Methodology
How we score an EV charging site
Hardware is not the bottleneck. Site selection is. Every site we assess goes through three engines: a spatial model that scores the ground it sits on, a planning layer that reads where the area is heading, and a financial model that turns both into a payback period.
Get my site scoredFree · report by email within 3 working days
How it works
You give us a location and a budget. We return a ranked, defensible answer.
You submit the form
Location, plot size, existing power connection, budget and objective — the form below.
The engine scans your area
We lay a grid around your point and score every cell against POI density, road access, grid proximity and competing chargers.
We simulate the economics
Your budget is converted into an optimal charger mix, then run through a 15-year cash flow in three scenarios.
You get a report by email
Score, heat map, competitor list, recommended configuration, payback and IRR. Within 3 working days.
Spatial site score
We lay a grid over the area around your site and score every cell from 0 to 100. These are the criteria that move the number most — alongside dozens of secondary signals. The weights behind them are calibrated per area type and stay in-house.
What moves the score
Road access
Not just whether a road is nearby, but what kind — a highway frontage and a back lane are worlds apart, and the model knows the difference.
Demand density
The resident and commuter base within realistic reach — the people who will charge here week after week, not the ones who pass once.
Anchor points
Places that hold a driver long enough for a charge to finish. A mall and a convenience store both count — but nowhere near equally.
Grid proximity
The criterion that quietly kills the most projects. A site can win on every other axis and still be unaffordable once the connection is costed.
Existing competitionsubtracts
Real stations actually operating nearby — pulled from a live national inventory with connector counts and power ratings, not guessed from a public map.
…plus dozens of secondary signals: dwell-time class of each anchor, road hierarchy, one-way and median constraints, water masking, and area typology.
Reading the score
Scores are calibrated on a fixed national scale, so a 72 in Da Nang means the same thing as a 72 in Hanoi.
80 – 100
Prime
Build here.
60 – 79
Strong
Viable; check the grid cost.
40 – 59
Marginal
Only with an anchor tenant.
0 – 39
Weak
Look elsewhere.
Why the score is reproducible
Globally anchored grid
Cell centres are fixed to an absolute origin, so the same physical patch of ground is always the same cell — the score does not shift because you moved the map.
4.4 km data padding
We load data well beyond the area being scored, so cells at the edge see their true surroundings instead of an artificial void.
National calibration
Normalisation uses a fixed nationwide scale, not a scale relative to whatever area you happened to select. Scores compare across provinces.
Water masking
Cells that are 60% or more water are dropped. A river bend scores well on access and demand; you still cannot build on it.
The engine
Multi-model scoring
One model that is confidently wrong is worse than no model. So we never run one. Independent methods score the same cell from different angles, and where they disagree is where a human analyst looks first.
Rule model
A fully traceable model where every point of the score can be walked back to a specific shop, road segment or substation. This is the layer that has to explain itself.
Learned from proven sites
Decades of petrol-station siting decisions are already priced into the map — real money, real survivorship. We learn that national pattern and ask how far your site falls short of it. Exceeding it is never punished.
Agreement as confidence
Where the layers agree, the score is solid. Where they diverge, that gap is itself the finding — and it is flagged in your report rather than averaged away into a number that looks confident.
Every figure in your report carries the layer it came from and the data it was computed on. No number arrives without a source.
Planning & growth potential
A site is not scored on what is around it today alone. A charging station is a 10-year asset — what matters just as much is what the area is scheduled to become. This is the layer most site-selection tools skip entirely.
Land-use zoning
What the parcel and its neighbours are zoned for. A site facing a plot zoned for a future residential block is a very different bet from one facing land already built out.
Pipeline projects
Approved infrastructure and property projects in the catchment: ring roads, interchanges, industrial parks, malls and residential developments — with their scheduled timing.
Official land pricing
The provincial land-price schedule that governs what you will actually pay in rent or land-use fees — read from the current official resolution, not from listing-site asking prices.
Growth trajectory
Whether the area is filling in or already saturated — and whether the demand curve arrives before or after your payback period ends.
Planning coverage differs by province — some publish machine-readable GIS layers, others only a signed decision document. Your report states which level applied to your site.
ROI simulation
A good location still needs to pay for itself. We size the station to your budget, then run a 15-year cash flow against it.
1. Sizing & CAPEX
- Chargers
- 11 / 30 / 60 / 120 kW at 32 / 163 / 200 / 300 M VND. Allocated to maximise kW per million invested.
- Transformer
- Load = total kW ÷ 0.85, then a 250 / 400 / 630 kVA class at 160 / 230 / 380 M VND.
- Switchgear & installation
- Fixed 320 M VND for MV + LV switchgear; installation 170 / 280 / 390 M VND by transformer class.
