One practice passed the AI pilot. Can the others meet the rollout date?

Find the dependency another build week won't fix

Our check compared drafts with partner-approved positions and linked each flag to supporting prior-deal material. That let lawyers inspect and address a standard-position issue before asking a partner to review the draft. The check passed its technical test, and the first group adopted it.

But two groups hadn’t finished approving their positions. Partners were handling closings, so approval work competed with immediate client work. We’d also proposed credit for build and adoption work against billable targets without checking when the compensation process could approve it. That decision sat outside our delivery window.

We also owned a technical gap. Two position types needed for credit agreements were missing, which caused the check to decline documents it needed to handle. During the remedy period, we captured the positions with a partner, obtained approval, and tested the change. The share of finance documents declined fell from 14% to 3%, but repairing that gap couldn’t authorize billable credit or complete every group’s approvals. That is why more engineering work didn’t get the rollout to 70%.

For your launch, separate missing technical coverage from missing firm decisions. The technical owner can repair a comparison rule. The practice partner has to approve the position, and the person with authority over targets has to decide how the work is credited. A date becomes credible only when those different owners can deliver what the next step needs.

Define the documents behind the launch percentage

In the example, Practice A wants to launch on November 16. Its check can handle standard purchase agreements, but the team discovers an amendment type with no approved positions. A promise covering all of A’s documents would now depend on work that hasn’t been completed.

A’s chair can narrow the initial scope to standard purchase agreements if that subset is worth launching on its own. Record the amendment exclusion before committing, give those documents a manual review route, and keep their volume visible beside the launch result. Otherwise, removing unsupported documents from the count could make the percentage rise while leaving the same work unresolved.

Define coverage as eligible documents that completed the check before first partner review, divided by all eligible documents first sent for partner review during the agreed measurement window. Count each document once. Include eligible documents that bypassed the check or were declined as failures to complete that route. A successful rerun before first partner review can count, with the earlier decline retained in the issue log.

For example, 32 checked purchase agreements out of 40 eligible purchase agreements gives 80% coverage. If 10 amendments were excluded in advance, show those 10 separately. That result supports a statement about purchase agreements, not a claim that 80% of all the practice’s documents used the check.

A’s narrow scope gives the chair a concrete decision about where to use partner time. The larger investment decision still starts with the firm’s goal, because a rollout that uses more documents only matters when it improves the work that limits that goal.

* Eight-week delivery, or twelve weeks when substantial data foundations are required, as agreed before work begins. If a covered stage fails its agreed benchmarks, we provide up to four additional weeks of remedy work at no charge, then waive or refund that stage’s fee if it still fails. Successfully delivered stages remain payable. Agreed client-caused delays pause the clock. Delivery guarantee terms.

Put the owners and decision dates beside each practice

Now carry the same example into a launch register. The dates, hours, and targets below are illustrative planning choices. The practice chair, knowledge owner, technical owner, and person who controls work targets should fill in the record together, because each can confirm a different dependency.

Practice B’s technical team could keep building while waiting, but it couldn’t decide which unapproved positions should govern new work. And an agenda slot doesn’t guarantee that the compensation decision will pass. Promising a launch immediately after that meeting would skip both the decision itself and the time needed to test and introduce the approved change.

Instead, B’s chair can choose an earlier authorized decision, a smaller independent scope, or a later commitment. If none is available, retain the current review process and stop treating more build effort as the solution to that delay. A shared spreadsheet is enough to manage this decision if it makes the dependencies and owners clear. You don’t need another AI tool to resolve approval authority.

If the positions themselves aren’t usable yet, our approved drafting positions guide shows how to capture the rule, its source, and its exceptions. Once those inputs exist, the separate handover exercise tests whether the receiving team can change, evaluate, and recover the workflow.

Agree what counts as delivery before the date arrives

A coverage target measures whether work used the route. It doesn’t establish that the check was correct or that partner effort fell. Before A commits, its chair and technical owner need acceptance tests for the actual purchase agreements, including departures from the standard, valid exceptions, and missing information. Keep the source permissions and lawyer decisions required by the practice in that test.

Then measure whether the route adds checking work. Compare active associate checking and partner correction time on comparable documents, and keep waiting time separate. If A reaches 80% coverage by asking partners to redo every check, the use target can pass while the intended review improvement fails. Separate measures let the chair see which part needs repair.

Give the operating owner a document log with the practice, document ID and type, eligibility reason, first partner-review time, check completion time, decline or bypass reason, and check version. The owner can then reconcile the coverage count against the actual review queue instead of relying on logins. Keep client material in the firm’s approved systems and use permitted references in the register.

Before launch, name who reviews the evidence, the deadline, and the action for a missed benchmark. Record how changes to scope or dates require agreement, and preserve the original commitment when recording a revision. If a technical gap prevents the agreed work, assign a repair and re-test. If a required firm decision is still missing, escalate it to its owner and reassess the commitment under the agreement.

In our engagement, later billable-credit approval and more completed position sets helped coverage reach 74%. That recovery didn’t change the 52% delivered at the end of the remedy period or undo the refund. Keeping both checkpoints visible prevents a later improvement from rewriting what was delivered on time.

Copy one record for each practice

Use the following blank structure in your existing project register. The practice chair owns the commit, narrow, or defer decision. Each dependency owner supplies the evidence for their field, and the operating owner keeps the document counts current.

Keep an unknown as an unresolved dependency with an owner. If A loses its booked reviewer to a closing, reopen its row before launch. The positions may still be approved, but no one is available to finish validation or handle exceptions, which changes whether the agreed route can start. A substitute with authority and protected time can restore readiness. An unchanged date on the register can’t.

Make the next commitment with the person who can unblock it

Take the next practice due to launch and complete its row with the chair. Identify the earliest unresolved dependency that prevents the next step, then take that specific decision to its owner. Leave with a supported commitment, a smaller agreed scope, or a deferred launch with a named review condition. That is the usable outcome of the meeting.

A credible rollout protects partner time by exposing missing approvals before the team plans around them. If your larger goal is higher partner profit or growth over the next 3–18 months, assess that rollout alongside the other demands on the firm. Its priority depends on what most limits the goal and how the released capacity would change the work. Our core engagement starts with that wider decision, then carries the agreed implementation and transfer.

* Eight-week delivery, or twelve weeks when substantial data foundations are required, as agreed before work begins. If a covered stage fails its agreed benchmarks, we provide up to four additional weeks of remedy work at no charge, then waive or refund that stage’s fee if it still fails. Successfully delivered stages remain payable. Agreed client-caused delays pause the clock. Delivery guarantee terms.