Your Outsourced Project Is Late: A Three-Question Decision Tree, What Penalty Clauses Really Recover, and a 45-Day Playbook

The Contract Said June 30. Today Is August 10.

You signed a NT$1.8 million contract for a custom order management system. Payments have reached 70% of the total, meaning NT$1.26 million has already left your account. The weekly report has said "overall progress 85%" for seven weeks running. Marketing has a fall campaign locked for September 15, and two of your staff have spent six months doing nothing but cleaning legacy data for this system. The question in front of you is not whether to lose your temper. It is three concrete decisions: whether this money is recoverable, whether to switch vendors, and at what point you cut your losses. Here is a process you can run in one week.

Four Things Most Owners Get Wrong About Delays

• Wrong assumption one: a delay is the vendor's fault. PMI's 2021 Pulse of the Profession, based on 3,950 project professionals, found that globally only 55% of projects finish on time, 34% experience scope creep, and 9.4% of invested funds are wasted on poor project performance. Scope creep is the single biggest killer of on-time delivery, and scope is almost always something both parties grew together.
• Wrong assumption two: a penalty clause protects you. Taiwan's public procurement convention sets late-delivery liquidated damages at one-thousandth of total contract value per day, capped at 20% of the contract (see the Taichung City Construction Bureau's schedule-overrun penalty procedure). Applied to that NT$1.8 million contract, 30 days late yields NT$54,000. On top of that, Article 252 of Taiwan's Civil Code lets courts reduce liquidated damages they consider excessive. A penalty clause is negotiating leverage, not a recovery instrument.
• Wrong assumption three: percentages measure progress. The "85% syndrome" exists because the percentage is self-reported by the builder and no acceptance definition exists. The only meaningful metric is features that have passed your acceptance test divided by total contracted features.
• Wrong assumption four: switching vendors gives you a clean restart. Under Article 511, a client may terminate a work contract at any time before completion, but must compensate the contractor for losses arising from that termination, which courts have read to include payment for completed work plus the profit forgone on the unfinished portion. Walking away is not the same as not paying.

The Three-Question Delay Decision Tree

Do not start with "should I fire them." That question is too large to answer. Ask three smaller ones in order and the answer surfaces on its own.

• Question one: is this a single slip or a trend? Pull the last three weekly reports and compute, for each, milestones actually accepted divided by milestones promised. Three consecutive periods under 70% is a trend, not an accident.
• Question two: how is the responsibility distributed? Tabulate three things across the last eight weeks: how many times requirements changed, how many days the vendor waited for your answers, and how many days your assets or account credentials arrived late. Compute client-caused delay days divided by total delay days. Above 40%, any vendor would be equally slow.
• Question three: does remaining value still exceed rebuild cost? Use this formula: Rebuild cost S = (amount already paid that cannot be reused) + (catch-up hours for a new team × day rate) + (operational loss during the switch). Remaining value R is what the system can still earn or save you inside its commercial window.

Four exits follow. If the trend holds, client-caused delay is under 20%, and R exceeds S, switch vendors. If client-caused delay exceeds 40%, fix your own side first: freeze scope, name a single decision owner, rebuild the milestone plan. If this is a single slip and the vendor can produce a recovery plan with testable deliverables, sign a recovery memorandum with a 30-day observation window and an exit clause. If R is below S and the commercial window has already closed, stop the bleeding: settle, reclaim your assets, bridge with off-the-shelf SaaS.

One Decision Tree, Three Companies, Three Answers

• 15-person e-commerce brand (custom OMS, NT$1.8M, 8 weeks late). The campaign date is the real deadline and R is concentrated entirely before September 15. The answer is not to switch vendors but to cut scope to protect launch: keep ordering, inventory and fulfilment; move returns and reporting to phase two. If the contract states that completion by a specific date is an essential element, Article 502, paragraph 2 allows rescission plus damages, which is the strongest card you hold in a scope-cutting negotiation.
• 80-person traditional manufacturer (ERP rollout, NT$6.5M, 5 months late). Running the numbers shows 45% of delay days are client-caused: three departments each send a different manager, and every meeting produces a different conclusion. The answer is to fix yourself first. Appoint one decision owner, freeze requirements for 60 days, and rewrite acceptance criteria as executable test cases.
• 5-person SaaS startup (mobile app, NT$900K, 12 weeks late). The vendor's lead engineer resigned and nobody remaining can read the codebase. Bus factor is one, the trend is unambiguous, and client-caused delay is under 15%. The answer is to switch, and to switch while the code is still legible.

What One Month of Delay Actually Costs You

Using the 15-person e-commerce case, here is a month of delay itemized:

Compare that with the NT$54,000 in liquidated damages accruing over the same month: a ratio close to nine to one. This is why delay is not a problem you solve with penalty clauses. It is a problem you solve with decision speed. Every extra month costs you roughly half a million in cash flow and a full quarter of market position, not the delta on a penalty calculation.

Vendor Health Scorecard: Ten Dimensions

Score each item 0 (absent), 1 (partial) or 2 (fully in place), for a maximum of 20. 14 and above is recoverable. 11 to 13 means an observation window. 10 or below means start evaluating a replacement.

