How Operations Leaders Decide to Fix, Switch, or Insource Underperforming Vendors

Early in building our robotic handwriting system, we leaned on outside vendors for a few key components rather than building everything ourselves, mostly because we were small and self funded and could not do it all at once. One vendor kept missing delivery targets on a part that fed directly into our production line, and each miss meant customer orders slipped further behind.
The trigger that made the call clear for us was simple. We asked whether the miss was a one time problem or a pattern, and we tracked it for exactly two cycles before deciding. One miss can be bad luck, a supply chain hiccup, something outside anyone's control. Two misses in a row on the same commitment told us it was a pattern we could not fix with a phone call and a promise to do better. At that point we made the decision to bring the work in house rather than keep switching vendors and hoping for a different result.
That single rule, one miss you watch, two misses you act, kept us from either overreacting to a bad week or waiting too long while customers felt the delay. Bringing that piece in house also ended up being one of the better decisions we made, because it gave us more control over quality and timing than any vendor relationship could have. Sometimes the fix is not a better vendor, it is not needing one at all.
Rick Elmore, Founder/CEO, Simply Noted (simplynoted.com)

My rule is simple. If a partner can't give us what our customers need and it's slowing us down, we stop waiting and build the piece ourselves, if we can do it cheaply enough to maintain.

As a digital insurance broker, Eprezto depends on insurers. Some of them have no API, so there is no way to connect their system to ours. For renewals, that meant someone had to go and get the renewal premium by hand. We didn't wait for those insurers to modernize. We built bots that log into the insurer's portal with our own credentials and pull the renewal premium for us. It saved a lot of manual time and made renewals faster and more reliable.

The other side of that decision is knowing when to drop partners instead. We used to show eight insurers on our site. Every one connects differently, with its own quirks and its own maintenance, and that eats engineering time. We cut it to between four and five. Fewer connections, less to maintain, and customers decide faster.

So the trigger for me is maintenance cost versus value to the customer. If a partner is worth keeping but the connection is the problem, fix it in-house. If the partner isn't adding enough to justify the upkeep, let it go.

The trigger that made the call clear for us was the second miss on the same specific commitment, not a second miss overall. A vendor missing two different things in two different ways over a quarter is often just normal variance in a working relationship, and remediation conversations are worth having. A vendor missing the exact same commitment twice, after we already had the fix-it conversation the first time, told us the first conversation did not actually change anything, whatever they said in it.

We went through this with a support-tooling vendor whose SLA response times slipped, we raised it, they promised a fix, and three weeks later the same metric slipped again in the same way. That repeat was the signal, not the original miss. We switched providers within the month rather than having a third conversation, because a vendor who cannot fix the thing they already agreed to fix once is unlikely to fix it after a second promise. Bringing it in-house never made the shortlist for us since the tooling itself was standard, just the vendor's execution on it was not.

The trigger I use is two consecutive misses on the same metric after a written fix plan. One miss gets a conversation. A second miss gets a written plan from the supplier with a date and one number. If the same number is missed again after that plan, the decision is already made, and the only remaining question is switch or insource.

The case was a content vendor writing French articles for a client's site. Deadlines slipped, then quality slipped, and each time there was a reasonable explanation. After the second miss we agreed a plan in writing: eight articles a month, delivered by the fifth, with a native review before delivery. The next month they delivered five, on the eleventh. The rule made the call for me before the client noticed, because the client only ever saw articles going live on the agreed schedule, filled in by our own team that month.

Switch or insource depends on whether the work repeats. This work was monthly and identical in shape for several clients, so we hired a French writer instead of finding another vendor, and the recurring risk disappeared. If it had been a one-off project, we would have switched suppliers. The value of the two-miss rule is that it removes the hope from the decision; suppliers who are going to improve improve after the written plan, and the ones who are not become visible in exactly one month.

If there have been two occurrences of a supplier's problem even after putting a corrective measure into effect, then the second occurrence will be considered a supplier's decision rather than an error or an accident. At LAXcar, this means reducing the work a supplier does, replacing them with another supplier, and doing more ourselves before the client feels the effects of the problem.

One supplier kept missing staging times for their vehicles while handling event transport requests. After the first occurrence, a discussion about the reporting time, driver communication, backup, and escalation took place. In the next assignments after a similar occurrence, we decreased the amount of work assigned to that supplier before any complaints came from the client. I tend to look at the leading indicators - late staging, late confirmations, and no-shows rather than excuses. One incident can be remedied. Two incidents after that mean an unreliable system.

