- Outsource OSP design to reduce project cost against a same-scope baseline
- Buy queue capacity, not the average drawing rate
- Freeze design-ready inputs before anyone prices volume
- Keep change visible so the external team cannot become a memory hole
- Choose when to outsource OSP design to reduce project cost
A finance lead will often treat the drafting rate as the whole decision. That number is easy to put in a cell. It is also the part of OSP cost that moves least once a permit reviewer starts marking exhibits. The hours that follow a changed pole-owner ruling never sat in that cell.
The comparison that actually changes spend has to price accepted design, not sheets produced. Source repair belongs in it. Exception rulings the owner still has to make belong there too, plus the exhibit revisions that follow a route that moved after the first submittal. Drop those and the file is a rate sheet wearing a make-or-buy title.
We start with a same-scope baseline so both columns describe the same outputs. Then we price the queue the designers actually see, which arrives as dumps rather than a staffing-model week. After the inputs freeze, every later change still has to be readable when a different engineer opens the native file.
Outsource OSP design to reduce project cost against a same-scope baseline
To outsource OSP design to reduce project cost is to buy defined engineering output from an accountable provider and compare total delivery cost against the internal alternative over a stated demand period. The United States Government Accountability Office Cost Estimating and Assessment Guide, GAO-20-195G, published March 2020, supplies the 12-step frame for that comparison.
GAO-20-195G publishes no OSP prices. It still defines what a reliable estimate has to show once a buyer asks which route miles the quote actually covers. Documentation is the first test most make-or-buy files fail, because the quote never states which route miles or pole quantities the price covers. We refuse to argue a rate until that coverage is written.
GAO-20-195G also lays out a 12-step estimating process, and a buyer needs mainly its two ends. Step 1 defines the estimate's purpose. Step 12 updates the estimate with actual costs once the work has been performed. A file that never records why it exists, and never replaces forecast hours with the hours the queue actually consumed, leaves the next renewal with nothing to reconcile. Price only the drafting rate and the unassigned work comes back at internal cost.
The internal column carries more than salary. Recruiting belongs there wherever new seats are required. Software licenses sit on that side with the standards upkeep that keeps a seat productive after it is filled. Intake that turns raw survey and GIS records into design-ready route geometry belongs there too, along with the QA/QC cycles that rarely appear in a forecast. Idle capacity after the peak is a real line. Those values come out of the buyer's own cost records. We will not invent them.
Compare the same work product on both sides
The external column extends past the provider's invoice. Someone on the buyer side still issues standards and still rules on exceptions. Native files still get checked against the approved route geometry. Acceptance still needs a signature. A bid that leaves those interfaces unpriced moves work back across the boundary at full internal cost the moment the first comment cycle starts and the permit clock is already running. We price the interfaces in the same file as the drafting volume.
Every output gets a responsible party before any rate is compared. So does every handoff. That includes the quantity basis and the review turnaround the buyer commits to. It also includes the person who resolves conflicts in the source hierarchy when the as-built disagrees with the survey. Rework accumulates in those unassigned gaps. That is expensive.
Same-scope test: if field-data intake appears on only one side of the ledger, the comparison is biased before any rate is argued. Owner review fails the same way. Repair the scope boundary first, then price it.
Buy queue capacity, not the average drawing rate
Field packages do not arrive on a smooth weekly average. The queue the designers actually see is a series of dumps rather than a staffing-model week. A pole owner release can dump a week in one afternoon. Construction questions against an issued drawing jump the queue because a crew is waiting. Average load still backlogs. A team sized to the peak carries idle seats.
The United States Government Accountability Office Schedule Assessment Guide, GAO-16-89G, published December 2015, is the companion volume to GAO-20-195G. It presents ten best practices for a reliable schedule. Two of those practices settle a make-or-buy argument before the rates do: capturing all activities, plus maintaining a baseline schedule so captured work survives a change order. Applied to a design queue, that ranking puts the package that holds the permit need date above the one with the lower drawing rate. A late exhibit lands on construction.
The Federal Highway Administration Guide to Project Coordination for Minimizing Work Zone Mobility Impacts, FHWA-HOP-16-013, addresses coordination of work inside the roadway. It sets no design fees. Its coordination premise still reaches an OSP queue, because need dates that nobody sequenced across permits and make-ready work raise exposure even where every individual drawing was cheap to produce. We keep the throughput baseline tied to demonstrated performance on the first releases so a later change in source quality is attributed to intake rather than treated as a mysterious stall.
