What the APA is asking districts to measure
The American Psychological Association's recommendations reject the metric most districts have defaulted to for a decade: minutes on screen. Instead, the guidance asks districts to evaluate what students are actually doing during that time, what activity the technology is replacing, and whether the tool measurably helps or hinders learning outcomes rather than simply occupying attention for a measurable block of the school day. That is a materially harder evaluation to run than counting minutes, and it is exactly the evaluation most procurement processes have skipped.
The recommendations extend past the classroom too, covering the content students access, background screen exposure they are not actively engaging with, and adult device use during shared family routines that shape how children relate to screens generally. For a body that has spent years fielding parent anxiety framed almost entirely around screen time totals, this is a deliberate pivot toward a harder, more specific question that vendors have mostly been allowed to avoid answering directly: does the tool actually work, and can that be shown.
Why screen time became the wrong proxy
Screen time is easy to measure and easy to regulate, which is exactly why it became the default metric even though it tells a district almost nothing about learning. A student watching a lecture video, chatting with an AI tutor, and passively scrolling an unrelated app all register identically on a screen time dashboard, despite producing wildly different educational value, and district technology committees have spent years debating limits built on that flawed equivalence rather than on what the screen time actually produced.
APA chief executive Arthur Evans Jr. put the stakes in plain terms: "Educational technology has the potential to profoundly advance students' learning and abilities, but parents, teachers, and school districts need tools that are safe, supported by evidence, and can genuinely help students learn." That framing shifts the burden onto vendors to prove value, not onto parents to police usage minutes, and it hands districts language they can put directly into board policy and procurement criteria instead of the vague safety assurances most vendor pitches currently rely on.
The one week retention bar
The most concrete piece of the guidance is also the most demanding for vendors: the APA recommends products demonstrate knowledge retention or skill transfer that lasts at least a week after use, rather than being assessed only at the moment a lesson or session ends. That is a meaningfully higher bar than the engagement metrics, session length, and completion rate figures most edtech sales decks lead with today, and it cannot be satisfied with the same analytics dashboards vendors already ship as standard reporting.
A week-long retention requirement forces vendors to run actual learning science studies rather than product analytics pulled from their own usage logs, and it exposes the gap between tools built around engagement loops designed to maximize time on task and tools built around durable learning that survives once the session is over. Districts that adopt this standard in procurement language will filter out a meaningful share of the current AI tutoring and adaptive learning market almost immediately, simply because most vendors have never run the study this bar requires.
How this lands next to the moratoriums
The timing is not incidental. New York City and Los Angeles Unified, the country's largest and second largest districts, have both imposed yearlong moratoriums on student facing AI tools for 2026-27 while they build out governance rather than continue deploying without it. The APA guidance gives those districts, and the smaller ones watching how they resolve this, a substantive framework to apply once the pause ends rather than defaulting back to the same screen time limits that got districts into this evidence gap in the first place.
Districts that restrict now and districts that deploy now are converging on the same underlying demand, whether they say it explicitly or not: proof, not promises. A moratorium buys time to build that proof requirement into the next procurement cycle before any new contract gets signed. Deployment without it, as several smaller districts already moving quickly are learning the hard way, produces exactly the evidence gap the APA is now asking every buyer in this market to close before their own next renewal cycle arrives.
What it means for vendor evidence packages
Any edtech vendor selling into K-12 or higher education should assume RFPs will start requesting retention and transfer data within a school year or two, not five years out once the standard has had time to become optional in practice. Vendors that already run independent efficacy studies have a real, durable advantage over competitors who do not; vendors relying on engagement dashboards and testimonial case studies will need new evidence infrastructure fast, and building it from scratch is a research undertaking, not a marketing exercise that can be assembled before the next sales cycle.
This also changes what counts as a credible pilot going forward. A six week engagement pilot that never checks whether students retained anything after the tool stopped being used will not satisfy this standard, no matter how strong the usage numbers look in the final report. Enterprise buyers evaluating any AI-in-learning vendor, corporate L&D included, should ask right now whether a follow-up assessment exists at all in the vendor's proposed pilot design, because most current pilots on the market are simply not built to answer that question.
The procurement shift enterprise leaders should plan for
Treat this as a preview of what corporate learning and development procurement will demand too, on roughly the same timeline school districts are now facing. The same weak proxy, completion rate instead of retained skill, dominates enterprise training evaluation today, and the same fix applies directly: require vendors to show that a skill still transfers a week or a month after training, not just that employees clicked through and finished a module before the deadline their manager set for them.
Build the retention question into your next AI-in-learning vendor evaluation regardless of whether you are buying for a school district or a Fortune 500 training organization with a very different budget and buyer sophistication. The APA just gave every buyer in this category a credible, professionally endorsed reason to ask it out loud in a vendor meeting, and the vendors who cannot answer convincingly will be the first line items cut the next time training budgets tighten.
