Implementation Tiers and Data Quality
In short: you do not need perfect data to calculate a valid SCI for Web score. You need to say honestly how you got each number. This page describes three levels of data maturity (Entry, Standard and Advanced) as a way to plan improvements. They are guidance, not something the specification asks you to declare.
Tiers are guidance, not a requirement
The original draft required every organisation to select an implementation tier and report it. The proposed specification does not: there is no tier-selection step in the Clause 4 procedure and no tier item in the Clause 8 reporting list. What the specification does require is disclosure, for each component, of whether it was measured or calculated and which data sources, models, coefficients and allocation methods were used (Clause 8 item 5). Clause 10 adds that a score can always be calculated without cost, using modelled data where measured data is unavailable, provided the method is disclosed.
The tiers remain useful as a description of maturity. Whether "tier" should come back as an optional disclosure label is Open Question 1. Until it is settled, you may mention a tier in your report as an optional, informal label.
Tiers are always named, never numbered. The original draft used "Tier 1" to "Tier 3" for client-side measurement approaches, and the SWI guidance uses "Tier 1" for its best data, so numbers invite confusion.
Entry
For: teams beginning with SCI for Web, with limited measurement infrastructure.
Typical practice:
- automated measurement using freely available tools;
- server energy measured directly where possible (cloud provider APIs, infrastructure monitoring);
- client energy measured on reference devices, at least one per major device category;
- third-party services estimated with conservative defaults, or excluded within the Clause 7.6 limits;
M_networkpopulated with industry-default factors and a non-bytes allocation basis, or declared unpopulated only with a materiality assessment showing it below 5 % of embodied emissions (Clause 7.5);- annual-average carbon intensity;
- a measurement period of at least 7 consecutive days (see Measurement Periods and Traffic);
- full disclosure of methods, tools and limitations.
Typical data sources: cloud provider energy or carbon APIs; browser developer tools for client profiling; reference values for device embodied emissions; industry-default coefficients for third-party services.
Standard
For: teams with an established measurement practice seeking better accuracy.
Typical practice:
- server and client energy measured directly across representative scenarios;
- client measurements weighted by device mix, with at least two devices per major category;
- third-party services included using vendor data or a documented estimation method;
M_networkcalculated from industry-default factors with a disclosed non-bytes allocation basis;- carbon intensity at daily granularity or finer;
- a measurement period of at least 30 consecutive days or one complete business cycle;
- results checked against published benchmarks where available;
- full disclosure of methods, data sources and uncertainty.
Typical data sources: infrastructure instrumentation and telemetry; browser profiling across device types; vendor disclosures or API-based estimates for third parties; regional grid carbon intensity data (see Carbon Intensity and Data Sources).
Advanced
For: organisations with sophisticated measurement capability and sustainability commitments.
Typical practice:
- continuous measurement of server-side infrastructure;
- real-user monitoring for client energy, with statistical sampling;
- comprehensive third-party inclusion, with vendor engagement;
- hourly carbon intensity, enabling carbon-aware optimisation, with geographic and temporal distribution;
- continuous measurement with regular (monthly or quarterly) reporting;
- independent validation or audit of the method;
- public disclosure of method, data sources and results;
- device-specific lifecycle assessment data for embodied emissions where available.
O_network at every tier
O_network is optional at every tier: no level of maturity implies
that it is expected. If you choose to report it, the method follows the
Clause 7.3 hierarchy whatever your tier: direct measurement first, a
hybrid model second, and a data-transfer model only where neither is
available. Better data simply makes the more direct methods reachable.
See Network Emissions.
The original draft listed "industry default coefficients for network
energy" as an Entry-tier requirement and "SWDM or equivalent" as a
Standard-tier requirement, implying that O_network was expected. These
guidelines remove that implication.
Methods by tier
Clause 8 item 5 requires you to disclose, for each component, whether it was measured or calculated. This table shows typical choices at each level:
| Component | Entry | Standard | Advanced |
|---|---|---|---|
O_server | Measurement or calculation | Measurement preferred | Continuous measurement |
O_network (optional at every tier) | If reported: Clause 7.3 hierarchy | If reported: Clause 7.3 hierarchy | If reported: Clause 7.3 hierarchy |
O_client | Measurement (reference devices) | Measurement (device mix) | Measurement (real-user monitoring) |
M_server | Calculation (defaults or LCA) | Calculation (LCA databases) | Calculation (specific LCA) |
M_network | Calculation (industry defaults), or unpopulated with a materiality assessment below 5 % | Calculation (industry defaults) | Calculation (vendor-specific) |
M_client | Calculation (defaults) | Calculation (device-specific) | Calculation (device-specific LCA) |
Third-party emissions are sub-totals within the operational components, not a separate row. Development lifecycle emissions are assessed separately and never added to the score; see Development Lifecycle Emissions.
Carbon intensity granularity by tier
| Tier | Typical temporal granularity |
|---|---|
| Entry | Annual average |
| Standard | Daily average or finer |
| Advanced | Hourly |
Clause 8 item 7 requires the spatial and temporal granularity of your carbon intensity data to be disclosed, whatever you choose.
Please submit any comments you have here.