Most AI-program ROI presentations to CFOs fail not because the technology is doubted but because the assumptions aren’t defensible. The CFO believes the vendor’s number could be true for some operator. The CFO doesn’t believe it has been computed correctly for theirs. This chapter walks through how the 15% free-cash-flow uplift and 40% OPEX reduction numbers are actually built, with a worked example and the sensitivity points so you can defend the number end-to-end without the vendor on the call.
The structure: where the 15% comes from, where the 40% comes from, what they sometimes double-count, what the investment side actually looks like, and what kills the model. Then a link to the live calculator at the end so you can plug in your own numbers.
Where the 15% free-cash-flow uplift comes from
Three components, additive, expressed as percentage of baseline FCF.
Component 1: Recovered deferred production. The biggest line. When the daily plan is ranked by economic impact, wells with the largest dollar deltas between forecast and actual production get visited first. Operators we work with typically capture 30–50% of the previously deferred production within the first quarter. The math: (deferred-production rate × % captured × commodity price × working interest) summed across the well set, then annualized.
Component 2: Operator hour reallocation. When 25% of miles get cut from the route plan, the time recovered is real time, not paper time. Operators do not shrink, they redirect: more time on high-value wells, less time on routes that don’t matter. The math: (hours/operator/day saved × loaded-labor cost × number of operators × 250 working days). The CFO will want a conservative redeployment factor here, typically 50–70% of the saved hours, since some of the time gets re-absorbed into administrative work.
Component 3: Reduced intervention costs. The 50% reduction in mean time to resolution for economic well-failure repairs translates into both faster production recovery (already counted in Component 1) and lower repair costs (separate line). Lost production from longer downtime is the larger of the two; pure repair-cost savings run 5–10% of intervention budget, not 50%.
Worked example: 2,000-well mid-tier operator
Take a mid-tier operator with 2,000 active wells averaging 10 bbl/day per well, 25% net working interest, and assume $80 WTI realized after differentials. Baseline daily production is 20,000 bbl, $1.6M/day, $400K/day to net interest, roughly $146M/year of net revenue.
If pre-WorkSync deferred production runs at 5% of total (a conservative figure for an operator with disconnected systems and habit-based routing), and WorkSync recovers 40% of that within six months, the captured incremental is 5% × 40% = 2% of total production. On 20,000 bbl/day, that is 400 incremental bbl/day, $32K/day at $80, $8K/day to net interest, roughly $2.9M/year of recovered cash flow.
Add ~$300K/year of redeployed operator hours and ~$200K/year of intervention cost reduction and the total cash-flow lift lands around $3.4M on a baseline of around $20M of unleveraged free cash flow at this scale, which is roughly 17%, in line with the 15%+ figure we observe in the field. Different operators with different deferred-production profiles or different price-realization will land differently. The framework holds; the inputs vary.
The CFO doesn't doubt the technology. The CFO doubts that your assumptions hold. Show the inputs, defend the inputs, and the output defends itself.
Where the 40% OPEX reduction comes from
The 40% OPEX number is specifically the field-operations transformation outcome from the Work Order Management module — what an operator measures against a Maximo or SAP implementation as the comparison point. The components:
- ▸Truck-roll cost. 25–35% reduction in site visits + miles driven. Per-visit cost includes fuel, vehicle wear, driver hours. Translates directly to the field-OPEX line.
- ▸Tank gauging labor. 60% reduction in manual gauging via predictive Liquid Management Index. Per-gauge labor × number of gauging events × cost-per-hour.
- ▸Maintenance reaction-time savings. 50% reduction in MTTR translates to less crew time on each individual repair (planners aren’t hunting for parts and crew schedules).
- ▸Software-rationalization savings. Operators replacing 3–5 disconnected tools with one platform see real licensing and integration-maintenance savings. Often the largest line item in the 40% but the most operator-specific.
