44,237 miles of fiber network designed. Every route starts the same way — someone needs accurate engineering, fast. We've built that machine, and it runs across all 50 U.S. states.
There's a version of this story we've seen play out in at least a dozen states: an ISP or electric co-op secures funding, hires a construction crew, and hands them a design that nobody actually stress-tested against the field. The crew shows up and the splice point locations don't work because the engineer didn't account for the drainageway crossing that shows up in the field but not on the satellite imagery. The 144-count feeder they specified runs short because the fiber allocation model assumed 40% take rate and the actual serving area has 58% of addresses in MDU clusters that need independent feeder paths. The permit drawings aren't formatted to the county's submission standards, so the ROW approval is rejected and needs to be redrawn. Now you're six weeks behind before a trench has been dug.
Undersized fiber counts are probably the most expensive design error we get called in to fix after the fact. When an HLD locks in a 96-count distribution cable for a corridor that ends up needing 144 fibers — because the splitter architecture was designed for 40% take rate and the real number is closer to 68% — the cost isn't just redesign fees. It's a second cable pull, a second round of splice labor, and if the conduit was undersized too, it's a second bore. We've seen that scenario add $180,000 to a project that was budgeted at $1.1 million.
Node placement that can't survive the LLD is the second one. An HLD that puts a fiber distribution hub at a location that looks clean on a map but sits on private property with no easement, or in a flood zone that requires special enclosure specs, or 400 feet from the nearest accessible conduit run — that node location has to move. When it moves in LLD, the serving area boundaries shift, the drop lengths change, the optical budget gets re-run, and you're looking at a partial redesign of 3–4 miles of plant. Do that once or twice on a project and you've burned the contingency.
Construction packages that contractors can't work from are the third category. We've reviewed packages from competitors where the pole attachment heights were estimated rather than field-measured, the splice case locations were shown at landmarks that don't exist on the route, and the bill of materials was rounded to the nearest 500 feet of cable. Contractors pad their bids when they can't trust the design — sometimes by 20–25% — because they've been burned by bad packages before. Accurate design pays for itself in the gap between a bid based on a solid package and a bid based on a vague one.
We run a five-phase process from route feasibility through construction support. Each phase has defined inputs, outputs, and quality gates. Nothing goes to the next phase without a QA pass on the previous one — because errors that get through early are exponentially more expensive to fix late.
We start with the coverage area — address points, existing infrastructure, ROW corridors, and aerial vs. underground tradeoffs. HLD locks in the network architecture: route corridors, fiber counts, node placement, splitter hierarchy, and serving area boundaries. Optical budget is modeled at HLD, not left for LLD to sort out. We've learned the hard way — from reviewing other firms' work — that HLD decisions made without optical budget modeling create LLD problems that are expensive to fix.
Our field survey teams use Katapult for GPS strand mapping, pole inventory, and attachment height measurement — photo-documented and QA'd before anything touches the design tools. We measure existing attachment heights; we don't estimate from the road. A field-verified 26.3-foot attachment height versus an assumed 27-foot standard can change a pole loading analysis result from passing to failing. That's not a hypothetical — it's happened on projects we've inherited from other firms.
Low-level design is where architecture becomes construction reality. Every splice point, every drop location, every reel cut length, every attachment height, every conduit schedule. We produce LLD in AutoCAD and deliver in the format the client's construction team or contractor needs. On aerial routes, make-ready engineering runs in parallel — flagging poles near NESC loading limits before the LLD is finalized so the construction package reflects actual make-ready scope.
Scaled permit drawings for county ROW, state DOT highway crossings, railroad crossing agreements, and USACE waterway crossings — formatted to each jurisdiction's submission standards. We've submitted permit packages in enough states to know that a drawing formatted for Florida DOT won't work for TXDOT, and what works for one North Carolina county won't satisfy the adjacent one. We don't use generic templates. See our permitting service page for full detail on permit types and timelines.
Design doesn't end when the construction package ships. Field conditions change. Contractors encounter obstacles that weren't visible in the survey — a buried abandoned conduit, a pole that's been replaced since the field data was collected, a property owner who won't grant access for the planned route. We provide RFI support during construction and deliver as-built documentation once the network is in the ground, updated to reflect actual field conditions.
Our fiber network design scope runs from the first desktop feasibility review through construction-ready delivery and as-built update. Here's what that covers in full — because "fiber design" means different things to different firms, and you deserve to know exactly what you're getting.
