SparkMeter
0-1 SaaS For Top Innovative Company with 100,000+ Smart Meters from Rural to National Grids
Role: Service Designer
Duration: January 2021 - February 2023
Industry: Renewable Energy
Market: African and Asian Developing Countries, The US
Challenge: Migration, Discovery, Design Sprint, Design System, Product Design, Offline App
Troubleshooting branch power losses in field, credit: SparkMeter
Overview
SparkMeter is a SaaS startup on affordable renewable electricity to underserved small-scale mini-grids in rural areas and distribution utilities that are part of national grids. The company produces smart meters and helps modernize cloud-based grid-management software and advanced metering infrastructure (AMI) systems mostly in African, Asian, and the US areas. With 100,000+ smart meters for 500+ deployments sold in 30+ countries, SparkMeter’s plug-and-play technology supports flexible billing systems and real-time remote asset monitoring even with no internet connectivity.
I joined SparkMeter as a Service Designer when they migrated users from outdated base stations to new hardware (and related software, of course). Next, we discovered new market opportunities for energy distribution and scaled up Koios, a platform for mini-grids, to strategic the 10-module UDSM suite for distribution utilities, with GridScan and GridFin tools.
Outcomes
2021 – the top 10 most innovative energy companies with the smart metering system Koios for developing countries
2022 – scaled technology beyond SparkMeter’s hardware led to raising $10M
2023 – another funding round for a product-market fit test of the UDSM suite, a SaaS concept, and its success in the Distributech conference
Links
Mini-Grids: Delivery First
Yes, we moved directly to the development. The Double Diamond doesn’t work here, we had a month to provide ready-for-dev basic prototypes. It was an urgent need when the startup faced the delayed commitment in face to investors for Koios + Nova release. This happened because the previous design agency spent a lot of time to understand the software logic. So my journey as the only designer in the team started with some low-fidelity wireframes and the simple UI kit at hand. And SparkMeter was an engineering-driven company initially by the way, so we didn’t have access to users. This process was as much about collaborative iteration as it was about rapid solution demanded.
Challenge
In a small village far from the national power grids, access to electricity is isolated in their own generation source, branch management by totalizers, and smart meters for end-point customers. It is called mini-grids – small-scale power grids that generate and distribute electricity to serve rural areas, typically for Sparkmeter at that time, in Africa.
Imagine a boy in Nigeria who wants to prepare a meal but can’t afford even $5 a month for power, it’s impossible cost for a household, so they aim to steal electricity
For local vendors with a mission to empower communities, electricity theft and lacking tools to efficiently manage systems are daily realities, so they collect the meter reading manually once per several months
Technical operators are from the same village where the mini-grid is set up and often are barely certified to install and troubleshoot smart meters and base stations, and they don’t speak English
Initial SparkMeter’s MVP equipment and software are not efficient enough which slowed response times for proactive and reactive interactions in off-internet areas, so grid maintenance is costly and time-consuming
Hypothesis
How might we migrate mini-grid operators from Thundercloud to Koios to simplify small-scale grid operations and consumer access by improving energy management, enabling remote monitoring anywhere, and integrating seamlessly with local payment systems in low-income communities?
SparkMeter customers install smart meters to end users and grid monitors for branches to read power metrics, like consumption and losses, credit: SparkMeter
Diagram of asset relationships in SparkMeter to illustrate data and operational flow, credit: SparkMeter
Thundercloud system we had to redesign to migrate customers to renewed Koios and Nova interfaces, credit: The Ptolemy Project
The Koios’s redesigned interface, credits: SparkMeter
Audit of Koios and Nova interfaces uncovered many inconsistencies not just in UI but in entire user flows of reading data and managing assets, mainly because Nova runs autonomously of Koios cloud, credit: SparkMeter
Design System
One of the first tasks was to set up a solid visual language. The initial timeline was tight, with overlapping design and development needs, so I ran in parallel mockup and components creation. I crafted the atomic design system from scratch – from typography layout to complex inputs and controls. The main issue was legacy styles implemented on the platform, so it’s difficult to achieve pixel-perfect development. I included in the design system the snapshot catalog and content strategy guidelines to manage communication on updates with developers.
