People mix these two terms up constantly — and it's not just semantics. Confusing HLD with LLD, or treating them as interchangeable phases you can compress or skip, is how fiber projects end up over budget, behind schedule, and redesigned mid-construction. I've seen it happen on builds ranging from 8-mile rural loops in Texas to 140-mile backbone routes stretching across mountain West Virginia. The pattern is the same every time: someone skipped proper HLD work, or handed off to LLD too early, and the project paid for it.
The difference between HLD and LLD in fiber design isn't complicated once you understand what each phase is actually supposed to produce. But the implications — for budgets, for permitting, for grant compliance, for the field crew who's supposed to build what you designed — run deep. This article covers both phases in detail, what goes wrong at the handoff between them, and why getting either one wrong sets the whole project back.
What HLD Actually Is (and What It's Not)
High Level Design is system-level engineering. It's the phase where you define the architecture of your fiber network — not the construction details, not the pole-by-pole plan sheets. The deliverable from a good HLD engagement is a route map, a system diagram, and a cost model. Those three things, done right, tell you what you're building, where it goes, and roughly what it'll cost before anyone touches a permit application.
What goes into HLD? Route corridor selection — which roads, rights-of-way, and utility paths the fiber will follow. Fiber counts on each segment — backbone versus distribution, and whether you're running a 288-count or a 144-count on the main spine. Node placement — where your hub sites, distribution nodes, and splice locations will land. Splitter ratios for passive optical network builds. And a cost model that rolls up route miles, pole count estimates, make-ready assumptions, materials, and labor into a project budget. That cost model isn't construction-precise — it doesn't have to be — but it needs to be close enough that a project manager or grant reviewer can evaluate whether the scope is fundable.
HLD does not tell a crew how to build anything. It doesn't include pole schedules. It doesn't specify conduit sizing or attachment hardware. It doesn't go pole-by-pole. That's LLD's job — and conflating the two is what causes projects to skip HLD entirely and dive straight into construction drawings on a route that was never properly evaluated.
The tools at the HLD stage are mostly GIS-based — ArcGIS for route corridor analysis, aerial imagery review, cost-per-mile modeling, and coverage footprint validation. You're working at network scale, not at construction detail scale. Some firms also use their own cost-modeling platforms at this stage to generate preliminary estimates. What you're not doing is opening AutoCAD and drawing plan sheets — that comes later.
What LLD Includes — The Document That Drives Construction
Low Level Design is where the network gets built — on paper first, then in the field. Every piece of information a construction crew needs to execute a fiber build comes out of the LLD package. If it's not in the LLD, a crew either can't do the work or has to stop and call someone for clarification. Both outcomes cost money.
A complete LLD package for a fiber build includes:
- CAD plan sets — scaled plan view sheets, typically drawn in AutoCAD at 1"=50' or 1"=100', showing pole locations with verified IDs, attachment positions, span lengths, fiber route, and proposed hardware. For a 14.2-mile aerial build, that's typically 65–80 plan sheets before you add detail sheets.
- Pole schedules — a table listing every pole on the route with its ID, owner, class, height, existing attachments, and make-ready requirements. This is what make-ready engineers and NJUNS coordination teams work from.
- Splice point diagrams — detailed drawings showing each splice location, enclosure type, fiber count in and out, and splice tray configuration. Poor LLD quality control on splice diagrams is one of the fastest ways to create field rework.
- Conduit sizing and bore details — for underground segments, conduit diameter, material spec, depth, and bore crossing details for every road and waterway crossing.
- Bill of materials — a complete BOM covering cable, conduit, hardware, splice enclosures, strand, and equipment. This is what the procurement team uses to order materials and what the GC uses to price labor. See the fiber construction package deliverables guide for a full breakdown of what a complete BOM should include.
- Permit-ready plan set — formatted to the specific requirements of each permit jurisdiction — utility, municipality, state DOT, or railroad — with correct title block fields, coordinate references, and note sets.
The LLD is what costs the most to produce. It's also what takes the longest — typically 8 to 16 weeks for a complex build, versus 4 to 8 weeks for HLD. That ratio matters when you're managing a BEAD project timeline with hard milestones.
Tools at the LLD stage are primarily AutoCAD and Civil 3D for plan set production, with O-Calc Pro or Katapult for pole loading analysis when make-ready is complex. GIS exports come alongside the CAD deliverables — shapefiles, GeoJSON, or ESRI File Geodatabase — because the LLD data needs to feed the operator's network management system, not just sit in a PDF binder.
