The customer segment we chose not to serve
One segment we declined a year ago. Why the decision held, what we gave up, and the conditions under which we'll revisit it.
A year ago we made a segmentation call that I haven’t written about in public. We decided not to serve government-prime RFP shops — the firms whose primary revenue is federal primes, eight-figure pursuits, 60-to-90-day cycles with full Shipley color-team discipline. This is the post about why we made that call, whether it held, and when we’ll revisit it.
Opinion piece. First person.
The segment, specifically
A “government-prime RFP shop” in the sense I mean: a firm with a dedicated proposal function of 8–20 people, a capture organization, a pursuit portfolio of federal task orders worth 10M+ each, a production rhythm that targets three to six major submissions per year, and a proposal budget measured in millions. Shipley is their handbook. APMP is their professional society. Their tooling choices are Privia, Virtual Acquisition Office, old Qvidian, custom SharePoint — sometimes all four at once.
This is the segment every incumbent in our category is optimized for. Loopio’s deep-enterprise product, Responsive’s proposalCenter, Upland Qvidian — they serve this segment.
Why we said no
Four reasons, in order of weight:
1. Product surface would bifurcate. A federal-prime product has requirements we don’t: FedRAMP-authorized infrastructure, CJIS compliance paths, air-gapped deployment options, IL4/IL5 hosting for certain pursuits, CAC/PIV authentication, specific document-control workflows. Supporting those requirements would mean building a parallel product. We wanted one product, not two.
2. Sales motion is different. Federal prime shops buy through a 9–18-month procurement cycle that we don’t have the sales infrastructure to run. Champions are VPs of Proposal Operations; approval chains include CIOs and FSOs. Our GTM is self-serve-then-sales-assisted. That motion doesn’t reach this segment.
3. Incumbency advantage is real. This segment has had Qvidian since the 1990s, Privia since the 2000s, Virtual Acquisition Office for federal workflow. Displacement costs are enormous — entire pursuit databases, pre-cleared templates, DoJ memoranda — and the risk of disruption to a billion-dollar revenue stream is too high. A rational buyer in this segment upgrades incrementally, not by replacement. We weren’t going to get the replacement.
4. Our grounded-AI thesis was further from their pain. The federal-prime segment’s top complaints aren’t about AI hallucination — they’re about SharePoint integration, color-team workflow, and B&P budget tracking. Our product leads with citations and retrieval. That story lands harder in segments where generic LLM output is the current failure mode.
What we gave up
Three things, honestly:
A revenue pool. This segment has customers paying six-figure annual contracts routinely. We chose a mid-market segment with four-to-five-figure ACVs. The revenue compounds more slowly.
Category credibility. Analysts grading the proposal-management category rank products partly on their federal presence. Forrester and Gartner both weight this. We’re not going to win a wave-leader position without federal primes, and we accepted that — we’d rather win the category we’re actually building in than compete for a quadrant position in the category we’re not.
Shipley optionality. Several Shipley consultants have asked whether we’d integrate with their methodology more deeply. The answer is we’re compatible with Shipley concepts — color teams, compliance matrices, capture plans — but we’re not going to build Shipley-specific workflow UI. That’s a federal-prime feature set.
Did the decision hold?
A year in, yes. Three signals:
First, the mid-market segment we chose — B2B software firms, regulated-industry SaaS, sub-billion health-tech, financial-services vendors — has the AI-hallucination problem we’re built for. They’re losing bids because of LLM-drafted responses that include fabricated case studies and generic past-performance claims. Our grounded-retrieval approach is pointed directly at that failure mode. The federal-prime segment is drafting by hand and has a different problem.
Second, the product has stayed coherent. One story, one set of metaphors, one pricing structure (in public). I don’t have to explain to a customer why a feature exists — the feature is either for them or it’s evidence that we’ve lost focus. We haven’t lost focus.
Third, the ACVs in our chosen segment are holding and the volume is scaling. Not a financial post and no specifics, but the shape of the curve is right.
What would make me revisit
Three conditions, any two of which would move the decision:
- A federal prime asks us to build a FedRAMP-moderate instance. This would have to be a customer with the contract value to justify the compliance investment.
- Mid-market saturation. If we close the obvious mid-market TAM and need a new segment to continue growing, federal primes are a natural candidate — but we’d serve them through partnership or a dedicated product, not by bolting federal features onto the current one.
- Incumbents ship a grounded-AI story that closes the mid-market gap. If Loopio’s Magic feature actually gets reliably truthful, our differentiator in mid-market shrinks. Federal primes become more interesting as a distinct category to compete in.
I don’t think two of these three move together in the next 18 months. Ask me again at year two.
Why I’m writing this in public
Because I’ve had six prospective customers ask me in the last quarter whether PursuitAgent does federal primes. The honest answer is “no, and here’s why.” Saying no clearly is a feature of how we run the company. Saying yes to every segment is how products lose coherence.