We evaluate each route segment for aerial vs. underground placement, existing pole infrastructure condition and capacity, ROW corridors, and cost per mile. In most rural builds, the answer is hybrid — aerial along existing electric co-op pole lines and underground through road crossings and areas where aerial isn't available. GIS-based corridor analysis using QGIS and ArcGIS identifies the most permittable options before we commit to a route. For longer projects, we model aerial vs. underground cost tradeoffs at a per-segment level so the HLD reflects realistic economics, not wishful thinking.
HLD establishes network architecture — route corridors, fiber counts, node placement, splitter architecture, serving area boundaries, and optical budget. It's the strategic foundation that every downstream phase builds on. We design HLD to be constructable, not just presentable. An HLD that looks clean on paper but puts a fiber distribution hub at a location with no practical easement path is a problem waiting to happen. We've fixed enough of those to be rigorous about constructability review at the HLD stage.
PON architecture is where the network's long-term economics get locked in. Wrong splitter placement, wrong splitter ratio, or a cascade design that wasn't modeled for optical loss creates problems that persist for the entire life of the network. We design GPON (ITU-T G.984), XGS-PON (ITU-T G.9807.1), and NG-PON2 architectures. For new builds in 2026, we default to XGS-PON — it supports 10 Gbps symmetrical service and the ODN architecture is essentially identical to GPON at the fiber plant level. Locking in GPON on a new build today creates an upgrade project in three or four years. Our optical budget modeling keeps every design path under 28 dB end-to-end for GPON, verified against OLT minimum receive sensitivity before HLD is finalized. A poorly placed cascade splitter stage can add 3–4 dB of insertion loss — that comes directly out of the margin you need for long rural spans and connector aging.
Pole-by-pole or conduit-by-conduit construction documents: splice case locations, strand assignments, reel cut lengths, attachment heights, conduit schedules, drop locations, and bill of materials. LLD is what the crew works from in the field. We produce LLD in AutoCAD and can deliver in GIS-compatible formats for clients running network inventory platforms. For aerial routes, LLD runs alongside make-ready engineering so that both packages are ready simultaneously — not sequentially.
GPS strand mapping, pole inventory, attachment height measurement, photo documentation, and GIS integration — using Katapult and documented to QA standards before it touches the design. We don't design against desktop data when the project budget supports field verification. The projects that produce the cleanest construction packages — fewest change orders, least field rework — are the ones where the field survey was thorough.
Scaled permit drawings for every crossing type and ROW jurisdiction on the route — formatted to that jurisdiction's specific requirements. County highway departments, state DOTs, railroads, USACE — each has its own format, its own required notes, and its own review process. We manage the drawing production and can coordinate permit submission and tracking through our utility coordination team.
For aerial routes, pole loading analysis using O-Calc Pro is performed during or immediately after field survey — not after the construction contract is signed. Knowing make-ready scope at the design phase lets you budget accurately and sequence construction realistically. We've seen projects where the make-ready scope discovered late added three months to the schedule and 18% to the construction budget.
The final package includes everything the construction contractor needs to build without calling the engineer for clarification: route design drawings at the plan-and-profile or schematic level, splice diagrams, conduit and equipment schedules, crossing details, permit drawing set, pole loading analysis reports, and a complete bill of materials tied to specific construction sections. As-built updates are delivered within 30 days of construction completion in the agreed format.
Tools we use: QGIS, AutoCAD, Katapult, O-Calc Pro, ArcGIS, and GIS platforms including IQGeo and GE Smallworld for clients running inventory systems. We select tools based on project requirements and utility specifications — not because we have a single-tool workflow. Fiber network design outsourcing works best when the engineering firm can match your existing toolchain.
The HLD/LLD distinction is sometimes treated as a formality — two names for the same deliverable at different levels of detail. It isn't. They're genuinely different engineering exercises with different purposes, different audiences, and different failure modes.
HLD is the decision layer. Route selection, fiber counts, node placement, splitter architecture, and serving area boundaries — these decisions are made at HLD. Change them after LLD starts and you're doing partial redesign work. That's expensive. HLD also validates the optical budget at a path level — feeder loss, distribution loss, splitter insertion loss, drop cable loss, and connector budget — before the network architecture is locked. An HLD that shows 29.4 dB total optical loss on the longest path has a problem. Catching it at HLD costs hours to fix. Catching it at LLD costs days. Catching it at commissioning costs real money and real schedule.