Name the cost mechanism before you claim a saving
| Cost mechanism | Internal-team exposure | External-provider exposure | Owner control |
|---|---|---|---|
| Capacity peaks | Queue growth or peak payroll | Reserved capacity or surge pricing | Release forecast and commitment terms |
| Standards and tools | Owner builds the capability | Provider capability plus setup | Criteria file |
| Field-data defects | Designers absorb diagnosis | Questions or scoped remediation | Input acceptance test |
| Design revisions | Staff opportunity cost | Included cycles or change control | Comment classification |
| Idle capacity | Owner pays for fixed seats | Provider manages its portfolio | Ramp-down terms |
| Interface overhead | Coordination with no contract edge | Formal handoffs and owner review | Responsibility matrix |
External capacity lowers cost inside a narrow set of conditions. The buyer would otherwise stand up temporary seats it cannot keep busy past the peak, or spend senior specialists on repeatable production of route geometry and permit exhibits that a criteria file already defines. The economics reverse where scope changes weekly, or where the owner's standards live in one person's head instead of a criteria file. Unopened reviews reverse them too. We say that before we quote.
One limitation of the process we recommend sits in plain view. Draftech adds production capacity with in-house engineering staff. No provider manufactures the owner's governing criteria. A queue stalled on unresolved exceptions still produces an invoice that looks like a production problem. An agreement should name that dependency and reserve buyer hours for answering technical questions inside a stated window. We will not hide that trade-off inside a blended rate.
The queue model closes with a capacity plan that ties forecast input releases to provider throughput and tests owner review capacity against construction need dates. Throughput baselines are project-specific. Buyer and provider set one from demonstrated performance on the first releases, then revise it when quantities or source quality move. That revision is Step 12 working on a smaller file.
Freeze design-ready inputs before anyone prices volume
Define what design-ready means for this owner
Input acceptance states what a design-ready package contains for this owner: route limits, the field dataset, the coordinate reference and vertical datum, the source hierarchy that governs conflicting records, plus every open condition still waiting on a decision. That last item is the one buyers skip. A designer cannot price around a missing exception. A blended rate absorbs the gap until the first review cycle exposes it against a permit submission date that construction is already holding. We surface the gap at intake.
Federal Highway Administration subsurface utility engineering program guidance carries one instruction into OSP work: obtain useful utility information early enough to avoid conflicts and unnecessary relocations. A late or thin field dataset therefore gets priced as an input risk with a named owner. The Federal Highway Administration Program Guide: Utility Relocation and Accommodation on Federal-Aid Highway Projects, FHWA-IF-03-014, supports the same posture where relocation drives the route. Neither FHWA-IF-03-014 nor FHWA-HOP-16-013 sets OSP design fees.
Before production quantities are committed, one representative input package runs the whole path. Records get imported and photos get associated with the correct features before route geometry is checked against the source hierarchy the owner named as governing for that corridor. A sample output returns to the buyer. A clean sample means the interface is accepted as written. A failed sample means the interface needs correction before any unit rate locks. We will not skip the sample. A failed source-hierarchy check at this stage is cheaper than a unit rate that absorbs missing exceptions for the rest of the engagement.
Output acceptance needs the same precision. Our OSP design package release guide separates issued plans from calculations and native source files. The contract names which are required, who accepts each one, plus the objective check acceptance runs against. Acceptance review carries the owner's engineering judgment. Drafting defects belong to the provider's QA pass ahead of release. We keep those two judgments apart.
Pilot before volume: approve the input dictionary and one representative output before the unit rate locks. If the sample surfaces a missing owner decision or an unresolved source conflict, repair the interface and re-price. Contingency buried inside a lower-confidence rate is then paid on every package that follows.
Keep change visible so the external team cannot become a memory hole
Change control preserves the baseline and names the cause. A route expansion consumes design hours. A revised field dataset does the same, yet those two causes carry different money and different approvers. Each entry records the initiating evidence and the affected outputs. It also records the cost treatment, the need date, plus the revised baseline. We keep that log in the buyer's custody.
Retained approval functions get priced where they sit. A provider agreement lists every approval the buyer keeps and attaches a response window to each. It also states what happens when a window lapses. Those hours stay on the buyer's books after the drafting volume moves out, so a proposal that omits them reads cheaper only until the first review cycle is measured against the window the buyer actually needed. We put the hours on the face of the file.
Classify before you schedule: blend revision causes into one rate and the original fee stops describing the work by the third release. Name the payer and the approver first. Then put the hours on a calendar.