• Milestone acceptance rate over the last three reporting periods
• Whether deliverables are operable (a staging environment versus screenshots)
• Whether you can see commit history and commit frequency in the repository
• Whether every requirement change goes through a written change order with re-estimated hours
• Median response time to messages (deduct above 24 hours)
• Key personnel stability, specifically whether the bus factor is greater than one
• Whether automated tests and defect regression records exist
• Whether the repository, cloud accounts and domain are registered under your company
• Gap between payment progress and accepted-value progress (deduct above 20%)
• Whether bad news is volunteered or only surfaces when you ask

How We Take Over, and the Cases We Decline

ScriptWalker offers four engagement models for delayed-project rescue. Technical due diligence costs NT$45,000 over five working days and produces a takeover feasibility report, catch-up hour estimate and asset gap list, with no obligation to continue. Rescue project works on frozen scope, fixed price and one testable deliverable every two weeks. Advisory oversight is a monthly retainer where we write no code and instead review your existing vendor's deliverables and timesheets, which suits owners who want to keep the incumbent. Post-takeover retainer covers ongoing maintenance once the system is stable.

In these situations we will tell you to hire someone else:

• The matter has moved to lawyers or litigation and what you need is an expert appraisal report. We are an engineering team, not an appraisal body.
• The repository and cloud accounts are not under your name and the incumbent flatly refuses to hand them over. Without assets there is no takeover, only a rewrite.
• You expect us to cover six months of the incumbent's work in two weeks. Catch-up effort typically runs 30% to 50% of the hours already sunk. That is a physical constraint, not a pricing position.
• Your available budget is below 60% of rebuild cost. A rescue that runs out of money halfway does more damage than no rescue.
• Requirements are still unfrozen but you want a fixed price and a fixed date. That combination is exactly what caused the first delay.

The 45-Day Stop-the-Bleeding Playbook

• Days 1 to 7: secure the assets. Obtain repository owner rights, cloud and domain administration, a full database backup and source design files. In parallel, issue a written scope freeze notice and name one decision owner.
• Days 8 to 21: establish the facts. Run technical due diligence, complete the scorecard above, calculate R and S, and put three options on the table with the incumbent: recover, assist with handover, or terminate by agreement.
• Days 22 to 35: re-contract. Sign either a recovery memorandum or a transition contract. Rebuild milestones as one testable deliverable every two weeks, tie every payment to acceptance, and write in an exit clause.
• Days 36 to 45: see the first result. The first acceptable build must reach staging. If it does not, execute the exit clause and grant no further extension.
• Day 90 review. Re-run the scorecard and compare. If the score has not improved by at least four points, the recovery has failed.

Self-Assessment Checklist

• ☐ I can state the ratio of accepted features to total contracted features
• ☐ I have the last three weekly reports and their milestones are testable, not percentages
• ☐ I have calculated client-caused delay days as a share of total delay days
• ☐ I know the gap between amount paid and value accepted
• ☐ The repository owner account belongs to my company
• ☐ Cloud hosting, domain and payment gateway are all registered to the company
• ☐ I hold a full database backup taken within the last 30 days
• ☐ The contract states that completion by the specified date is an essential element
• ☐ Payments are tied to staged acceptance rather than monthly or percentage instalments
• ☐ I know the exact date the commercial window for this system closes
• ☐ I can calculate the true cost of one month of delay
• ☐ I have issued a written scope freeze
• ☐ The project has exactly one decision owner
• ☐ I know who the vendor's key personnel are and whether they are still employed

Frequently Asked Questions

The vendor has said "just two more weeks" three times. Should I believe the fourth?

Do not decide based on belief. Decide based on testable output. Reply with a specific question: at the end of these two weeks, which three features will I be able to operate in staging? Put the commitment in writing, tie it to acceptance, and state that a handover process begins if it is missed. Once promises become concrete, most teams revise their own forecast, because they know an inspection follows.

My contract has no penalty clause at all. Do I still have leverage?

Yes. Under Article 502 of the Civil Code, where completion is delayed for reasons attributable to the contractor, the client may demand a reduction in remuneration or compensation for delay damages, and may rescind the contract entirely if the deadline was an essential element. Your more practical leverage is the unpaid balance. The moment you convert remaining payments to acceptance-first, your negotiating position reverses.

Can I simply terminate and withhold the final payment?

The two routes have very different consequences. Terminating under Article 511 requires no reason but obliges you to compensate the contractor for losses arising from termination. Acting under Article 502 requires you to prove the delay was attributable to them. Gather evidence before choosing a route; the cost of documentation is far below the cost of compensation.

Will switching vendors make things slower?

It will add time, but usually less than you fear. Technical due diligence takes about five working days, and catch-up effort typically runs 30% to 50% of hours already sunk. What actually destroys your schedule is six months of "two more weeks" repeated four times, not the switch itself. The deciding factor is the comparison of R against S, not frustration.

Should I keep paying while the project is late?

Do not stop paying. Restructure instead. Split the remaining balance so that each testable two-week deliverable carries its own payment. A full stop gives the vendor grounds to down tools and weakens your legal position; acceptance-first payment preserves both your cash and your leverage.

What to Do Next

If your project is more than four weeks late, the most expensive option available to you is to keep waiting. We offer a free 30-minute online diagnostic. Bring your contract, the last three weekly reports and your payment record, and we will run the three-question decision tree with you on the call and tell you whether this is a recovery, an oversight problem or a stop-loss. If you want to go deeper, technical due diligence follows, and the report is yours to keep with no obligation to hire us for the build.

• Email: [email protected]
• Phone: 0916-224-047
• LINE: @ufv9089p