Atomic approach from styles to complex components for dashboard widget, credit: SparkMeter
Content guidelines for design and development, credit: SparkMeter
High-fidelity mockups in snapshot catalog for up-to-date pages, credit: SparkMeter
The Main MVP Feature: Tariff Management for Pre-Paid Billing
The payment flow was the most important for the seamless start. Pre-paid low-income consumers were in the majority within mini-grids in rural African areas. Traditional billing methods fall short there due to the high meter reading costs and bill collection. The pre-paid model makes it easier for technicians to monitor the grid remotely and minimizes the need for on-site interventions.
So the system allows users to purchase electricity credits in advance, and this balances energy consumption with their financial capabilities. Consumers can buy electricity credits via mobile money and understand their balance via SMS messages because they live typically in areas with limited banking infrastructure. When a consumer's credit is exhausted or any meter invasion is detected, the system can remotely turn off smart meter service, and it can be reconnected promptly upon credit top-up.
Payment delays occur as the Nova base station syncs with Koios in the cloud not in real time, credit: SparkMeter
Examples of data with different communication requirements, credit: Intechopen
SME Interviews: Migration from Thundercloud to Koios+Nova
SparkMeter grew up. But the DA1000 infrastructure lacked the capacity to handle a large-scale API integration and support for 1,000+ meters for a single site, and there was a need to launch several base stations to exchange the data from all meters in a grid. The limitation was a bottleneck in scaling and maintaining consistent performance to support real-time data syncing and high-load environments. There was also extensive validation of feature compatibility with Nova and Koios as migrating customer data, tariff structures, and historical usage records from legacy systems to avoid risks of data corruption and loss.
We interviewed 5 SparkMeter’s customer support experts to learn about typical workflows within Koios and Nova software, some interviews took several meeting to complete. Customer training sessions were conducted by support, paired with interactive demos and guiding materials, to bridge the knowledge gap and give more confidence in using the new tools. A phased data migration strategy was executed in stages with sandbox environments for testing, data mapping, and verification protocols as downtime could disrupt critical operations for energy providers.
Site migration lifecycle, credit: SparkMeter
User flow of setting payment permission to sales account, credit: SparkMeter
The Main Post-MVP Feature: Large Deployments
Sprint by sprint, we transitioned SparkMeter’s maturity from an engineering-focused to a user-facing vibe. And client demand increased as well. Large deployments require complex hierarchies, such as multiple base stations and sub-branches, and careful mapping. As energy providers’ size could be thousands of meters, the systems managing these meters must support bulk configuration, provisioning, and monitoring without performance degradation. This time, we prioritized user needs and experiences from the available knowledge from proxy stakeholders and customer support experts, still with limited access to end-users. Our team – including the COO, Director of Product, Product Manager, and myself as the solo designer – experimented a lot to refine user flows based on collected customer insights in Productboard and best market practices.
Early mockup from a brainstorming on large-scale deployment management, credit: SparkMeter
Concept Test: Installing High-Capacity Meters
Unlike standard meters, high-capacity meters often require specialized training for installation teams because phase alignment and upstream/downstream component connectivity can complicate deployments. Initially, meter configuration was provided by SparkMeter during manufacturing based on customer company details in Salesforce. However, due to reducing delivery time for a device set from several months to several weeks, all configuration should be conducted in fields by technicians while installing meters. Field conditions, such as limited internet and power availability, can impact provisioning and performance testing, so the Product team proposed technicians an innovative solution – mobile provisioning devices (MPDs).
It was required to learn from real technicians the concept validity of configuring meters during installation by uploading pre-defined settings such as tariffs, communication parameters, and unique meter IDs. It makes sure proper integration with the grid management system by searching connectivity, synchronizing data, and performing initial functionality diagnostics. If there is no internet, the MPD facilitates offline provisioning, storing data locally until connectivity is restored. This resolved the main usability problem – field technicians no longer need to climb on a pole with a laptop, they can configure meters with a small black device.