We also use HLD to validate fiber count adequacy. An 83-count fiber in a distribution cable sounds like it covers a 64-split PON network — but if the architecture uses a cascade 1:4 × 1:16 design with two of those locations serving as FDH entry points, the actual fiber demand at the splice point is 96 counts plus express fibers. HLD is where you catch that, not when you're ordering cable. See our FTTH HLD design mistakes guide for the specific errors we see most often.
LLD is the construction layer. Splice point locations at specific pole or handhole numbers, individual drop locations tied to address points, reel cut lengths for each cable section, attachment heights for aerial strand, conduit section assignments, and equipment specifications down to enclosure model numbers. LLD is where a good HLD proves itself — and where a bad HLD creates expensive detours. We've worked on LLD projects where the HLD had node locations that couldn't survive contact with actual property lines or terrain. Moving a node in LLD is a 2–3 day event, not a 20-minute edit.
Architecture selection at HLD determines the fiber plant design — and each architecture has real engineering tradeoffs that affect construction cost, operational flexibility, and upgrade path. Getting this decision right at the start matters. Here's how we think about it.
GPON is the dominant PON standard in the U.S. market today and still makes sense for projects where the client already has GPON OLT equipment, where XGS-PON equipment isn't available in the required form factor, or where state program requirements specify GPON. ODN architecture is a passive splitter hierarchy — the fiber plant itself is protocol-agnostic, so a GPON network can be upgraded to XGS-PON by swapping OLT and ONT electronics without touching the fiber. We design to ITU-T G.984, with optical budget targeting <28 dB total loss on the longest path. Splitter ratios of 1:32 for rural and 1:64 for suburban or urban density. See our GPON network design guide for the detailed architecture breakdown.
XGS-PON (ITU-T G.9807.1) is our default recommendation for new builds in 2026. Ten gigabit symmetrical, ODN architecture essentially identical to GPON at the fiber plant level, and equipment pricing has come down significantly in the last 18 months. The incremental design cost to specify XGS-PON instead of GPON is minimal. The incremental cost to upgrade from GPON to XGS-PON after the fact — at OLT, ONT, and potentially SFP module level — is real. We've been recommending XGS-PON as the default for greenfield builds since 2024 because the upgrade economics are clearly superior.
Active Ethernet is a point-to-point fiber architecture — one fiber strand per customer, terminated on a switch port at the hub or central office. It's more expensive in fiber count and switch port cost than PON architectures, but it offers dedicated bandwidth per subscriber with no shared downstream contention, and it's the preferred architecture for enterprise and business service where SLA guarantees matter. Design considerations are fundamentally different from PON — no splitter hierarchy, different fiber allocation model, and a different hub site infrastructure requirement. For last mile fiber design in mixed residential/commercial corridors, we sometimes design hybrid plants with a PON distribution layer for residential and active Ethernet on business service paths.
| Factor | GPON | XGS-PON | Active Ethernet |
|---|---|---|---|
| Max downstream speed | 2.5 Gbps shared | 10 Gbps shared | 1–10 Gbps dedicated |
| Fiber per subscriber | Shared (1:32–1:64) | Shared (1:32–1:64) | Dedicated strand |
| Optical budget target | <28 dB | <29 dB (Class N1) | N/A — active electronics |
| Upgrade path | Swap to XGS-PON (electronics only) | Future NG-PON2 upgrade path | Port speed upgrade |
| Best fit | Existing GPON operators, legacy equipment | New greenfield FTTH builds | Enterprise, MDU, business service |
| Design complexity | Medium — splitter hierarchy design | Medium — same as GPON at ODN level | Higher — more fiber, more switch design |
BEAD-funded fiber networks require engineering documentation that goes substantially beyond what a standard ISP deployment demands. The NTIA BEAD program requires subgrantees to demonstrate coverage of all unserved and underserved locations at minimum 100/20 Mbps, document cost-per-location economics, and deliver HLD and LLD packages in formats that satisfy state broadband office reporting requirements. Those requirements vary by state — and program administrators are increasingly sophisticated about what constitutes adequate engineering documentation.
The documentation standard that matters for BEAD isn't just technical accuracy — it's traceability. State broadband offices want to see that every funded address point was identified from FCC fabric data, that the HLD demonstrates coverage of each unserved and underserved location, and that the cost-per-location model is tied to the actual design rather than a desktop estimate. We've reviewed BEAD engineering submissions from subgrantees whose engineering documentation was rejected not because the design was wrong, but because the address verification methodology didn't align with NTIA's location-based requirements. See our guide on BEAD engineering requirements for the specifics.
