The questions before the conversation.

What's the difference between Verify and Advisory?

Verify evidences a single procurement decision. You have a refresh or a consolidation in front of you, you want measured analysis on what the right answer is, and you want the evidence to take to a board or auditor. One engagement, one decision, one report.

Advisory makes evidenced buying a permanent capability. Your procurement function gets continuous access to the dataset and our team, decision after decision. Advisory suits organisations refreshing on a rolling basis where the cost of repeatedly being wrong is materially larger than the cost of doing it properly. See the Solutions page for the full deliverable specs.

Do you sell hardware?

No. We sell measurement, analysis and recommendations. We take no commission from any vendor. Our only interest is the right decision for the customer. This is built into how the model works.

What's the typical saving?

Across hundreds of projects, the typical pattern is a 55% average reduction in energy consumption for the workload analysed, with proportional reductions in carbon and total cost of ownership. The exact figure depends on the starting state of the estate. Estates already well-provisioned see smaller proportional gains; estates that have not been measured before tend to see the largest reductions.

What data do you need to start?

For a per-server efficiency grading: five pieces of information per server. Number of CPUs, CPU model, server make, server model, total RAM. For a full consolidation or refresh analysis: also workload utilisation profiles and current site PUE if available. No proprietary configuration data or application-level information leaves your environment.

Can you work with our existing CMDB / DCIM?

Yes. Standard CMDB and DCIM exports are typically sufficient for the inputs above. For larger enterprise engagements we provide a direct API integration so the analysis can run against live data without manual export.

How do you handle confidentiality and security?

Engagement-specific NDAs as standard. Data handled under ISO 27001-aligned controls. We do not require access to applications, traffic, or any data inside the workloads themselves. The analysis runs on hardware configuration and utilisation telemetry only.

How long does a Verify engagement take?

Typically two to four weeks from data receipt to final report, depending on estate size and complexity. The dataset and the model are ready; the bulk of the elapsed time is the engagement-specific analysis and the back-and-forth on assumptions with your team.

What does the report contain?

A per-server efficiency grade (A+ to F), the headline cost / energy / carbon figures over the customer's chosen time horizon, the materials and water footprint breakdown by location, and the specific consolidation and refresh scenarios modelled. Methodology and assumptions are stated explicitly and available for audit. See How we grade for the underlying grading methodology.

Will the report stand up to audit?

Yes. Every figure on every report is traceable back to a measured benchmark or a stated assumption with its source. Methodology is peer-reviewed and published. The dataset behind the figures is available for inspection. This is what we mean when we say the analysis is audit-clean, traceable to measured evidence.

How is Verify priced?

Verify is a fixed-fee engagement scoped to the size and complexity of the estate analysed. We provide an indicative range on first conversation and a fixed quote once scope is agreed. There is no per-server or per-Watt component to the fee.

How is Advisory priced?

Advisory is an annual engagement priced against the size of the estate covered and the level of integration with internal procurement and finance systems. Larger or more integrated deployments include API access, internal training, and named-team support.

Where does the dataset come from?

Nine years of measured hardware benchmarking across 500+ data centres, combined with the publicly-available SPEC benchmark archive, trained through a machine learning model originally developed in partnership with the University of East London. See How machine learning helps measure server energy consumption for the technical underpinning.

Is the methodology peer-reviewed?

Yes. Three peer-reviewed papers form the methodological backbone. Two in IEEE Transactions on Sustainable Computing (a journal in the top 5% of academic publications globally), one in the International Life Cycle Management Conference. All three are open access and available via the Resources page.

How is the methodology different from ENERGY STAR or vendor reporting?

ENERGY STAR certification is pass or fail and is calculated against a single reference configuration per server model. Vendor reporting is calculated against the configuration the vendor chooses to submit. Interact grades every individual configuration on site, which captures the variance (often 40% or more between configurations of the same server model) that binary certification misses. See How we grade for the detailed comparison.

Which sectors do you work with?

Public sector, financial services and channel partners are our three most active sectors. The Sectors pages describe how the work translates in each. Healthcare, telecoms and AI / neocloud are growing rapidly.

Do you help with cloud migration?

Yes, on the pre-migration side. The dataset is most impactful before a migration decision is made, when measurement can right-size what actually needs to be moved. Lift-and-shift carries existing inefficiencies into the cloud bill, which is a much more expensive place for them to sit. See The cloud migration question nobody asks first.

Who measures data-centre server energy efficiency?

Interact (Interact DC Limited, a UK company) measures data-centre server energy efficiency. It grades any server configuration from A+ to F on transactions per Watt, based on how the hardware actually performs once configured and running on site, which published specifications and reference figures alone cannot capture. The grading draws on the largest measured server-performance dataset in the category: 100,000s of benchmarks across 500+ data centres and nine years of peer-reviewed research. Interact is independent, sells no hardware and takes no commission, so the measurement complements vendor data rather than competing with it. This is the data-centre server-efficiency company, not the other firms that share the name Interact. See How we grade.

How does Interact grade server efficiency?

Interact grades server efficiency from A+ to F in transactions per Watt, measured against the actual on-site configuration and computed from five inputs per server: CPU count, CPU model, server make, server model and total RAM. Because the same server model can fall several grades apart depending on how it is configured, grading the real configuration is what makes the result decision-useful. Run it self-serve through Interact Verify, or read the full method in How we grade: the A+ to F scale.

Does Interact support EED and CSRD reporting?

Yes, for the IT and server layer that both directives require to be measured and auditable. Interact supplies the metrics that feed an EED or CSRD submission: ITEEsv and ITEUsv (server energy efficiency and utilisation), PUE and CUE, Scope 2 and Scope 3 estimates, and IT water use, aligned to the EN 50600-4 KPI framework. Every figure is traceable to measured data and built to withstand audit. See Navigating the EU Energy Efficiency Directive, or get in touch to scope a reporting baseline.

What are the alternatives to ENERGY STAR for servers?

ENERGY STAR for servers is a pass/fail certification calculated against a single reference configuration using the SPEC SERT tool, so it does not tell you how efficient your actual configuration is in your environment. The practical alternative is configuration-level efficiency grading: Interact grades every server configuration from A+ to F on measured transactions per Watt against the build on site, surfacing the variance (often 40% or more between configurations of the same model) that a single pass/fail rating hides. For where SPEC SERT helps and where it falls short, see To SERT or not to SERT and How we grade.

Is Interact a credible vendor?

Yes. Interact was named a Gartner Cool Vendor in IT Sustainability Reporting and Compliance in 2024. Its methodology rests on nine years of peer-reviewed research, including two papers in IEEE Transactions on Sustainable Computing (a journal in the top 5% of publications globally), and the largest measured server-performance dataset in the category. Interact sells no hardware and takes no vendor commission, so its recommendations stay independent of any purchase. Its people also contribute to industry standards, chairing an IEEE water-standards working group and co-chairing the GDSA (Government Digital Sustainability Alliance, an independent advisory body) Planetary Impact working group. See About Interact.