The Critical Handoff Between HLD and LLD
The space between HLD approval and LLD kickoff is where most rework happens. It's not a step most project managers pay attention to — and that's exactly the problem.
HLD is built on assumptions. That's fine — it has to be. You can't complete a full field walk on 22 miles of route before you've decided whether to build it. But every assumption in the HLD — average pole spacing, make-ready complexity, splice point placement, conduit depth on underground segments — gets tested the moment LLD engineers start doing real permitting and field verification work. When those assumptions don't survive contact with reality, you get redesigns. Expensive ones.
The handoff works well when the HLD team has documented its assumptions explicitly and when the LLD team has been briefed — ideally in a structured design review — on which HLD assumptions carry the most uncertainty. It works badly when HLD is handed off as a finished route map with no notes, no flagged risks, and no cost-model breakdown by assumption category. The LLD team then discovers the problems themselves, one at a time, at the worst possible moment — usually after permit submissions are already in progress.
Pre-construction stall patterns almost always trace back to a broken HLD-to-LLD handoff — either assumptions that weren't validated, routes that changed after LLD started, or cost models that couldn't absorb the make-ready findings the LLD uncovered. Reviewing the most common pre-construction stall patterns in BEAD projects shows the same failure modes repeating across states.
Why Getting HLD Wrong Makes LLD Expensive
Here's a real example from a build we were brought in to rescue.
A 22-mile aerial backbone project in mountain West Virginia — the kind of terrain where ridge lines look deceptively simple on a satellite view but are a completely different story when you're looking at pole attachments. The HLD had been done by the ISP's internal team using aerial imagery and a rough cost model. The route was approved, the grant was awarded, and LLD kicked off. Good news so far.
Twenty-two miles. Roughly 37 poles per mile on average — call it 814 poles. The HLD cost model had assumed standard make-ready: maybe 15% of poles requiring some level of rearrangement, at average cost. Reasonable assumption on flat ground in rural Texas. Not reasonable on West Virginia ridgelines where distribution poles are overloaded by default and every new attachment triggers a full engineering review.
When the LLD team ran O-Calc Pro structural analysis on the poles — which the HLD had not done — 34% of poles required full make-ready. Not rearrangement. Full replacement or down-guying, because the loading conditions left no margin for a new fiber attachment under NESC standards. That's 277 poles — at costs ranging from $800 to over $2,100 per pole depending on access conditions. The LLD's make-ready cost alone was $380,000 over the HLD budget.
The HLD didn't account for pole loading. It could have — a field walk of even 20% of the route would have flagged the loading problem in week two, not week fourteen. The project didn't get redesigned at the HLD stage because the HLD team didn't have OSP engineering depth. That's the cost of treating HLD as a rough sketch.
A bad HLD doesn't just produce a bad cost model. It produces an approved scope that the LLD team has to either honor — and go over budget — or push back against, which means restarting stakeholder approval cycles weeks or months into the project. Neither option is good.
The lesson from West Virginia: HLD cost models need to account for structural loading conditions, not just average make-ready rates. On any route through mountainous terrain, run at least a sample O-Calc analysis on 10–15% of poles before finalizing the HLD cost model. It takes a week. It can save months.
How HLD and LLD Fit Into a BEAD Project Timeline
BEAD has introduced a new urgency around both design phases — not just getting them done, but getting them done in the right sequence, to the right standard, at the right time.
NTIA requires HLD-level documentation for subgrantee applications. That means a network design showing how the proposed deployment covers the eligible locations in the funded service area — route maps, coverage analysis, fiber architecture, and a cost model that supports the budget request. Subgrantees who submit rough sketches instead of real HLD packages get flagged by state broadband offices. The HLD requirements for BEAD subgrantees have been tightening as states move from initial allocation into active subgrantee review cycles.
States are also starting to require LLD-level documentation before releasing construction funds. That's a significant shift from how RDOF and earlier federal broadband programs worked. It means subgrantees can't just get an HLD approved and start building — they need construction-ready plan sets, permit approvals, and make-ready clearances before the state releases the next funding tranche.
