead combined — not just base pay.
Use your own payroll number here, not an industry average. Loaded labor costs vary widely by market, firm size, and how case-manager roles are structured.
A firm reclaiming 50 hours a week nets real money. At $38 an hour in loaded cost, that's roughly $7,900 a month in staff time. That's before any revenue impact gets added in.
That figure only covers time spent on calls the system fully resolves. Partially-automated calls save less time per call.
That's often what a generic AI receptionist delivers — it gathers information, but a human still finishes the conversation. Track those calls separately during a pilot, rather than folding them into the same number.
Step 3: Calculate what missed calls are already costing you
This is the number most ROI conversations skip entirely. Industry research shows missed calls carry real cost. Between 62% and 78% of prospective clients hire the first attorney who actually speaks with them.
Separately, roughly 35–50% of inbound calls at PI firms go unanswered during business hours. After-hours volume gets missed at an even higher rate, since most firms don't staff overnight coverage at the same level.
Each missed intake call risks $5,000 to $50,000 in lost contingency revenue, depending on case type and severity. That range is wide because case values vary enormously across PI practice areas.
You don't need to assume every missed call converts into a case. Even a conservative estimate works: one previously-missed call converting to a signed case per quarter. Attach a real contingency-fee value to that single case, and the number still moves the calculation.
Pull your own average case value from last year's closed matters if you have it. That number will be more accurate than any industry range, including the one above.
Step 4: Add up what you're already spending
Compare the AI voice agent's cost against what you're currently paying to cover the same call volume, not against zero. Most PI firms already pay for some form of coverage.
Answering services like Ruby, Smith.ai, and Lex typically run $3,000 to $5,000 a month for comparable coverage. That cost doesn't shrink as your CMS documentation burden grows.
Most are billed per minute or per seat, regardless of what actually gets logged into the case file.
Overtime or after-hours staffing costs belong in this column too. Include them if that's how your firm currently covers weekend and evening calls.
Don't forget the hidden cost of manual documentation. Time your team spends copying call notes into the case file is real cost. No invoice reflects it directly, but the payroll hours are the same either way.
Don't skip the total cost of ownership
The number on a vendor's pricing page is rarely the whole cost. Before you plug a monthly figure into the formula, check for setup fees and minimum contract terms. Also check per-minute overage rates once you exceed a plan's included minutes.
Ask specifically what happens at your projected call volume, not the vendor's example volume. A plan that looks affordable at 500 minutes a month can get expensive fast. Run past that cap, and overage rates start compounding.
Also account for the migration cost. Porting phone numbers, retraining staff, and running a pilot alongside your existing system all take real time. That time has a dollar value, even as a one-time cost rather than a recurring one.
Ask, too, whether pricing changes as your call volume grows over the contract term. Some vendors lower the per-minute rate at higher volume; others raise the platform fee at renewal regardless of usage.
None of this should scare you off the calculation. It just means the "new tool cost" line should reflect what you'll actually pay in month three. Use that figure instead of the number on the pricing page in month one.
Step 5: Run the formula
Once you have the four inputs above, the calculation is straightforward. Add up the value, subtract the new cost, and divide by that new cost.
ROI = (Time value recovered + Revenue recovered + Current spend eliminated − New tool cost) ÷ New tool cost × 100
Everything in the numerator is value. Everything you'd still pay for the new system comes out of it.
The result is a percentage return, not just a raw dollar figure. That makes it easier to compare against other spending decisions your firm is weighing this year.
A worked example: a 120-case PI firm
Here's how that formula plays out for an illustrative firm with 120 active cases and one dedicated case manager. Treat this as a template, not a benchmark — swap in your own numbers before you draw conclusions.
- Case-manager hours reclaimed
- Monthly value: $7,900
- How it's calculated: 50 hrs/week × 4.33 weeks × $38/hr loaded cost
- Answering-service cost eliminated
- Monthly value: $4,000
- How it's calculated: Mid-range of the $3,000–$5,000/month legacy-vendor bracket
- Missed-case revenue recovered
- Monthly value: $4,167
- How it's calculated: One recovered case per quarter, at a $12,500 average value, divided across 3 months
- New AI voice agent cost
- Monthly value: –$3,500
- How it's calculated: Illustrative monthly cost, roughly 50% below legacy answering services
- Net monthly value
- Monthly value: $12,567
- How it's calculated: Sum of the line items above
At a $3,500 monthly tool cost, that's a net monthly value of $12,567. Divided by the tool's own cost, that works out to roughly 359% ROI.
Even if you drop the revenue-recovery line entirely, the time-and-spend savings alone still clear the tool's cost several times over. That's worth checking against your own numbers.
The revenue line is the one most worth stress-testing on your own firm. It's the most volatile input in the whole formula, and the easiest one for a vendor to inflate.
Why the same formula produces different numbers for every firm
Two firms with the same case count can land on very different ROI figures. A firm already paying $5,000 a month for a legacy answering service starts from a higher baseline. One paying nothing and simply eating missed calls starts from zero on that line.
Case value matters just as much as case count. A firm handling catastrophic injury cases with six-figure settlements will see a much larger revenue-recovery line. A firm running high-volume, lower-value soft-tissue claims will see a smaller one, even with more total cases.
