A Dear Colleague Letter With Teeth
The U.S. Department of Education used a Dear Colleague Letter this week to tell states and districts something that should not be controversial and yet reshapes how billions of dollars in edtech spending gets justified. Assistant Secretary Kirsten Baesler wrote that technology in classrooms must be purposeful, instructionally sound, and tied to measurable student outcomes. The letter draws a sharp line between recreational technology, the kind students use for entertainment, and instructional technology bought with the expectation that it moves learning forward. That distinction sounds obvious until you look at how most districts actually evaluate the tools they already own, which is almost entirely through usage dashboards showing logins, minutes on task, and feature adoption.
For the enterprise reader, the interesting part is not the K-12 mechanics. It is the philosophy: adoption is a poor proxy for evidence, and a vendor's growth metrics tell you nothing about whether the product does what it claims. Baesler framed responsible design as the baseline expectation, arguing that buyers need to know whether a tool works and why it works before they renew. That same question sits unanswered in most corporate learning stacks today, where L&D teams report completion rates to the board and call it proof of value. The federal government just put that gap in writing for K-12. Corporate buyers should assume the same scrutiny is coming for them.
From Login Counts to Learning Outcomes
The guidance sets explicit expectations for vendors, not just buyers. Companies selling into schools are now expected to design products with high-quality instruction in mind, provide independent evaluations of effectiveness, and share randomized controlled trial evidence wherever it exists. The letter also asks vendors to minimize unnecessary screen exposure and to share data on how their tools perform in real classroom practice, not just in pilot demos. Districts are encouraged to build these expectations directly into procurement, treating an RCT or a third-party study the way a finance team treats an audited statement: table stakes for a serious conversation, and not a nice extra a good sales deck can substitute for.
This matters because edtech procurement has historically rewarded the wrong signal. A platform with high login counts can still fail to move test scores, and a district under budget pressure often cannot tell the difference between a tool that works and one that is simply well marketed. The guidance gives procurement officers cover to ask harder questions and gives finance teams a federal reference point when a vendor pushes back on evidence requests. Enterprise buyers evaluating any learning platform, whether for a school system or a 10,000-person workforce, should recognize the pattern: usage data becomes a vanity metric the moment it stands alone, without an outcome you can independently verify.
Five States Already Run This Playbook
The letter singles out Arkansas, Indiana, Louisiana, Michigan, and Texas for building performance-based contracting models into their edtech purchasing. In practice, that means renewal and payment terms tied to whatever outcome the tool was bought to improve, reading gains, math proficiency, teacher time saved, rather than to a subscription calendar. These states did not wait for federal guidance to demand this; they built the leverage into their contracts first, and Washington is now using them as proof of concept for everyone else. That sequencing is instructive. The states that moved first got to define what evidence, and how much of it, would count.
For a PE-backed SaaS buyer or a large enterprise L&D function, the lesson translates directly. Whoever writes the outcome clause into the contract controls the leverage in the renewal conversation, and a vendor who resists specifying a measurable outcome up front is telling you something about how confident it is in its own product. Procurement teams that keep contract terms tied to seat counts and feature access, rather than to a defined result, will find themselves negotiating on the vendor's terms indefinitely. The five-state model is a template any buyer can borrow: pick the metric that matters, put it in the contract, and make renewal conditional on it.
Corporate Learning Vendors Should Read This as a Preview
Nothing in the letter applies to corporate LMS or AI-tutoring vendors selling to employers, and no one should expect that to last. The same evidence gap exists in workforce learning platforms that exists in K-12 edtech: heavy investment in completion rates and engagement scores, and comparatively little independent verification that the training changes what employees can actually do on the job. Enterprise buyers who have watched federal guidance migrate from K-12 into higher ed and then into regulated industries should treat this letter as an early signal rather than someone else's problem. The vendors who build evidence generation into their product now will have an answer ready when a customer, or a regulator, finally asks for it.
This is also a competitive opening. A learning platform vendor that can produce a genuine before-and-after skills assessment, rather than a completion certificate, has a defensible answer the next time a CFO asks whether the training budget is producing anything measurable. Buyers evaluating LMS or AI-skills platforms this renewal cycle should ask vendors directly whether they have ever run, or would agree to run, an outcome study on their own product. A vendor that treats the question as reasonable is worth a longer look. A vendor that treats it as an unusual request is telling you how the next difficult conversation will go.
Where the Guidance Falls Short
The letter carries real weight as guidance, but it creates no enforceable national standard, and several education researchers flagged that gap quickly. Digital Promise's Jean-Claude Brizard noted that edtech adoption has too often been measured by access, logins, or frequency of use rather than outcomes, and welcomed the shift, though the guidance still leaves each state to decide how, or whether, to act on it. Districts with strong procurement teams and existing performance-based contracts will move quickly. Districts without that capacity may treat the letter as aspirational language rather than a mandate, a gap that risks widening the distance between well-resourced systems and everyone else.
Enterprise buyers should not mistake the absence of federal enforcement for the absence of pressure. Evidence requirements travel through procurement networks fast once a handful of large, sophisticated buyers start demanding them, because vendors that build the reporting infrastructure for one big customer tend to offer it to the rest. The five states named in the letter will likely become reference customers other states call before their own renewals. The same dynamic will play out in enterprise learning: once a few large employers demand outcome data as a contract condition, the capability becomes a standard line item on every vendor's roadmap.
What to Do Before Your Next Renewal
Start with an inventory question most L&D and IT leaders cannot currently answer: for each learning platform under contract, what independent evidence exists that it changes a measurable outcome, and who verified it. If the answer is a completion rate or a satisfaction survey, that is a usage metric wearing an outcome's clothing. Build a short evidence requirement into the next RFP or renewal negotiation: ask for a third-party study, a controlled comparison, or at minimum a defined before-and-after skills measure tied to the specific capability the tool is meant to build, whether that is coding proficiency, sales onboarding speed, or AI literacy.
Tie a portion of the renewal to that evidence rather than to the subscription date alone, following the five-state model the Department just held up as the standard to beat. Vendors will resist at first, and that resistance is useful information about how much confidence they actually have in their own outcomes. The buyers who set this expectation now, ahead of any mandate, will spend the next few renewal cycles negotiating from data instead of a sales deck. The federal government just told K-12 to stop paying for adoption alone. Enterprise learning budgets should start asking the same question of every vendor on the books.
