The 90-day path from Era 2 to Era 4

Chapter 10 explained where the 15% free-cash-flow uplift comes from. This chapter explains how the operator gets there. Ninety days, three phases, with the proof points landing in sequence so the program funds itself before the board has to revisit the original commitment. We have run this rollout pattern across upstream, midstream, and gas- utility deployments enough times that I can describe it to you week by week. Whichever AI vendor you pick, the structure should look something like this. The integration realities and change-management requirements don’t depend on the vendor.

Week 0, scope and metrics

The Week-0 conversation matters more than the rest of the rollout combined. The job is to define what success looks like before any code ships. Three things have to be on paper:

• ▸The success metric. One number. “5% deferred-production capture in basin X by Week 6” is a metric. “Improve operations” is not. The metric should map to one of the proof points the executive committee will see at Day 90.
• ▸The baseline. The same metric, measured today, on paper. Without a baseline, any improvement is invisible at executive readout. With a baseline, the delta is the program’s case for continued investment.
• ▸Who owns the metric. Not the vendor. Not the IT lead. The operations leader whose number it is. If the VP Ops doesn’t own the success metric, the program is a side project, not a transformation.

The most common Week-0 failure: skipping it. Operators who move directly into integration without the metrics on paper invariably succeed at “deploying software” at Day 90 rather than at “moving the cash-flow number.” Both happen, but only the second one funds Phase 2.

Weeks 1–2, integration window

The integration mechanics are the day-by-day breakdown from Chapter 4: SCADA tap (Day 1–2), ERP/CMMS/GIS schema mapping (Day 3–4), reconciliation against historical data (Day 5–6), validation and go-live for one basin (Day 7). Week 2 is the tail of the integration plus the first scoring-model calibration runs.

By the end of Week 2, three things should be true:

• ▸SCADA, ERP, CMMS, and GIS data are flowing into the unified asset model. The reconciliation agent has built the canonical asset graph and flagged any conflicts for engineer review.
• ▸The scoring model has run nightly against the last 30+ days of historical data. The output is a list of ranked work that the superintendent can verify against their own intuition for the basin.
• ▸The Week-0 baseline metric has been confirmed in the new system — same number, same calculation, same period, regardless of how the operator was tracking it before.

Weeks 3–4, ranked plan goes live in one basin

Field crews start receiving the ranked plan on the truck-cab tablet at 6 AM. The foreman reviews and approves the day’s plan in the same five-to-ten minutes that used to be the morning phone tree. The pumpers run the plan. The first real feedback loops close: completed work orders feed back to the scoring model, observed outcomes adjust the priority weights, the next morning’s plan is informed by yesterday’s execution.

Week 3 is the cultural test. Operators who have been running their day from a yellow pad and a phone for fifteen years are now running it from a tablet. The expected reaction is skepticism: that well isn’t actually critical, the system doesn’t know our basin. The superintendent’s job during Weeks 3–4 is to surface those disagreements as scoring-model adjustments rather than as “the system is wrong.” Every operator override in Weeks 3–4 is data the model uses to calibrate the scoring weights for that basin’s specific behavior.

The ranked plan in Week 3 is not the final ranked plan. It's the starting position for the conversation that shapes the final ranked plan in Week 4.

By Week 4, override rates should be falling, the superintendent should be tuning weights instead of second-guessing every assignment, and the field crews should be calling the plan “today’s plan” instead of “the system’s plan.” That language shift is the one I look for as the leading indicator that the rest of the rollout will work.

Weeks 5–6, measure and Week-6 executive readout

The Week-0 metric is now measured against four weeks of ranked-plan execution. The Week-6 executive readout is the make-or-break moment for the program, and the structure of the readout should be predictable enough that the executive committee can absorb it in 20 minutes:

• ▸Where the metric started, where it is now. One slide.
• ▸What changed in the work loop to drive the move. Specific. “The top 50 wells by deferred-production capture got visited 1.4× more than baseline; the bottom 50 got visited 0.6× as often.”
• ▸What didn’t work, honestly. Every Week-6 readout should have a section on what surprised us, where the model needs more calibration, where the field is still pushing back. If your readout doesn’t have this section, the program is being sold rather than reviewed.
• ▸What Phase 2 looks like. Either expand to additional basins, expand to additional modules (Field Safety, Route Optimization), or both. The Phase-2 ask should be specific and trade-able.

If the metric is trending right, this readout is the document that authorizes Phase 2. If the metric is not trending right, the readout is what surfaces the issue early enough that the operator can either course-correct or stop the program before the next phase commits more spend. Both outcomes are healthy. The unhealthy outcome is the readout that papers over the metric and asks for more time.

Weeks 7–12, expansion

Phase 2 runs the same playbook against a different scope. Three patterns, depending on the operator:

Pattern A: another basin. Same module (Work Engine), same architecture, new asset base. Integration is faster the second time (3–4 weeks vs. the first 6) because the connectors and reconciliation patterns are already in place. The basin-specific scoring calibration takes most of the time.

Pattern B: another module. Same basin, new module. Adding Field Safety on top of the ranked work loop turns qualification enforcement and dynamic hazard scoring on. Adding Route Optimization tightens the daily route plan against geography and crew qualifications. Each module typically takes 3–4 weeks to stand up because the data foundation is already there.

Pattern C: M&A integration. For serial acquirers, Phase 2 is often the just-acquired asset coming onto the same ranked work loop. The reconciliation agent does the schema-harmonization work that an internal integration team would otherwise spend six to eighteen months on. The result: Day-1 ops visibility on the new asset, vs. the Day-180 visibility most operators settle for.

The Day-90 board readout

The deliverable to the board at Day 90 is not a deck about WorkSync. It is a before/after on the operator’s metrics: the deferred production captured, the trip-cost reduction, the TRIR trend, the payback period actually achieved. The board should be able to read it in five minutes and make the continued-investment decision in thirty seconds.

The structure that works:

• ▸Slide 1. What we said at Day 0. What we delivered at Day 90. Side by side.
• ▸Slide 2. The cash-flow line item, with payback period.
• ▸Slide 3. What changed in the field that caused the move. Operator quotes, ideally.
• ▸Slide 4. What we recommend next, with the budget ask. Year-2 BETTER tier expansion or full-platform commitment.

That’s a four-slide deck for a decision that lives with the operator for years. Most board members will not read past the first two if the numbers move.

What this rollout is not

It is not a 12-month enterprise pilot with a steering committee, a system integrator, and a milestone-review cycle. The reason mid-tier operators can run this in 90 days is precisely that the architecture is built around integration that ships on Day 7 and a paid-pilot framework that puts everyone in a yes/no decision in 60 days. The 12-month pattern is what super-major procurement processes require. It is also why super-majors take longer to compound the AI advantage. The mid-tier is fast precisely because it doesn’t have to satisfy a 200-person procurement committee for a $40K decision.

Chapter 12 closes the guide with the vendor landscape, where this fits, and where it doesn’t. The credibility chapter — written last so we earn the right to opinion.