Before the Next AI Renewal, Run an Input-Readiness Review

An AI license renewal should open with an input-readiness review: pull per-grade usage data and ask whether your youngest students can physically type fast enough to use the tool you are paying for. Fourth graders averaged about 12 words per minute on the 2012 NAEP computer writing pilot. At that rate, one chatbot exchange eats a class period.

TL;DR: Before renewing an AI tool for the elementary grades, set three numbers side by side: seats used, the grades that used them, and typing fluency in those grades. One analysis of 48 districts found a median 70 percent of student software licenses completely unused. Elementary teachers adopt AI tools at about half the rate of English language arts and science teachers. If usage is thin in grades 3 through 5, keyboarding is the cheaper problem to fix first, and it is measurable inside one semester.

What an input-readiness review is, and what it is not

An input-readiness review is a one-page check, run before a renewal signature, that puts three numbers beside each other: how many licensed seats were used, in which grades, and how fast students in those grades type. It is not a program evaluation. It never asks whether the AI tool is any good. It asks whether the children holding the seats can operate it.

That question now sits in a governance file as well as an instructional one. NASBE counted 34 states with statewide AI guidance by the end of 2025, most of it advisory. Ohio went further and is among the first states to require rather than suggest: every district, community and STEM school must adopt an artificial intelligence policy by July 1, 2026, and districts may adopt the state model as written. So a 2026 renewal happens under a policy that is either already in your board packet or about to be demanded of you.

The review takes an afternoon of an assistant superintendent's time, needs no vendor, and produces a table a trustee can read in the meeting.

A renewal you cannot explain in three numbers is a renewal you cannot defend in open session.

The renewal numbers worth pulling first

Utilization data is the least flattering thing in your contracts folder and the most useful. GovTech reported the Penn Center for Learning Analytics review of 48 districts: a median 70 percent of student software licenses were completely unused, and 97.6 percent were not used to the levels intended. LearnPlatform's 2026 Edtech Top 40 puts the surrounding clutter in numbers: districts maintain access to an average of 3,001 unique digital tools while students and educators each interact with about four.

Adoption is also uneven in a way that matters for elementary renewals specifically. RAND's American Educator Panels reported that in the 2023-24 school year, 25 percent of teachers used AI tools for instructional planning or teaching and nearly 60 percent of principals used them for their own work, with English language arts and science educators roughly twice as likely to adopt as mathematics or elementary teachers.

Number	Source	What it changes about the renewal
Median 70 percent of student licenses completely unused, across 48 districts	Penn Center for Learning Analytics, reported by GovTech	Treat thin usage as the expected outcome, not the surprise
3,001 tools accessible per district; about four used per person	LearnPlatform by Instructure, 2026	The seat you renew competes with 3,000 others for the same 40 minutes
25 percent of teachers and nearly 60 percent of principals used AI tools in 2023-24	RAND American Educator Panels	Adult adoption is not the bottleneck at the top of the building
Elementary and mathematics teachers about half as likely to adopt as ELA and science teachers	RAND American Educator Panels	Your weakest usage sits exactly where typing fluency is lowest

Renew on utilization by grade, and the elementary seats are usually the ones that cannot be justified.

Why input friction reads as disinterest in a usage report

Usage dashboards report behavior, never capability, and from the district office the two look identical. A fourth grader who abandons a prompt after nine words and a fourth grader who was bored produce the same row.

Start with the physical constraint. On the 2012 NAEP computer-based writing pilot, fourth graders averaged about 12 words per minute against about 30 at grade 8, and wrote about 110 words on the computer compared with about 159 on the paper assessment. Wisconsin's Department of Public Instruction publishes the practitioner rule of 5 words per minute times grade level at 80 to 95 percent accuracy. Call that a convention, because that is what it is: no validated national cut score for student typing speed exists, and any vendor who quotes you one is inventing it.

What input friction looks like in the data

• Median session length collapses in grades 3 to 5 while login counts stay respectable, because students open the tool and stall at the text box.
• Student-authored text per session is short in those grades and gets shorter as the week goes on.
• Teachers run the tool from the projector on one adult account instead of assigning student accounts.
• Usage climbs sharply at the first grade band where your students already type, usually grade 6 and up.