I watched a 3PL partner miss their 99% accuracy target three months in a row while running my e-commerce brand. The trigger that forced my hand wasn't the metrics themselves--it was when their excuses shifted from "we're fixing this" to "industry standards are actually 97%." That's when I knew they'd stopped trying.

Here's my rule: You get one miss to explain what broke. You get a second miss to show me the fix is working. If there's a third miss, I'm already talking to their replacement because they've shown me who they are. The mistake most founders make is waiting for customers to complain before acting. By then you've burned trust you can't buy back.

When I built my fulfillment company to $10M, we had a packaging supplier who kept delivering corrupted boxes. First month, machinery issue--fine. Second month, they blamed their upstream supplier. I gave them 30 days to fix it or I'd split their volume with a backup. They didn't believe me. I moved 60% of our business in week five.

The decision tree is simpler than people think. Can they fix the root cause in under 60 days? If yes, remediate with clear milestones. If no, switch providers immediately. Bringing work in-house only makes sense if you're dealing with a capability gap across all providers or the work is core IP.

The single clearest trigger for me is the accountability shift. Partners who own mistakes and show you their corrective action plan are worth keeping. Partners who rationalize, deflect, or reset expectations downward have already quit on you--they just haven't told you yet. I've seen too many brands lose customers because they were loyal to vendors who stopped being loyal to quality.

At Fulfill.com, we see brands switch 3PLs after 18 months on average. The ones who wait longer almost always say they wish they'd moved sooner. Your supplier's performance problem becomes your customer's experience problem faster than you think.

One missed target is a problem; a missed recovery commitment is a decision. I first separate capacity, communication and capability. If the supplier still has the capability and alerts us early, we agree on a written recovery plan covering upcoming orders, quantities, delivery dates and an escalation contact. If they miss a customer-critical milestone in that plan without warning, I begin shifting volume before a customer promise breaks. I bring work in-house only when it is recurring, strategically important and capable of keeping the required equipment and skills productively used. Otherwise, an alternate supplier preserves more flexibility. The decisive trigger is failure to escalate inside the agreed risk window. Late material can sometimes be managed; late information cannot. Volume can be transferred gradually, giving the supplier a fair opportunity to recover while backup capacity protects customers.

In order to make the decision to either fix, relocate, or bring an unsuccessful partner in-house, it is important to distinguish whether the problem concerns a lack of capacity or abilities. Personally, having managed global delivery teams for the past 20 years, I can confidently say that capacity issues are mostly fixed through enhanced planning and improved cooperation while capability issues prove persistently impossible to solve through additional supervision. If it takes more time to manage the work done by the partner than to do it ourselves, it is already evident that the partnership has lost its value.

To avoid waiting for the negative effects of the partner's failure to reach our client, I have a simple criterion, which I call the Velocity of Correction. As soon as the partner fails to meet the required benchmark, we provide them with a remediation plan and set a relatively short deadline for improvement; in case they admit their mistake but time to implement the changes exceeds our operational cycle, they become a liability instead of an asset. If a system fails to meet us two times in a row regarding one particular type of a failure, it is immediately decided whether to switch the supplier or do the work ourselves.

Waiting to find out that a customer complains about poor service is something that the companies dealing with clients cannot afford. At the moment a negative dip in quality of service becomes noticeable to the customer, all the costs incurred from the failing supplier have already accumulated.

A supplier relationship is worth developing, not discarding at the first sign of trouble. My starting point is to understand the cause of a failure and work with the supplier on a realistic correction.

At Eurodita, we manufacture timber building kits and timber joinery in-house while buying selected components, including insulated glass units and hardware. One component can affect the readiness of an entire kit, so supplier development and customer commitments have to be considered together.

I would give a supplier the opportunity to recover, with clear expectations, a named owner and an agreed checkpoint. The important question is whether performance is improving and the supplier is open about what remains unresolved.

My limit would be reached when there is no longer a credible, verifiable recovery plan that protects the next customer commitment. At that point, I would move to a qualified alternative. Bringing production in-house requires a separate assessment of capability, capacity and investment—not simply frustration with a supplier.

The principle is patience with genuine improvement, but a firm boundary when continued patience puts the customer at risk.

My rule is that I bring work in-house when the thing being outsourced is something my clients think they're buying from me. Everything else I'm happy to remediate or switch.

I've run Green Planet Cleaning Services in the Bay Area for 16 years. The clearest version of this for us was subcontracting. Early on, when demand spiked beyond what my own crews could cover, the industry-standard answer was to hand overflow jobs to independent contractors. It's how a lot of cleaning companies scale, and on paper it's elegant — you only pay for the work you get.

