Request a Demo
Get in touch with us and we will provide a solution that meets your exact requirements
Interviewer.AI · Sustainability Impact Suite
Methodology & Sources
Everything behind the numbers: the formulas, the regional parameters, and the published data they're built on. If a figure appears on the impact page or in your PDF report, you can trace it here. All figures are modeled estimates from your inputs and published regional averages — not metered measurements.
← Back to the impact calculatorThree principles
📏
Estimates, not audits
These are directional figures for ESG communication and planning, built from your inputs and published averages — not a verified emissions audit or metered measurement.
🌍
Regional, not global
Every parameter — travel habits, salary costs, grid carbon intensity — is set per region (North America, Europe, Asia Pacific), never a single world average.
⚖️
Net, not gross
The electricity used by Interviewer.AI's own servers is calculated and subtracted before any CO₂ is reported as avoided. No cloud-washing.
The calculations
1 · Travel avoided (km)
interviews × one-way distance × 2
Each remote interview replaces one round trip to your office.
2 · Transit emissions avoided (kg CO₂e)
travel avoided × regional transit factor
The transit factor blends each region's typical mode mix — car-heavy in North America, rail-heavy in Europe, mixed in Asia Pacific.
3 · Our server footprint (kg CO₂e)
interviews × 0.42 h × 0.5 kW × 1.2 PUE × regional grid factor
0.42 h is the average 25-minute interview; 0.5 kW is our per-interview compute draw — a conservative internal engineering estimate under periodic review against actual infrastructure usage, not a measured value; PUE 1.2 covers data-centre overhead. Also reported gross as "Your AI footprint" — see below.
4 · Net CO₂ avoided
transit emissions avoided − our server footprint (never below zero)
This is the only carbon figure we report as "avoided".
5 · Team time saved (hours)
interviews × 1.15 h × panel size
1.15 hours covers scheduling, hosting, and note-taking per panelist per in-person first round.
6 · Candidate time saved (hours)
interviews × (round-trip distance ÷ regional travel speed + 0.5 h waiting)
7 · Cost savings (US$)
team hours × regional team rate + candidate hours × regional candidate rate
8 · Paper saved (pages)
interviews × 2 pages × panel size
9 · Everyday equivalents
cars = net tonnes ÷ 4.6 · trees = net kg ÷ 21 · FTE years = team hours ÷ 1,880
Reporting your AI footprint
What the footprint figures cover
Included: the estimated server electricity used to run your interviews (formula 3's energy term: interviews × 0.42 h × 0.5 kW), a 1.2× PUE uplift for data-centre overhead (cooling, power distribution), and conversion to CO₂e using your selected region's grid carbon intensity. Reported as total kWh, total CO₂e, and per-interview intensity.
Excluded: candidate and interviewer devices, network transmission, and embodied emissions of hardware — immaterial at interview scale and not reliably attributable per interview.
Gross vs. net: the "Your AI footprint" section and the PDF report show this footprint gross, so your ESG team can lift it directly into an AI-usage disclosure. Everywhere savings are reported, the same footprint has already been subtracted — nothing is counted as avoided until our own compute is paid for.
Regional parameters
| Parameter | N. America | Europe | Asia Pacific | Basis |
|---|---|---|---|---|
| One-way distance (default) | 32 km | 18 km | 22 km | Typical urban commute ranges; editable in the calculator. |
| Transit factor (kg CO₂e/km) | 0.21 | 0.12 | 0.14 | Blend of mode-specific factors from the UK Government conversion set (average car ≈ 0.17, rail ≈ 0.035, bus ≈ 0.09), weighted by each region's typical mode mix. |
| Grid intensity (kg CO₂/kWh) | 0.38 | 0.23 | 0.55 | US EPA eGRID; European Environment Agency; APAC blend of Singapore EMA (≈0.41), India CEA (≈0.71), Australia NGA (≈0.66) and IEA/Ember data. |
| Team rate (US$/hour) | 55 | 45 | 25 | Loaded hourly cost of hiring-team time, benchmarked to regional salary levels. Interviewer.AI assumption. |
| Candidate rate (US$/hour) | 22 | 18 | 10 | Value of candidate time, benchmarked to regional wage levels. Interviewer.AI assumption. |
| Travel speed (km/h) | 50 | 38 | 32 | Typical door-to-door urban travel speed by dominant mode. Interviewer.AI assumption. |
Sources
Greenhouse Gas Conversion Factors for Company Reporting
UK Government — DESNZ / Defra
The annually updated, publicly documented factor set behind our transit emissions: average car ≈ 0.17 kg CO₂e/km, national rail ≈ 0.035, bus ≈ 0.09. We blend these into one factor per region based on its typical mode mix.
gov.uk/government/collections/government-conversion-factors-for-company-reporting
eGRID — Emissions & Generation Resource Integrated Database
US Environmental Protection Agency
Source for North American electricity carbon intensity (≈ 0.38 kg CO₂/kWh national average), used to price our server footprint when you select North America.
Greenhouse Gas Emission Intensity of Electricity Generation
European Environment Agency
Source for the European grid factor (≈ 0.23 kg CO₂/kWh EU average).
Asia-Pacific national grid factors
Singapore EMA · India CEA · Australian Government · IEA / Ember
Our Asia Pacific grid factor (0.55 kg CO₂/kWh) blends published national figures: Singapore Energy Market Authority (≈ 0.41), India Central Electricity Authority CO₂ Baseline Database (≈ 0.71), Australia's National Greenhouse Accounts factors (≈ 0.66), cross-checked against IEA and Ember regional data.
ema.gov.sg · cea.nic.in · dcceew.gov.au · iea.org · ember-energy.org
Greenhouse Gas Equivalencies Calculator
US Environmental Protection Agency
Basis for the "cars off the road" equivalent: a typical passenger vehicle emits ≈ 4.6 tonnes of CO₂ per year. The tree equivalent uses the commonly cited ≈ 21 kg of CO₂ absorbed per mature tree per year, consistent with EPA tree-growth equivalency data.
Data-centre efficiency (PUE)
Google Data Centers · Uptime Institute
We assume a Power Usage Effectiveness of 1.2 — deliberately less optimistic than Google's published fleet-wide average (≈ 1.1, where our workloads run) and well below the Uptime Institute's reported industry average (≈ 1.55), so our server footprint errs on the side of overstatement.
google.com/about/datacenters/efficiency · uptimeinstitute.com
Interviewer.AI assumptions
Internal — conservative and configurable
Per-interview compute draw (0.5 kW — a conservative internal engineering estimate under periodic review against actual infrastructure usage, not a measured value), team time per panelist (1.15 h), candidate waiting time (0.5 h), regional travel speeds, hourly rates, pages per résumé (2), and the 1,880-hour working year are Interviewer.AI engineering and market estimates. Each is chosen conservatively, documented here, and adjustable for enterprise deployments.
Assumptions & limitations
What's counted: avoided candidate round trips, avoided team and candidate hours, avoided printing, and our own server electricity (reported gross as your AI footprint, and subtracted before anything is reported as avoided). What's not counted: avoided meeting-room energy, candidate devices, and network transmission — excluding the first understates your savings; excluding the last two is immaterial at interview scale and keeps the model simple and defensible.
Published factors are updated periodically by their issuers; we review our defaults against the latest releases annually. All figures are modeled estimates from your inputs and published regional averages — not metered measurements. Our methodology and factor sources are fully documented on this page for use in ESG reporting.
Ready to see your own numbers?
Run the calculator, or talk to us about your hiring and sustainability goals.