What genuine disinterest looks like

• Flat, low usage across every grade band, including grades where typing fluency is not in question.
• Teachers who used the tool in September and stopped by November, in every subject and every building.
• Long sessions with plenty of typed input and no assignment attached to any of them.

The standards are oddly quiet on this. Common Core names keyboarding at grade 3 in W.3.6, "using keyboarding skills", and attaches nothing measurable to it; W.4.6 at grade 4 is the first place it becomes checkable, asking students to type a minimum of one page in a single sitting. The skill gets named in the same year typed testing usually begins, and measured a year after that. Our own grade-by-grade typing benchmarks for US schools exist because the standards decline to give you one.

Before you read low elementary usage as low interest, prove the students could type the prompt.

The review, in five steps

1. Pull seat-level usage by grade for the last two complete quarters, not year to date. Year to date hides the September spike.
2. Ask the vendor for median session length and median student-authored words per session, by grade. If the reporting cannot produce student-authored volume, put that sentence in the renewal memo. It is a fact about the product.
3. Take a typing sample: three minutes of grade-level text, 25 students per grade band, words per minute and accuracy recorded by the teacher who administers it.
4. Compare the sample against 5 words per minute times grade level. Flag any band running more than 5 WPM under.
5. Decide by rule, using the table below, and write the rule into the memo before anyone sees the result.

Step five is the step districts skip, and it is the one that makes the review defensible. A decision rule written after the data arrives is an opinion with a table attached.

Sequencing: renew, trim, or teach typing first

Usage in grades 3 to 5	Typing sample against the convention	Decision
Healthy	At or above 5 WPM per grade	Renew as written; the input question is closed
Thin	More than 5 WPM below	Renew fewer elementary seats, move the difference into keyboarding minutes, retest in one semester
Thin	At or above	Input is not your problem; look at teacher training and the AI policy your staff is waiting on
Healthy in grades 6 to 12, thin below	Below only in grades 3 to 5	Shift seats up a grade band and buy keyboarding instruction down a band

Each row is a motion somebody can move at the next meeting. Two of them cost less than the renewal you were about to sign.

The objection I take seriously

Here is the strongest case against everything above. Speech-to-text has become good enough that a nine-year-old can hold a conversation with an AI tutor without touching a keyboard, so keyboarding is a shrinking constraint and the seats are the better buy. That is half right, and the half it gets right matters: if conversational tutoring is your only use case, you can defer keyboarding and lose little this year.

It breaks on the graded artifact. Texas requires an extended constructed response essay in every grade 3 through 8 reading language arts assessment, administered online. New York phased computer-based testing in until all of grades 3 through 8 were required from spring 2026. Voice input does not write the essay inside the state's test delivery platform, and the student who dictated all year meets the keyboard for the first time in April.

One more concession, because a superintendent will ask for it: I cannot point to a study showing that keyboarding instruction raises AI tool utilization. That link is reasoning from a physical constraint, not evidence. What is evidenced is the constraint and the assessment requirement.

Defer keyboarding if you only care about the conversation; do not defer it if your students will be graded on typed writing.

What the review costs, and what school access costs

The review needs no purchase. School software does, and the honest version is short. Meta Typing Club is free for personal use, meaning an individual learner, a family, or a parent working with their own child. School, district and organizational access is commercial and priced on request. An independent teacher evaluating on their own account gets one class of up to five students at no charge, which is enough to form an opinion and far too small to be a pilot. The moment a school's students are on it, that is school use, and it goes through your data privacy agreement first, paid or not.

What the product does today, so you can price the work honestly: more than 2,400 structured lessons; a teacher or parent creates each student account directly, with no email required for the child and no age or date of birth ever collected; homework assignments with optional due dates return per-student completion, accuracy, words per minute and stars. The dashboard reports words per minute only. There is no Clever, ClassLink or OneRoster integration, which means no roster import and an adult creating 25 accounts by hand.