What gets double-counted. Deferred production is FCF, not OPEX. If your CFO sees the same dollar in both columns, the model is wrong. Working capital tied up in tank inventory (the 70% inventory reduction figure) is balance-sheet, not income-statement — it shows up in CFO conversations as cash conversion, not as OPEX.
The investment side: software, services, change management
The CFO’s second question, after “defend the uplift,” is “what does it cost to get there.” Three components again, all of which need to be in the model.
Software license. WorkSync GOOD-tier modules land in the low-$20K-to-$40K-per- year range per module. The Work Engine GOOD package, which most upstream operators start with, is around $40K including the free Data Hub foundation. The full closed-loop platform (all 17 modules) sits around $400K ARR, but most operators will land at GOOD or BETTER tier first and expand. Every tier is priced below typical VP signing authority for mid-tier operators, no procurement committee required.
Implementation services. Quoted separately from software. Most cost is in the first two weeks (integration plus Data Hub stand-up); the rest is training and adoption support. For operators with a clean stack (Enertia, Quorum, AVEVA PI, Inductive Automation Ignition, ArcGIS, Maximo) this typically runs at the lower end of the range. Custom-SCADA or 1990s-era accounting systems push it higher. The 6-week paid pilot model ($15–25K, credited to license) lets the operator test scope before committing to full deployment.
Internal change management. Not zero. A foreman moving from a phone-tree morning to reviewing a ranked plan needs training. A superintendent tuning the scoring model needs the framework. Pumpers adopting the truck-cab tablet need the workflow and the confidence to question the plan when their judgment disagrees. The internal hours are real — expect 1–2 weeks of senior superintendent time per basin during rollout, plus a multi-month adoption tail. Not a separate line on the invoice; a line in your operating budget.
Payback period in plain numbers
Continuing the worked example. The 2,000-well operator at Work Engine GOOD tier ($40K license + ~$50K implementation + ~$30K internal change management for a single basin = $120K total first-year investment) against $3.4M of annual cash-flow lift hits payback inside the first quarter. The OPEX line items are gravy on top.
The smaller the operator, the faster the relative payback but the smaller the absolute dollars. The largest operators we work with (5,000+ wells across multiple basins) see full-platform payback in 4–6 months but have 8-figure multi-year value capture, which is the conversation that gets the program from VP-signing-authority into board-room material.
What kills the model and what saves it
What kills it. Sustained sub-$50 oil. The 15% FCF lift is computed on a commodity-price assumption; if WTI sits at $40 for an extended period, the absolute dollars contract and the payback period stretches. The math still works in percentage terms because every component scales with price, but the CFO conversation gets harder when the absolute number shrinks. Counterintuitively, a low-price environment is also when AI compounding matters most — the operator running the AI-driven loop survives the price downturn at higher net cash flow than the operator that didn’t.
What saves it. M&A integration. An operator that closes a deal and brings the acquired asset onto the same ranked work loop within the first quarter realizes the FCF uplift on the new asset base immediately, instead of waiting six to eighteen months for an integration project to harmonize the CMMS and accounting stacks. For serial acquirers, this is the line item the CFO wants on every deal model going forward.
Take it to the calculator
The worked example above uses placeholder inputs. Yours will differ. The live ROI calculator at /roi-calculator takes your well count, average production, working interest, and current OPEX line items and produces a number in your data. If that number is defensible to the CFO at internal review, the next conversation is the 6-week paid pilot — Week 0 success-metric scoping, Week 6 executive readout, $15–25K credited to license. That structure puts the operator in a yes/no position in 60 days, not the 12-month enterprise pilot cycle that traditional vendors require.
Chapter 11 picks up exactly there: the week-by-week structure of a 90-day rollout that takes you from spreadsheet to production-grade ranked work plan, with the metrics defined up front and the proof points landing in sequence.
The 90-day path from Era 2 to Era 4
Week-by-week. What integrates first. What ships in week 1. What you measure in week 6. What you take to the board at day 90.