K-12 outbound only works when the signal is right. There are roughly 14,000 school districts in the United States. Two hundred of them are actively evaluating any given product category at any given moment. Emailing all 14,000 is spray. Emailing the 200 who are actually in market is signal-based outbound.
The problem is identifying the 200. That is what StarBridge, NationGraph, and Burbio each solve, imperfectly and differently. I spent the last two months evaluating all three for a K-12 EdTech outbound engine. This is the field review.
A district curriculum director receives 200 or more cold vendor emails a day. Their triage is not thoughtful. They tap through the primary inbox looking for internal district messages, then a few vendor names they already know. Everything else gets cleared.
Reply rates on unsegmented K-12 cold outbound are between 0.5 and 1 percent on a good week. Reply rates on signal-matched outbound (where the message references a specific budget line item or board minute the district just published) can hit 4 to 8 percent. That is a five-to-ten-x delta on the same send volume, from the same domain, with the same rep. The only variable is signal.
The signals worth catching in K-12:
None of this data is secret. All of it is public. The problem is that scraping 14,000 district websites, PDFs, and board portals is beyond what a seed-stage GTM team can do by hand. That is the wedge each of these vendors sells into.
StarBridge is the youngest of the three, LLM-native from day one. Its pitch is that it runs a language model across public district documents (board minutes, strategic plans, RFPs) and surfaces intent signals matched to your product category.
The signal quality is genuinely good when it hits. StarBridge caught districts we would have missed entirely with a manual scrape, including one district with a $200K allocated intervention budget buried in a board packet PDF that no other vendor surfaced.
The LLM matching is flexible. You can describe your ICP in natural language ("districts adopting Tier 2 math intervention in grades 3 through 8") and get a filtered feed. Compare to keyword-based scraping, which requires you to know the vocabulary the district uses in advance.
The team is responsive. My conversations with them included specific product feedback that shipped within weeks.
StarBridge is credit-limited. Every district search consumes credits. Priced pilots typically start around $2,000 to $3,000 per month for a modest credit allocation. For a seed team running high query volume, you will burn the monthly budget in the first two weeks and be locked out for the rest of the month, or forced to escalate to a larger plan.
The LLM-per-query design means costs scale with your usage, not with your revenue. That is fine at seed. It becomes a governance problem at Series A when you have multiple SDRs firing queries.
Coverage is uneven. StarBridge is stronger on districts that publish board packets to public portals and weaker on smaller districts running Google Drive or WordPress-based board archives that the crawler cannot reach.
Seed team, small SDR bench, need flexible ICP language, willing to trade cost per query for signal quality. Also pick it if your product category is not one the other two vendors already track natively.
Burbio is the established name, running for years before the current EdTech AI wave. Its model is manual: a curation team physically reviews district calendars, board agendas, and school schedules, then surfaces structured signals.
Data quality is high because it is human-curated. When Burbio says a district approved a $500K SEL curriculum line item, that number came out of a board minutes review by an actual analyst, not an LLM inference. For high-stakes outbound (six-figure ACV pilots) where a wrong number will burn a reputation, Burbio is the least risky source.
Coverage is deep on their fixed set. Burbio tracks around 6,000 districts by hand. If your ICP is within that set, you get well-structured signals with high recall.
Pricing is transparent. Multi-tier subscriptions typically ranging from $1,500 to $5,000 per month depending on volume and access levels. No credit games.
The 6,000-district ceiling is the ceiling. If your ICP includes smaller rural districts, charter networks, or independent schools, you are outside Burbio's tracked set and no amount of subscription upgrade fixes it.
Signal freshness is human-curation-limited. You cannot get a same-week signal on a Monday board minute if the analyst has not reviewed it yet. Latency on new signals is measured in days to weeks, not hours.
ICP customization is limited. You are subscribing to Burbio's schema, not describing your own. The vendor decides what counts as an intervention signal or a grant signal; you receive their categorization.
Enterprise-adjacent motion. High ACV. Cold call needs to reference exact dollar figures without margin for error. ICP fits inside the tracked 6,000-district set. You can wait one week for signal recency.
NationGraph is the newer entrant trying to combine breadth of coverage with LLM-based query flexibility. Its architecture is different from either of the other two: pre-scrape every district's public document surface into a warehouse, then let customers run LLM queries against the warehouse without per-query credit costs.
Coverage is the broadest of the three. NationGraph indexes documents from a much wider district set than Burbio (targeting most or all US districts, not just a curated 6,000).
The pricing model breaks the LLM-credit trap. A typical pilot runs around $750 per month for unlimited queries into the warehouse. If you have five SDRs each running dozens of daily queries, you are not tracking credit burn.
The pre-scraped architecture means signal freshness on tracked document types can be measured in hours after publication, not days.
The product is younger. Onboarding was slower for us than StarBridge (roughly 50 percent complete after two weeks of setup). Data schema is still stabilizing. Some categories of signal are less well-formed than in Burbio.
LLM query quality depends heavily on your prompt. NationGraph gives you a query interface with real flexibility, but you can spend a full afternoon tuning a query to get precision-matched output. Not a plug-and-play tool.
Coverage of smaller districts is technically there but is uneven. The scrape catches districts that publish to standard board portal formats better than districts that do not.
Seed-to-Series-A team scaling SDR count. Wants unlimited query volume without per-query costs. Willing to invest a week in prompt tuning to get a repeatable query surface. ICP includes districts outside the top 6,000.
Signal-based outbound is only half the answer. The other half is what you do with each signal.
The framework we settled on:
The tiering is what lets you use signal-based outbound at scale. If every SDR is calling every district, you burn the SDR bench inside a month. If only the 99-plus districts get the SDR call, you get the same pipeline coverage on a fraction of the effort.
None of the three vendors solve everything. If I were building a K-12 signal tool from zero, the shape would be:
None of the three vendors ship all of that. Most teams end up stitching two together: NationGraph for volume, Burbio for high-ACV verification. Or StarBridge for LLM flexibility and Burbio as a fact-check layer.
The best K-12 outbound teams are not the ones that send the most email. They are the ones that pick the right 200 districts each quarter and route effort proportionally to signal strength.
Signal-based outbound is not about buying a better data vendor. It is about scoring your signals well enough that the SDR bench spends 80 percent of its time on 20 percent of the pipeline.
Pick your vendor. Layer the tier framework on top. Route the bench accordingly. The tools are not the answer. The discipline of tiering is the answer.
← Back to Notes