On data, the position is uncomfortable and I would rather you hear it from us. Our storage region and sub-processors are not documented, so we make no compliance claim, and you should not accept one from any vendor without paperwork. Ask us, and every bidder, for the four artifacts a district should demand from an edtech vendor: a signed data privacy agreement, a named sub-processor list, a written retention and deletion period, and a plain statement of what a child account stores. Across Meta Typing Club's 10,000 plus learners, daily practice tends to add about 10 words per minute a month; that figure covers the whole user base rather than an elementary cohort, so treat it as a planning estimate and measure your own grades 3 to 5.

Ask for the per-student quote in writing before the vote, and ask for the four artifacts in the same email.

Key takeaways for the renewal memo

• A median 70 percent of student software licenses across 48 districts went completely unused, so thin usage is the expected result rather than an anomaly.
• Districts hold access to about 3,001 digital tools while each student and teacher uses roughly four, which is the competition your renewed seat faces.
• 25 percent of teachers and nearly 60 percent of principals used AI tools in 2023-24, and elementary teachers adopted at about half the rate of ELA and science teachers.
• Fourth graders averaged about 12 words per minute on the 2012 NAEP computer writing pilot, against about 30 words per minute at grade 8.
• The 5 words per minute times grade level rule at 80 to 95 percent accuracy is the only widely published yardstick, and it is a convention rather than a validated cut score.
• Common Core names keyboarding at grade 3 and first measures it at grade 4, at one typed page in a single sitting.
• 34 states had statewide AI guidance by the end of 2025, and Ohio requires a district AI policy by July 1, 2026.
• The review needs one afternoon and no vendor; school access to typing software is commercial and quoted per student on request.

Questions superintendents ask about this

What does Meta Typing Club cost for a district our size?

School, district and organizational access is commercial and quoted on request, with nothing published and no self-serve purchase. Ask for a written per-student figure before your finance committee meets. Personal use stays free for individual learners, families, and a parent working with their own child.

Can one teacher try it before we spend anything?

An independent teacher can evaluate on their own account with one class of up to five students at no charge. That is an evaluation, not a pilot. Put a school's actual students on it and it becomes school use, which goes through your data privacy agreement first, paid or not.

Do I have to cancel the AI license to act on this review?

Rarely. Most reviews change the seat count and the grade band rather than the vendor. Renewing 200 seats in grades 6 through 12 instead of 500 across grades 3 through 12 frees money for keyboarding minutes, and you keep the tool your board asked you to adopt.

How long before typing speed moves enough to matter?

Plan on a semester. Across Meta Typing Club's 10,000 plus learners, daily practice tends to add about 10 words per minute a month, measured across the whole user base rather than one grade. A grade 4 cohort starting near 12 WPM and reaching the 20 WPM convention within a semester is a defensible target.

What would we have to sign, and will you sign our DPA?

Bring us your template and we will answer it plainly rather than in marketing language. We do not publish a storage region or a sub-processor list today, and we will not claim FERPA or COPPA compliance we cannot evidence. That answer should shape your decision, including a decision against us.

How much teacher time does this actually take?

Two costs are real. Practice runs in 10 to 15 minute blocks that a teacher schedules, and account creation is manual because there is no rostering integration, so a class of 25 means 25 accounts created by an adult. Homework assignment and per-student results are automatic after that.

Is this my decision or the board's?

The board votes budget and policy, your staff selects, and purchasing runs the process. The one-page instructional audit needs no vote and no vendor, which is why it comes first. Bring the board a written quote and a decision rule rather than a recommendation with a blank fiscal note.

Where to start before the renewal date

An input-readiness review answers one question in three numbers: seats used, the grades that used them, and typing fluency in those grades. Run it in the two weeks before your renewal date, write the decision rule before you look at the result, and take the table to the board. If grades 3 through 5 come back below the 5 words per minute per grade convention, request a per-student quote for structured keyboarding lessons by grade band and ask for the four data artifacts in the same message. The audit is the part you can start on Monday.