What actually happened was that the quality of a job stopped being something I controlled. A client doesn't experience "our subcontractor had a bad week." They experience Green Planet did a bad job. And the failure showed up on the client's side before it ever showed up on mine, which is the worst possible sequence — by the time I knew, someone was already deciding whether to keep us.

That was the trigger: the vendor's miss was invisible to me and visible to my customer. Once that pattern repeated, remediation was pointless, because the problem wasn't effort, it was that I had no ability to see quality before the client did. So we went the other direction and built a W-2 employee model. My people are hired, trained, paid, and accountable to me. It costs more per hour than contracting, without question. It also means the product my clients buy is the product I actually make.

The general rule I'd offer: outsourcing is fine where a miss is recoverable and invisible to the end customer, and dangerous where a miss becomes your reputation. Switch vendors for the first kind. In-source the second kind, and accept the margin hit as the price of controlling your own brand.

The signal I watch for is who tells me about the problem. If I'm consistently learning about failures from my clients rather than from my vendor, the relationship can't be fixed by a scorecard, because the visibility gap is the problem. That's my line, and it's saved me from long, polite improvement plans that were never going to work.

When a key tooling vendor or freelance partner repeatedly misses targets, the expensive mistake is waiting until a client launch fails and then making a panicked switch. Our rule is written remediation first, then a hard trigger. After the second consecutive miss on a dated weekly deliverable that the client path depends on, following a one-page fix plan with an owner and a deadline, we switch providers or bring the slice in-house while we still have a week of cover. Misses that would land on a live account do not wait. That trigger made the call clear on a reporting export partner who needed chasing every Friday for product-level data we had already scoped. Continuity mattered more than sunk familiarity. A supplier who needs chasing every Friday is the same cost as a dormant seat. Remediate once in writing. Switch before the customer feels the gap.

My rule is simple. If a miss would show up before a guest ever checks in, that vendor gets one shot to fix it, not three. Turnover cleaning runs on a tight clock. A booking starts in an hour, there's no time to remediate after a guest is already inside.

First miss, I ask if it's a training gap or a capacity gap. A training gap gets a clearer checklist and a second try, because that fix is fast. A capacity gap means the vendor is stretched, and a stretched vendor stays stretched next month too.

The trigger that ends coaching and starts switching is the same task getting missed twice, in two different homes. That tells me it's the process, and coaching a process rarely holds.

Bringing it in-house only makes sense once volume is steady enough to justify owning it, and the miss is something a guest would notice fast, like restocking or timing gaps. Occasional misses stay with a managed vendor relationship instead.

The goal is catching this before a review does. Once a host learns about a miss from a guest complaint, the internal trigger already fired too late.

When a supplier repeatedly misses, we remediate once with a written service target, then switch if the book path still breaks. Scheduling or payment tools that fail the $47 capture before a 60-minute hold on The Functional Medicine Process: What to Expect get a same-week fix window. No Business Associate Agreement is not remediable with charm. Relationship stays civil. Patients do not absorb vendor drama during camera-on.

When a key partner repeatedly misses live-file import or SLA targets that put Monday closings at risk, we switch providers rather than run another apology cycle. The trigger that made the call clear was simple: if open deals cannot import into a usable live file inside the Getting Started window of under 1 week, remediation ends and the work moves. Customer impact shows as stalled Monday files, so we do not wait for a third miss after closings already slipped. For the 1,700+ brokerages on the product, the file is the system of record. Partners who break that path lose the slot before brokers feel a second week of blocked deals.

When a key supplier for London pack-out repeatedly misses dated handoffs, I remediate once with a written new date and a named owner, then I switch or dual-source before customers feel empty carts.

The single trigger that made the call clear was the second consecutive miss that would break ship windows for a few hundred customers a month. We do not wait for a sold-out weekend and a flood of day-3 emails. One written fix window. If the carton still cannot meet the next pack-out slot, volume moves to a backup. Relationship stays civil. The shelf stays honest. On a small DTC team that rule beats hoping the partner suddenly finds capacity after ads are already live and inventory is already promised.

I use a two-miss rule on cable and connector suppliers before I switch.
If a key partner misses the agreed ship window twice in a row on the same SKU family, I open a second source and move new purchase orders across before stock hits zero. Remediating once is fine. Remediating after two misses just burns the buffer customers feel as "out of stock" on Type 2 or tethered leads. The trigger is operational, not emotional, so the call is made before the storefront shows empty.

The call on a failing supplier gets made late because the trigger is a feeling rather than a threshold. Missed targets accumulate, each one individually forgivable, until a customer feels the impact. In my practice the threshold has to exist before the pressure does.