Interconnection between smart meter, mobile provisioning device, and application which is installed in field technician’s phone, credit: SparkMeter
We started the mobile app design as a cross-platform platform, but typically field technicians use Android, credits: SparkMeter
Key Achievements for Mini-Grid Solutions
Operators access real-time data with remote diagnostics, which reduces troubleshooting time by 40% for small and large customer bases
Vendors’ collection rates are improved with clear pre-pay systems via local mobile money
Meters could be connected to operations in minutes even in low-connectivity areas
Performance Insights on consumption data are comprehensive enough to predict grid maintenance
Automated alerts in the Live Status widget ensure consistent energy delivery and in-time reaction by remote operation teams
Customer-facing summary about Koios capabilities, credit: SparkMeter
Distribution Utilities: The Discovery Foundation
As SparkMeter became mature with new investments, they saw an opportunity to expand into urban markets to support full-scale distribution utilities (DUs) connected to national grids. While mini-grids serve small, remote communities, distribution utilities operate on a much larger scale, managing complex, high-capacity grids that power urban and semi-urban industrial areas. Typically, such grids require scalable solutions for analytics, outage management, and billing. SparkMeter aimed to expand into this market with tailored SaaS tools called UDSM (Unified Distribution System Management). We collaborated to explore two of the modules – GridFin and GridScan.
Challenge
As we transitioned from mini-grids to large-scale distribution utilities, the complexity of the problems made us think about the different services:
Legacy platforms with segmented decades-old systems struggle to manage thousands of consumer accounts with several meters connected to each
Many utilities combine smart and non-AMI meters, so teams rely on extensive manpower for routine tasks, like field inspections and data collection
DUs struggle with technical and non-technical energy losses, like consumer theft and inconsistent loads because of outdated infrastructure, which directly impacts revenue by lack of real-time insights into grid performance
Teams lack advanced analytics for collecting and visualizing grid data, so it is difficult to forecast demand, address outages in time, and optimize resource allocation
Adapting to changing policies and regulations with environmental and performance standards in different regions adds a layer of complexity, especially when systems are not flexible enough for updates
Fragmented communication channels and a lack of real-time notifications prevent utilities from engaging new and existing consumers, which causes higher churn rates
Hypothesis
How might we modernize operations for distribution utilities with outdated infrastructure, fragmented tools, and limited data visibility to scale workflows and reduce energy losses so they can satisfy customers with low and high consumption and meet regulatory demands?
Typically, distribution utilities are parts of national lines and provide energy for manufactures and urban blocks, credit: SparkMeter
Persona and Service Blueprint Exploration
Together with Sparkmeter’s sales, we interviewed in-depth more than 200 CTOs and senior managers to understand key roles, workflows, challenges in distribution utilities, and their differences from mini-grid structures. We collected more than 1500 contacts based on their LinkedIn profiles and selected the higher roles in companies we were curious to learn more about. No matter of company size, there are commercial and operational roles. Typically on the commercial side, there are acquisition agents, sales teams, and local vendors. On the technical operation side, there are field technicians, operation and maintenance experts, data analysts, and sometimes IT support.
To visualize role hierarchies, team interdependencies, and communication flows, we compared roles within typical organograms to identify real SparkMeter user roles. To document end-to-end operations, like user touchpoints, tools, and backstage processes, and provide long-term opportunities for SparkMeter, we mapped 2 service blueprints to showcase the difference between mini-grid and distribution utility in strategic and operational requirements.
Persona template on user roles within organizations and their roles within SparkMeter products, credits: SparkMeter
Exploration of typical roles for energy companies, credit: Solar career map
Organograms to understand better how distribution utility needs are different from mini-grids, credit: SparkMeter
20+ user roles combined by journeys and workflows into service blueprints, credit: SparkMeter
Design Sprint to Address Losses in Distribution Utilities
We involved the main SparkMeter’s Product and Engineering stakeholders in collaborative workshops. The design sprint aimed to create within 6 months an actionable solution for DUs in low or lower-middle-income countries to identify and address customer theft on a large scale. The initial design sprint phases – Understand, Define, and Diverge – took 3 days while others – Prototype and Validate – were conducted over several weeks to properly iterate concept flows based on user feedback.