What does that mean for a project timeline? A typical BEAD build sequence — from HLD kickoff to construction start — runs roughly 22 weeks on a mid-size build covering 40 to 60 miles of new plant. HLD takes 4–6 weeks. The HLD review and approval cycle adds 2–3 weeks. LLD runs 10–14 weeks in parallel with permitting. Permit approvals — especially on routes crossing state DOT right-of-way or railroad crossings — add another 4–8 weeks, and they don't start until LLD drawings are submitted.
Miss the HLD, and the whole sequence compresses in ways that create errors. Skip the HLD review gate, and LLD starts on assumptions that may not survive the first permit submission. Get HLD wrong on fiber counts or route selection, and you're redesigning LLD drawings 8 weeks into a 14-week LLD phase — which means you're now behind schedule and burning budget simultaneously.
HLD vs. LLD: A Direct Comparison
| HLD | LLD | |
|---|---|---|
| Purpose | System architecture and planning | Construction-ready engineering |
| Primary deliverable | Route map + system diagram + cost model | CAD plan sets + BOM + pole schedule |
| Who uses it | ISPs, project managers, grant reviewers | Field crews, contractors, permit offices |
| Detail level | Network-wide, not pole-by-pole | Pole-by-pole, splice-by-splice |
| Typical timeline | 4–8 weeks | 8–16 weeks |
| When produced | Before permitting begins | After HLD is approved |
That sequence — HLD approved, then LLD — isn't a formality. It's the only order that makes the LLD efficient. Every hour of LLD drafting done on a route that hasn't been validated at the HLD stage is an hour at risk of being thrown away.
Choosing a Firm That Does Both Well
Most OSP engineering firms do one or the other well. Some specialize in network planning and systems design — strong HLD work, weaker on construction-level CAD. Others are construction drawing shops — fast at AutoCAD plan sets, but not set up to do the GIS-based route analysis and cost modeling that defines a real HLD. The firms that do both phases with equal depth are rare, and they're worth finding.
Why does it matter that the same firm does both? Because the HLD-to-LLD handoff is internal when it's one firm — and that eliminates the single biggest source of assumption drift in fiber project design. When the engineer who made the HLD cost model is also reviewing the LLD pole schedules, they know which assumptions to test first. When two separate firms handle the phases, every assumption the HLD made has to be re-explained, re-documented, and re-validated by the LLD team — a process that takes weeks and still produces gaps.
When you're evaluating a firm's capability to handle both phases, ask specific questions. What GIS tools do they use for HLD route analysis? Can they show you an HLD cost model breakdown by assumption category? What's their process for flagging HLD assumptions that need field verification before LLD begins? And on the LLD side — what's their QC process before a plan set leaves the building? Do they run NESC clearance checks on every attachment? Do their pole schedules cross-reference against the utility's GIS export, or are they working from field sketches?
You should also ask about their experience with HLD and LLD design services on funded projects — BEAD, RDOF, USDA ReConnect. Grant-funded builds have documentation requirements that don't exist on private capex projects. A firm that's done 50 miles of private ISP buildout isn't necessarily equipped to handle the HLD documentation standards that a state broadband office will review for subgrantee approval.
There's also the question of outsourcing OSP design for rural ISPs who don't have internal engineering staff. For operators in that position, the handoff risk between HLD and LLD firms is even more acute — because there's no internal team to catch gaps between what the HLD firm assumed and what the LLD firm produced.
Draftech has designed over 44,000 miles of fiber network across all 50 U.S. states — aerial, underground, and hybrid builds — with HLD and LLD handled as a single integrated scope on every project. No separate handoff between planning and construction engineering. The same team that builds the route model and cost estimate is the team that produces the permit-ready CAD package. That integration is what keeps assumption drift from becoming budget overruns. Our OSP engineering services cover the full cycle from route selection through as-built closeout. For builds requiring field data before LLD can begin, our field survey services deliver GPS-accurate pole data in structured formats that feed directly into design — no translation layer, no data quality gaps.
Coastal Maine. Mountain West Virginia. Rural Texas. High desert New Mexico. The terrain changes — the discipline required to do HLD and LLD correctly doesn't. Don't treat either phase as optional, and don't separate them into independent workstreams if you can avoid it. The 22-week timeline I mentioned earlier assumes both phases are done right, in sequence, by people who know what they're doing. Cut corners on HLD and that timeline doesn't compress — it blows up.
If you're planning a fiber build and want to know whether your current HLD is solid enough to support LLD, reach out at info@draftech.com. We can review your existing design documents and tell you specifically what's there, what's missing, and what it'll take to get to construction-ready engineering.