- Source-data correction: the field dataset or GIS record was wrong or incomplete, and production waits until it is repaired.
- Owner change: the buyer altered route limits or a governing standard after production started, which reopens the quantity basis.
- Authority comment: a permit reviewer required a revision the original package did not contain.
- Construction condition: field conditions after issue force a redraw.
- Provider defect: the output failed the acceptance check against the standards in force at release, so those hours return to the provider.
Keep the owner's design intelligence portable
The buyer should finish the engagement holding its own standards and its own decisions, with native design files and source records kept in agreed formats so review history and accepted as-builts travel with them into the next procurement. A flattened PDF set leaves the next team without the attributes it needs to continue the route. That gap surfaces during the following procurement. We will not treat a PDF as a complete handoff.
Knowledge moves through the job itself. Decision logs explain why a standard changed. Review comments show how the owner reads its own criteria in the cases that were genuinely hard. An exception re-explained in session after session belongs in the criteria file where the next designer will find it on the following package. Put it there. Native files that cannot be opened outside the provider environment leave the buyer without a portable record when the next contract is bid, which is when the missing attributes actually cost money.
A rising sheet count reads two ways. Productive delivery. Or design fragmenting into rework. The baseline plus the acceptance record are what separate productive delivery from rework that only looks like progress. Open the native files outside the provider environment. Trace one live decision from its source evidence through review status to the accepted release. Records that fail the trace leave the buyer without evidence to choose a delivery model on price. We run that trace before closeout.
Our fiber design services RFP template converts these boundaries into procurement language a bidder can answer line by line. Corridor selection sits upstream of all of it, which is where the fiber route analysis workflow belongs. Both come before the rate conversation. The corridor decision sets the quantity basis every bidder will price against. We will not quote against an unchosen corridor.
Choose when to outsource OSP design to reduce project cost
Construction program manager carrying a temporary peak: buy bounded production against a criteria file that is already written. Set the commercial terms in the order they will bite. Fix the quantity basis and its unit first. Reserved capacity and batch size belong in that same clause, because those two terms determine how the peak is actually staffed. Write the ramp-down that releases the commitment when the peak ends, because a commitment sized to peak volume outlives the peak. We write the ramp-down into the first draft.
Engineering manager whose packages sit on unresolved exceptions: hold the procurement. Repair the decision path first: an owner for each open exception, a published route the question travels, a response window, plus a default action when that window lapses. Measure the queue again over a full release cycle before anyone buys seats. Capacity bought into a decision backlog turns a governance gap into an invoice that reads as a production problem even though the designers were never the constraint. We will not staff around an unanswered exception.
Standards owner holding design knowledge you cannot publish: keep architecture and release authority inside your organization. Scope the provider to bounded packages executed against your criteria file. Custody of the criteria file stays with you. Native-file ownership does not transfer. The source hierarchy the provider must follow is yours, plus return of the decision logs at close. Bounded packages under those terms are still external capacity. We will execute them. We will not own your criteria.
Operations director with demand that has gone year-round: build the internal team. Fund it against recorded actuals rather than the forecast that opened the file. Steady load amortizes the licenses and the intake capability that made the internal column look expensive against a peak week. The criteria file gets a permanent owner. External capacity keeps a defined role, absorbing surge volume and giving an independent read on standards. That is a standing complement, not a substitute.
From the buyer's side of that comparison, Draftech is the external provider. Our OSP engineering is performed in-house by the partner-led Draftech team that stays accountable for it. Our fiber network design service runs source-data acceptance into HLD and LLD work and carries permitting deliverables and QA/QC through to a controlled release. We do not outsource that engineering. The buyer outsources it to us.
Construction, where it belongs to the same engagement, sits on a separate contract surface. Draftech delivers full turnkey construction through Draftech-managed subcontract crews working under our QA/QC and safety program, with single point accountability across both surfaces under the managed crew framework. Engineering stays in-house. We will not blend those labor models into one rate.
Before moving a full design volume to an outside team, a buyer can test Draftech on a qualifying route. Through our free design offer, Draftech engineers the first 20,000 linear feet at no cost, from feasibility and field survey through permit approval. A Draftech owner reviews each request before we commit the package.
If the comparison file is built, send the scope and the quantity basis to info@draftech.com with the owner standards, the available field data, its source hierarchy, plus the permit need dates. We answer with the input states we require. Deliverables we accept against are named in the same reply. Exclusions are stated before quantities are priced.