As a preparation for the design sprint, Director of Product and I conducted a detailed competitive feature walkthrough to identify our solution gaps and opportunities for differentiation
The core team helped us to clarify SparkMeter’s values, solution metrics, technical capabilities, and DU roles insights
During workshops, we designed intuitive user flows for grid operators to quickly identify and resolve suspected theft tickets
Finally, the core team was present during concept testing with 6 O&M leads and data analysts from selected client companies to align with real-world user needs
Design sprint canvas in Miro board, credit: SparkMeter
Low-fidelity concept flow of reporting theft while field technicians inspected meter, credit: SparkMeter
Low-fidelity concept page of theft monitoring on the end user level, credit: SparkMeter
GridScan
While Koios provides customer-level readings for immediate operational decisions, GridScan offers system-wide analytics on grid performance for strategic planning. GridScan calculates technical and non-technical losses in overall distribution lines dashboards, outages and voltage issues with real-time SAIDI and SAIFI metrics, energy delivered, and operational performance. With GridScan, data analysts easily handle growing demands from EVs, rooftop solar, and other electrification projects. Users also access geographic heatmaps with forecast-based transformer capacity and upgrade recommendations based on automated forecasts.
Smart meter network system architecture, credit: Intechopen
General dashboard to learn about estimated losses, credit: SparkMeter
GridFin
When Koios focuses on setting and applying basic tariffs for individual customers, GridFin enables tariff strategies for consumer groups with dynamic adjustments, forecasting, and detailed financial reporting to optimize revenue across the entire utility network. There are tools for payment tracking with automated alerts for overdue accounts and dynamic tariff adjustments based on consumption patterns, regulatory changes, or market demands. GridFin supports both prepaid and postpaid billing models with minimum revenue risks. Commercial staff can also generate financial summaries to forecast and improve cash flow.
We validated GridFin concept during the Distrubutech 2023 conference in Texas by showcasing its capabilities to industry leaders and potential clients. Huge success of the concept has proved our right track to the innovation of advanced revenue tracking in the challenges faced by modern DUs. You can dive into details to explore a Figma prototype presenting the GridFin in the flow where analysts modify a tariff rate scenario.
Custom rate creation in GridFin, credit: SparkMeter
Key Achievements for Distribution Utilities Solutions
Implemented role-based permissions and workflows in UDSM tools for commercial and operational strategies across thousands of endpoints
Optimized resource allocation and minimized system downtime with analytics for capacity planning and predictive maintenance
Reduced theft and inefficiencies for DUs by quantifying technical and non-technical energy losses with real-time data from transformers and meters
Automated revenue tracking and flexible tariff management improved cash flow and reduced overdue accounts for large-scale utilities
Integrated existing DUs’ reporting tools into the suite to adapt to maintain accountability with minimal operational disruption
Customer-facing summary about distribution utilities solutions, credit: SparkMeter
Lessons Learned
Start Research Early
We had to restructure several features after the post-release customer feedback in the early ages. Limited access to end users delayed critical insights, but we were lucky to have at least proxy stakeholder feedback for redesigns. In-depth interviews with users is essential to align complex solutions with real-world needs and avoid inefficiencies as soon as possible.
Simplify User Workflows
Early assumptions often missed operational nuances so unique complex workflows caused user confusion. Breaking processes into simple actionable steps and testing concepts early reduced support needs as products are easy to adopt. By creating tiered system access, we secured compatibility with both small mini-grids and large DU tasks.
Balance Innovation with Daily Routine
Interactive workshops were more effective than static demos for aligning stakeholders and validating product assumptions. Regular retrospective meetings helped refine processes, avoid repeating mistakes, and maintain clear communication between teams. Open feedback loops and co-creation align better project goals with daily challenges while discovering and implementing solutions.