Write the trigger into the agreement and into the internal scorecard. Two consecutive periods below the service standard opens a remediation plan with a fixed end date. Failure to clear remediation by that date moves the volume, with no further discussion. The rule removes the negotiation that keeps a weak supplier in place.

Choosing between switching and insourcing turns on whether the capability is differentiating. Rule-governed work goes to another provider, and work carrying institutional context comes inside instead. Set the threshold now, and the call makes itself before customers ever feel the miss.

My trigger is when a supplier problem starts threatening our response time or vehicle availability. In towing, a late part or unreliable external service can take a truck off the road, and that quickly becomes a customer problem rather than an internal inconvenience. I will usually give the provider a chance to correct the issue if communication is good and the failure is isolated. But once the same problem becomes predictable, I start lining up an alternative before service quality suffers. Loyalty matters, but reliability matters more when customers are depending on you during accidents and breakdowns.

Repeated misses without a communicated root cause is the trigger. If a partner can't tell me why something failed and what specifically changes next time, the miss is almost guaranteed to repeat.

At 3D Studio we had a rendering partner fall behind on three consecutive project cycles. Each time there was an explanation, but no structural change behind it. By miss two I had already started parallel-testing a backup. By miss three we had transitioned 80% of that work before a single client deadline was broken.

The remediate-vs-switch question comes down to one thing: switching cost vs. compounding cost. Remediation makes sense when the failure is isolated, the partner has margin to absorb the fix, and the relationship has data showing it's anomalous. Switch when the pattern is the relationship, not the incident.

Bringing work in-house is a separate and harder call. We've done it only when the volume justified a fixed cost and when the external dependency was showing up repeatedly in client conversations as a trust issue. Those two conditions together, not either one alone.

One rule I've held to: never let a partner's third miss be the first time they hear it's a problem. Feedback at miss one, a documented threshold at miss two, and a decision already made before miss three arrives. By the time customers feel it, you've already lost the window to act cleanly.

TKEG Expat is a corporate-services firm managing 120 companies across 22 jurisdictions, and almost all of our delivery rides on outside suppliers. Our single trigger is a missed statutory filing date. It is a government date nobody in the chain can renegotiate, which is exactly why it can not be walked back, and why it turns a remediation into a switch.

Because we hold no supplier-side contract to appeal to, the trigger has to be a government date instead of a service level. Every agreement on file is client-side, with no milestone, acceptance or payment-trigger field. Our operations system carries 279 dated obligations across 56 of the managed companies, so a supplier going quiet becomes visible early, as a date that moves. However, the date itself gets no second attempt. Ireland's Taxes Consolidation Act adds a section 1084 surcharge to a late corporation-tax return even though the tax may have been paid in full and on time, and section 363 of the Companies Act 2014 costs 2 financial years of audit exemption for a late annual return, unless it is the company's first one.

Because roughly two-thirds of what a client pays for a company setup is supplier and government pass-through instead of our margin, switching is normally the one we take, and our 141 suppliers cover 79 jurisdictions with overlap in the common lines. We only bring work in-house where we hold the registration ourselves, and only 24 of our 2,164 project line items have no supplier.

I don't decide based on the misses. Every partner misses eventually. I decide based on who finds the miss first.

If the partner tells us before we notice, with a cause and a plan, that relationship is fixable. They can see into their own operation, and they're willing to be honest about it. So we remediate: tighter targets, shorter check-ins, and a clear date to reassess. But if we keep discovering problems ourselves, in a customer ticket, a bad report, or a missed delivery, the partner either can't see their own failures or won't admit them. Neither gets fixed with a stern meeting. That's the trigger to switch.

The rule became clear with a data vendor at an earlier company. Their error rate wasn't worse than a competitor we'd used before. The difference was that the first three problems all surfaced through our support team, not their account manager. By the time they explained, customers had already felt it. We started a switch that month. The transition was painful but short, because we'd stopped waiting for a promise to fix a visibility problem.

Bringing work in-house is a separate question, and I'm much stingier with it. I only do it when the work sits in the part of the product customers judge us on, and I've already switched providers once for the same reason. If two outside partners can't get it right, it may be too close to our core to hand to anyone.

My advice: add one line to every partner review. Who found each issue first, us or them? After two quarters, the answer makes the decision for you.

When a supplier repeatedly misses targets we first attempt remediation by clarifying ownership for each artifact and enforcing a written definition of ready before the next handoff. In my work I focus on naming one person against each artifact and confirming a dated final export so there is no ambiguity at launch. The single trigger that made the call clear was when a partner could not point to a named owner and a confirmed final export for the at-risk artifact. If that condition could not be met, we moved the work to a different provider or brought it in-house to remove the risk to customers.