- BESS & solar
- Battery at 4 M VND/kWh, rooftop PV at 9 M VND/kW — the two components that shift purchasing into off-peak hours.
2. Energy & revenue
- Throughput
- Total kW × 24 h × utilisation × 365 days.
- Revenue
- Slow-charge kWh at 750 ₫, fast-charge at 800 ₫, plus the network settlement rate of 3,302 ₫/kWh.
- Grid cost
- Only the shortfall after BESS and solar self-supply, priced on the off-peak / normal / peak tariff mix of 1,653 / 2,755 / 4,298 ₫.
- Operating cost
- 2% of CAPEX for warranty, straight-line depreciation (chargers and transformer 10 years, solar 20), software, and land rent if the site is leased.
3. Financial metrics
- Payback period
- CAPEX ÷ first-year profit after tax.
- IRR
- Internal rate of return on a 15-year cash flow, including the battery replacement in year 11.
- ROE
- Average annual profit after tax over total invested capital.
Utilisation falls as the station grows
This is the single assumption that most often breaks an EV charging business case. A 240 kW site in the sweet spot runs busy; a 900 kW site in the same catchment simply has more idle bays. We model utilisation as a step function of total installed power rather than one flat rate.
| Total installed power | Utilisation | Effective hours/day |
|---|---|---|
| ≤ 150 kW | 25% | 6,0 h |
| 150 – 350 kW | 20% | 4,8 h |
| 350 – 600 kW | 15% | 3,6 h |
| > 600 kW | 10% | 2,4 h |
Three scenarios, every time
A single number hides the risk. We bracket it.
Optimistic
Utilisation one tier above base, dedicated transformer already on site, no land rent.
Base
Utilisation from the power tier, grid upgrade costed from your declared connection status.
Risk
Utilisation one tier below base, transformer upsized one class, and a competitor opening within 1 km in year 2.
Foundations
The data underneath
Every input is a real, dated dataset — not an estimate typed into a spreadsheet.
>18.000
charging points nationwide
18,685 operating car-charging stations, merged from multiple national sources and de-duplicated by station ID — this is the competition layer.
34
provinces covered
Full nationwide coverage under the current administrative map — any address in Vietnam resolves to a scoreable point.
131k
searchable places & streets
An offline OpenStreetMap index of Vietnam — 114k streets and 16k named places.
≤ 440 m
finest grid cell
Cell size adapts to the area, from roughly 440 m up to 1.8 km, capped at 1,500 cells per run.
The sources behind every score
We publish where the data comes from, not the weights we apply to it.
Competition
- National charging-station inventory — 18,685 operating car-charging stations, refreshed
- Connector count, power rating and AC/DC split per station
Movement & demand
- Road network by hierarchy, nationwide
- Residential and apartment building footprints
- Retail, hospitality, health and education anchors
Grid
- Substation locations and distance modelling
- Hourly national tariff structure across the load day
Planning
- Provincial land-use planning portals and GIS services
- Official land-price resolutions in force
- Approved infrastructure and property project registers
What you receive
A PDF report by email. The scoring workbook itself stays on our side.
Site score with breakdown
Your 0–100 score, plus the individual score on each of the five criteria so you can see exactly what is dragging it up or down.
Neighbourhood heat map
A ranked map of the cells around you — often the best cell is a few hundred metres from where you were planning to build.
Competitor inventory
Every operating charging station within the catchment: operator, max power, connector count and AC/DC split.
Recommended configuration
The charger mix your budget actually buys, the transformer class it needs, and whether BESS and solar pay for themselves here.
Payback & IRR, three ways
Optimistic, base and risk — with the assumption that separates them stated explicitly on each.
Planning & growth read
What the surrounding land is zoned for, which approved projects land in the catchment and when, and the official land price that sets your rent.
CAPEX risk flags
What could blow the budget: distance to the substation, a transformer sitting just over a class boundary, or a site too small for the charger count.
Request an assessment
Tell us about your site
Five short sections. The more precise your answers, the tighter the scenario range in your report.
Scope & limitations
- The score ranks locations against each other; it is not a permit, a grid connection agreement, or a land-use planning clearance.
- POI and road data come from OpenStreetMap and carry its coverage gaps, which are wider in rural districts than in cities.
- Planning data is read from official provincial sources, and its depth varies: some provinces publish machine-readable GIS layers, others only a signed decision. Where a layer is unavailable, an analyst reads the decision manually and the report says so.
- Financial figures are a model, not a forecast. Tariffs, equipment prices and land rent move; we date every assumption in the report.
- A final investment decision still needs an on-site survey and a written capacity confirmation from the utility.