Subgrantees that engage engineering before their award announcement enter the design phase without the capacity crunch that follows award announcements. Right now, every BEAD subgrantee in states where awards are imminent is competing for the same engineering resources. The firms that started pre-award are ahead. If you haven't started yet, the time to move is now — not after the award letter arrives.
Our BEAD fiber network design scope includes:
Should you build an in-house design team or outsource? It depends on your project pipeline, timeline pressure, and how many states you're building in. Here's how the comparison actually looks for a mid-size ISP or co-op running 2–5 projects simultaneously.
| Factor | In-House Design Team | Draftech |
|---|---|---|
| Time to first deliverable | 3–6 months (hiring + onboarding + tooling) | 2–3 weeks from project kickoff |
| Cost per route mile | Higher at low volume — fixed overhead | Variable cost scales with project pipeline |
| Geographic flexibility | Limited to where your staff can operate | Active in 22 states, deployable across all 50 |
| BEAD compliance expertise | Requires dedicated training per state | Designed BEAD networks across multiple states |
| 50-state coverage | Not practical without significant headcount | Available across all 50 U.S. states |
| Surge capacity | Constrained by headcount — hiring takes months | 600+ engineers — scale up without lead time |
| Software licensing | Additional cost — QGIS, AutoCAD, Katapult, O-Calc Pro | Included — no per-project software cost |
| Long-term cost at scale | Lower if you maintain 10+ designers at full utilization | Competitive — no benefits, turnover, or downtime cost |
The decision isn't binary. Many of our clients run a small in-house team for day-to-day project management and route oversight, and use Draftech for production capacity — the actual field survey, HLD, LLD, and permit drawing production. That hybrid model gives them oversight without the fixed overhead of a full design staff. Read more about how that works in our fiber network design outsourcing guide. For legacy operators considering a technology shift, our guide on copper to fiber migration engineering covers the specific design considerations that apply when overlaying fiber on existing copper plant.
Design cost per mile varies significantly by project type, terrain, underground vs. aerial topology, and permitting complexity. These are real-world ranges from projects we've delivered — not marketing estimates.
| Project Type | Typical Design Cost/Mile | What Drives the Range |
|---|---|---|
| Rural FTTH, aerial | $1,100–$1,400/mile | Co-op pole density, make-ready scope |
| Suburban FTTH, mixed | $1,250–$1,700/mile | Underground segments, denser permitting |
| Underground only | $1,600–$2,200/mile | Bore design, conduit layout, more crossings |
| BEAD project (full documentation) | $1,340–$1,750/mile | State documentation requirements, address verification |
| Electric co-op fiber | $1,050–$1,350/mile | Existing easement access simplifies ROW |
| Middle-mile / backhaul | $900–$1,500/mile | Fewer splice points, longer spans, but more crossings |
These figures include field survey, route design, permit drawings, pole loading analysis, utility coordination, and construction-ready deliverables. As-built documentation is typically scoped separately. Electric cooperative fiber design often comes in at the lower end of the range because existing easements along electric rights-of-way reduce the ROW research burden significantly. For a project-specific estimate, contact us with your route miles, approximate topology, and state — we'll give you a real number within 48 hours.
A lot of engineering firms deliver something they call a "design package" that's really just an HLD with a bill of materials attached. That's not a construction package. Here's what a construction-ready fiber design package from Draftech actually includes — and why each component matters.
FREE FIRST PROJECT
Active in 22 states. First 20,000 LF of fiber network design free — no commitment, no strings.
Get Free Design →ARE YOU A FIBER DESIGN FIRM?
This page describes the service we deliver to clients. If you provide OSP fiber HLD/LLD production and are looking for a consistent subcontract pipeline, we have ongoing capacity needs across multiple states and network types.
Tell us your route miles, approximate topology, and state. We'll scope the project and give you a timeline and cost estimate — usually within 48 hours. We've worked across all 50 U.S. states and understand what the local utilities, ROW agencies, and state broadband offices actually require.
Contact Our Engineering TeamOr email directly: info@draftech.com — we reply within one business day. | First 20,000 LF free — no commitment.
SERVICE AREAS
Active in 22 states and deployable across all 50 U.S. states — including our highest-volume BEAD markets:
View all service areas →Draftech International provides fiber network design services across all 50 U.S. states — from small regional ISPs to Tier-1 carriers and BEAD-funded subgrantees. Contact our engineering team to discuss your project. | 15280 NW 79th CT, Suite 102, Miami Lakes, FL 33016 | 305-306-7407