A key leave-in supplier that slips restock dates twice in a row on a hero jar gets one written remediation window. Miss again and we switch or pause the listing before customers hit empty carts.

I do not wait for a sold-out weekend to decide. The carton date is the trigger. In The UK Wash-Day Report 2026, https://zenvy-beauty.com/blogs/news/uk-wash-day-report-2026, wash days sat 4.8 days apart. Supply that cannot keep that rhythm becomes a customer problem fast.

The clearest trigger is not the missed target itself, but whether the partner identifies the problem before the customer discovers it. We have found that operationally mature suppliers surface a constraint early, explain its effect on specific orders, and offer realistic alternatives. Those behaviors preserve options, even when performance is imperfect.

A partner should be remediated when its data is trustworthy and its corrective actions can be verified quickly. Switch when updates arrive only after repeated chasing, because hidden uncertainty destroys planning. Bringing work in-house is justified when the required knowledge is highly specific, such as coordinating dimensions, finishes, packaging, and customer expectations, and outside teams repeatedly lose that context.

A missed date does not decide it for me, being told late does. I sit on both sides of this, because a clothing manufacturer is a supplier to its brands and a customer of its mills and contractors, and my rule is that a supplier who calls me the day a delivery slips stays, while one who lets me find out at the cutting table does not. The first kind costs me a week I can plan around. The second costs me a run, because the cut is scheduled, the operators are booked, the fabric is not there and the client hears about it from me instead of the other way round. That is the same standard I am held to, and it is a large part of how we held 97 percent on time delivery from 2021 to 2025. The honest part is that switching costs a re test on the fabric and a delay of its own, so remediating is almost always cheaper, and I have kept a supplier longer than I should have twice for exactly that reason. What I ask a new supplier now, before the first order, is who calls me when something slips and how quickly.

Q&A

How Operations Leaders Decide to Fix, Switch, or Insource Underperforming Vendors

Q&A

How Operations Leaders Decide to Fix, Switch, or Insource Underperforming Vendors

Act on Two Identical Failures

Balance Upkeep Against Customer Value

Replace Vendors After Recurring Commitment Breach

Demand Written Plans, Then Bring Work Inside

Cut Allocation After Corrective Relapse

Judge Accountability, Not Excuses

Transfer Volume Upon Silent Milestone Breach

Gauge Correction Speed Against Operational Cycles

Require Verifiable Recovery to Guard Promises

Control Brand-Defining Services Directly

Impose a Fix Deadline With Cover

Separate Training Gaps From Capacity Constraints

Safeguard Patient Holds Through Rapid Repairs

Relocate Live Files at First Import Block

Dual-Source Dated Pack-Out Handoffs

Activate Backup Supply on Repeat SKU Delay

Codify Service Thresholds in Advance

Preempt Vehicle Downtime Through Alternate Providers

Parallel-Test Alternatives Amid Root-Cause Silence

Treat Statutory Date Lapses as Disqualifiers

Track Who Detects Issues

Assign Named Owners to Each Artifact

Pause Hero Listings When Restocks Falter

Favor Proactive Disclosure Over Chased Updates

Retain Suppliers Who Warn Early

Related records

Standing brief

Q&A

How Operations Leaders Decide to Fix, Switch, or Insource Underperforming Vendors

Q&A

How Operations Leaders Decide to Fix, Switch, or Insource Underperforming Vendors

Act on Two Identical Failures

Balance Upkeep Against Customer Value

Replace Vendors After Recurring Commitment Breach

Demand Written Plans, Then Bring Work Inside

Cut Allocation After Corrective Relapse

Judge Accountability, Not Excuses

Transfer Volume Upon Silent Milestone Breach

Gauge Correction Speed Against Operational Cycles

Require Verifiable Recovery to Guard Promises

Control Brand-Defining Services Directly

Impose a Fix Deadline With Cover

Separate Training Gaps From Capacity Constraints

Safeguard Patient Holds Through Rapid Repairs

Relocate Live Files at First Import Block

Dual-Source Dated Pack-Out Handoffs

Activate Backup Supply on Repeat SKU Delay

Codify Service Thresholds in Advance

Preempt Vehicle Downtime Through Alternate Providers

Parallel-Test Alternatives Amid Root-Cause Silence

Treat Statutory Date Lapses as Disqualifiers

Track Who Detects Issues

Assign Named Owners to Each Artifact

Pause Hero Listings When Restocks Falter

Favor Proactive Disclosure Over Chased Updates

Retain Suppliers Who Warn Early

Related Articles

Related records

Standing brief