Staffing structure changes the time-savings line too. A case manager already handling 150 cases efficiently has less slack to reclaim than one who is visibly underwater.
Practice mix matters as well. A firm with heavy Spanish-speaking client volume or multiple co-counsel relationships generates call patterns unlike a straightforward single-plaintiff PI practice.
A large share of mass-tort cases shifts the pattern again. That difference changes which calls are worth automating first.
Run the calculation with your own numbers before comparing your result to anyone else's. The formula is universal. The inputs are not.
A second example: a smaller firm
Not every firm runs 120 cases. Here's the same math applied to a smaller PI practice with 45 active cases and a part-time case manager role.
At that volume, reclaimed hours drop to roughly 20 a week. Current answering-service spend is closer to $1,500 a month on a lighter plan. Missed-case recovery slows to about one case every two quarters, given the lower call volume.
Run those numbers through the same formula, and the tool's cost still clears at a smaller but meaningful margin. That's often in the 100–200% range, rather than the 300%-plus seen at higher volume.
The lesson isn't that smaller firms should skip the calculation. ROI scales with volume. A firm below a certain size may do better starting with an entry-tier platform, rather than a full enterprise contract.
Questions to ask before you trust a vendor's ROI number
A vendor's ROI slide is only as good as the assumptions behind it. Before you accept one, ask the following:
- What call volume and case count is this estimate based on — yours, or their average customer's?
- Does the time-savings figure assume full automation, or a realistic resolution rate for your specific call types?
- Is the missed-call revenue estimate based on your actual contingency-fee range, or a generic industry number?
- What's included in "current spend" — just the answering service, or overtime and after-hours coverage too?
- Does the estimate account for a ramp-up period, or assume day-one performance?
- Can they walk you through the math behind the number, not just show the final percentage?
- Will they put a written estimate in front of you before you sign anything?
- What happens to the estimate if your actual call volume comes in lower than projected?
If a vendor can't answer most of these clearly, treat their ROI number as marketing copy. A number built on your firm's actual call data is worth far more than one built on an average customer.
How long it actually takes to see returns
Time savings show up fastest. Most firms see a measurable drop in case-manager call load within the first two to three weeks of a pilot. That's once configuration settles and the system handles steady-state call volume.
Revenue recovery from previously-missed calls takes longer to confirm. It depends on cases actually converting and reaching a fee event, which can take months in a contingency-fee practice.
Run your ROI calculation twice. The first pass uses time and spend savings alone — that's the number you can verify during a two-week pilot. The second pass adds revenue recovery, the fuller upside once the system has run a full case cycle.
Ask for a custom ROI estimate built from your own call data before the pilot starts. That gives you a number to check your first-pass results against.
Frequently asked questions
Should I include the cost of switching vendors in my ROI calculation?
Yes, if you're currently under contract with another service. Factor in any early-termination fees, and account for a short overlap period while your team transitions off the old system.
Does ROI look different for a firm that isn't currently using any answering service?
Yes. The "current spend eliminated" line in the formula drops to zero. The calculation still works, but the return leans more heavily on time savings and missed-call recovery.
How do I estimate my firm's average missed-call value if I don't track it today?
Pull your average signed-case value from last year's closed matters. Apply a conservative conversion rate — 10–25% is a reasonable starting range for calls that reach a live conversation. Adjust that range once you have real pilot data.
What if my firm's call volume is too low to justify a dedicated tool?
Run the calculation anyway. Even at lower volume, the time-savings math often still clears an entry-tier platform's cost. Case managers at smaller firms are usually stretched across more roles.
Should outbound calls count toward the ROI calculation?
Yes, if case managers currently place outbound calls for records requests or client check-ins. That time counts the same way inbound handling time does. It's hours a system can take off a case manager's plate.
Does firm size change which cost inputs matter most?
Somewhat. Smaller firms tend to see the time-savings line matter more, since one case manager often covers several roles.
Larger firms with dedicated case-manager teams tend to see the revenue-recovery line grow faster. Missed-call volume scales with total caseload.
How often should this calculation get revisited?
Rerun it any time call volume, staffing, or case mix shifts meaningfully. Most firms find once or twice a year is enough. A growing caseload changes every input in the formula, so last year's number won't necessarily hold.
Conclusion: Run your own numbers before you run a pilot
The ROI of an AI voice agent isn't a fixed number any vendor can hand you. It's a function of your call volume, your current spend, and how honestly you estimate the revenue side.
Build the calculation with your own inputs before you talk to a single vendor. That gives you a number to hold every pitch against, instead of accepting whichever slide looks most convincing.
Bring your call-volume breakdown, current spend, and loaded case-manager cost into every vendor conversation. Ask each one to model their ROI number against those same real inputs. Don't accept their own example scenario or a generic customer average instead.
For a PI firm weighing this against a generic AI receptionist, the same calculation applies. HelloCounsel runs this exact math during a pilot, using real call data, not an average customer's. A custom ROI estimate comes with every 2-week pilot, built from your firm's actual case volume and current spend.
Share this article
