02Product strategy
Delta Rising Foundation
- Role
- Product Strategy Consultant
- Dates
- February 2025 – August 2025
- Location
- Remote
This is the one project where I did the diagnosis, the product decisions and the go-to-market myself. Product strategy for Redwood Big Tree ID, an AI climate tool, across product management, strategy and GTM.
It rests on one market finding. Existing carbon measurement, reporting and verification platforms run on modelled and remotely-sensed data, and the field data underneath comes overwhelmingly from managed plantations — so the most carbon-dense forests are the least well measured. The product sits in that gap, and everything else followed from it.
+30%
Sprint velocity, as acting Scrum Master on a 14-person cross-functional AI team
15%
Cut from GTM launch-cycle time
Reading the market
Competitive analysis of the carbon MRV market
I read every platform through one consistent frame — value proposition, mechanism, product and data type, funding position, pricing model, revenue strategy, valuation — so the market could be compared like for like, and a gap would show up as a gap rather than as an opinion.
The product needed a defensible position in carbon measurement, reporting and verification, which meant establishing what existing platforms actually measure and where their underlying data comes from.
The analysis fed the product's positioning and its pricing approach, and it produced the finding the whole product rests on: because field data comes overwhelmingly from managed plantations, the most carbon-dense forests are the least well measured.
Read the approach
Problem
A climate product cannot claim a position it has not earned. Carbon measurement, reporting and verification is a market where platforms describe themselves in similar language, and where the interesting differences sit underneath that language — in what is actually measured, and in where the data that supports the measurement comes from.
Approach
I built the competitive analysis on a single frame and applied it to every platform without exception: value proposition, mechanism, product and data type, funding position, pricing model, revenue strategy, valuation.
Holding the frame constant is what makes an analysis like this usable. Comparing platforms on whichever dimension each one happens to lead with produces a summary of their marketing. Comparing them on the same seven dimensions produces a map, and a map has holes in it that mean something.
What it produced
The hole in the map was the data layer. Platforms run on modelled and remotely-sensed data, and the field data calibrating those models comes overwhelmingly from managed plantations — even, uniform, accessible stands that are straightforward to survey.
Old-growth forest is none of those things, which is why so little field data exists for it, and why the forests holding the most carbon are the ones the market measures worst. That asymmetry is the product's reason to exist, and it is what the positioning and the pricing approach were built on.
Building the product
Roadmap, prioritization and delivery
There was no roadmap, no KPI framework and no agreed basis for deciding which feature came next.
15%
cut from GTM launch-cycle time
I built the roadmap and the KPI framework from scratch, and underneath them a quantitative feature-prioritization model — so the order of work followed from the model rather than from whoever had the floor.
Sprint velocity rose 30%. I was acting Scrum Master on a 14-person cross-functional AI team, running the sprints with the engineers as product manager.
Read the approach
Problem
The product had a direction and no machinery. Nothing recorded what would be built in what order, nothing defined what success would look like when it was, and every sequencing decision was therefore an argument rather than a calculation.
Approach
I built the roadmap and the KPI framework from scratch, then a quantitative feature-prioritization model underneath them, so that sequencing had a defensible basis rather than resting on whoever made the case most forcefully.
I ran the sprints with the engineers as product manager, and took the acting Scrum Master role on a 14-person cross-functional AI team — which meant the prioritization model was not a document handed over at the start but the thing actually deciding what the team picked up each sprint.
What it produced
Sprint velocity rose 30%, and launch-cycle time for go-to-market came down 15%.
Both followed from the same change. A team that does not have to re-litigate what matters at the start of each sprint spends that time building instead, and a launch that is not waiting on a sequencing decision ships sooner.
Reaching the funders
Funder targeting through to grant submission
Reaching capital for a climate product means reaching particular people inside particular institutions, not the generic inboxes a funder list gives you.
139,000+
acres of redwood forest the solution is projected to conserve
Scope, not achievement — the area the tool would serve, in Redwood National and State Parks.
I scoped the population of capital sources, screened it on scale and mission fit, then used Clay waterfall enrichment to resolve each remaining institution down to the specific program officer whose remit actually covered the work.
Outreach ran through to grant submission — from a population of capital sources down to the named individuals whose remit covered the work.
Read the approach
Problem
A funder universe is easy to assemble and almost useless in that state. Scale and mission fit eliminate most of it, and what survives still resolves to an institution rather than to a person — and institutions do not read proposals.
Approach
I scoped the population of capital sources and filtered it on scale and mission fit, then used Clay waterfall enrichment to resolve the remainder down to the individual program officer whose remit covered this kind of work.
What it produced
Outreach ran through to grant submission.
What the filtering bought was not a shorter list but a list of people. A funder universe resolves to institutions, and institutions do not read proposals — program officers do, and only the ones whose remit already covers the work.
Designing the discovery
A second direction: California agricultural AI
I mapped stakeholders well past the end user — agronomists, university extension, water organizations, labor representatives, rural development groups — and wrote questions capable of disconfirming the idea rather than validating it, pairing surveys for breadth with interviews for depth.
A separate California agricultural-AI direction needed discovery before it needed a product — and the easiest discovery to run is the kind that collects agreement with the idea you already hold.
I flagged workforce impact as a consideration before anyone asked me to. On a product aimed at agricultural labour, raising that unprompted is the difference between a discovery process and a sales process.
Read the approach
Problem
Discovery is the cheapest part of a product to fake. Ask the end user whether they would like a tool that solves their problem and they will say yes, and the answer will have taught you nothing, because the question could only ever have returned one result.
Approach
I mapped the stakeholders well past the end user: agronomists, university extension, water organizations, labor representatives, rural development groups. An agricultural tool lands in all of those relationships whether or not anyone designed for them, so leaving them off the map does not make them absent — it only makes them a surprise later.
Then I wrote the questions to be capable of disconfirming the idea. That is a specific discipline: a question that cannot return an answer you did not want is not a question. Surveys carried breadth and interviews carried depth, because the two failure modes are different — a survey misses the thing nobody thought to ask about, and an interview misses how common it is.
What it produced
A discovery strategy that could have killed the direction, which is the only kind worth running.
I also flagged workforce impact as a consideration before anyone asked me to. On a product aimed at agricultural labour, that is not a compliance step at the end; it is one of the things discovery is for, and raising it unprompted is the difference between a discovery process and a sales process.
Built with
- Clay
- Agile/Scrum
- Product roadmapping
- KPI frameworks