
What Meta Connect 2026 Actually Changes for Teams Building AI ProductsRead More

A $150,000 EdTech quote and a $400,000 quote can describe what appears to be the same feature list. The difference often lies in what the price actually covers, from a demo-ready build to a production platform prepared for accessibility reviews, institutional integrations, and peak student traffic.
Cost moves further once the product has to support WCAG 2.1 AA conformance, a VPAT, LTI 1.3, SIS rostering, multi-tenancy, or high-stakes assessment workflows. Team location and delivery model matter too.
A realistic EdTech software development cost estimate starts with understanding what sits behind the number. This guide moves from cost ranges and estimation arithmetic into EdTech-specific cost drivers, build routes, post-launch run rates, engagement models, and a seven-question method for comparing vendor quotes. It explains what determines EdTech platform cost and how to budget for compliance, integrations, infrastructure, and scope risk. It is written for founders, product and engineering leaders, and institutional EdTech teams evaluating software development partners.
A focused EdTech MVP costs roughly $92,000 to $281,000 to build with an offshore team in 2026, a production platform for institutional buyers $274,000 to $896,000, and a high-stakes assessment system $548,000 to $1.68 million. Those figures come from a formula you can run yourself.
Every engineering quote reduces to the same formula: people, multiplied by months, multiplied by an hourly rate. Work it in that order and you can rebuild almost any number a vendor hands you.
A full-time engineer bills about 173 hours a month, which is 2,080 hours a year divided by twelve. Rates depend on where the team sits. A 2026 rate survey, drawn from 60 software development partners worldwide, puts junior developers in Asia at $24 to $31 an hour and seniors at $31 to $41. (Source: Price Pressure and Performance - Accelerance) In Central and Eastern Europe the same roles run $31 to $39 and $64 to $76. In Latin America, $33 to $45 and $60 to $75. Average rates fell year over year in all three regions: about 8% in Asia, 7.1% in Latin America, 4.4% in Europe.
Hiring the same people in the United States changes the arithmetic. The median annual wage for software developers was $135,980 in May 2025, according to the Bureau of Labor Statistics. Benefits account for 30.0% of total employer compensation costs in private industry, so the fully loaded cost of that median developer is about $194,000 a year, or roughly $16,200 a month, before recruiting, equipment, management time, or a seat sitting empty for three months.
A team is not all seniors, so use a blended rate: the midpoint across the junior and senior bands for your region. That gives about $33 an hour for Asia and about $54 for Central and Eastern Europe or Latin America. Every figure below is that rate multiplied by the hours in the same row, so you can check the arithmetic.
Product shape | Team and duration | Engineering hours | Asia (~$33/hr) | CEE or LATAM (~$54/hr) | In-house, US |
|---|---|---|---|---|---|
Focused MVP, one audience, no institutional integrations | 4 to 5 people, 4 to 6 months | 2,800 to 5,200 | $92k to $172k | $151k to $281k | $259k to $486k |
Production platform: multi-tenant, SSO, LTI, reporting, accessibility conformance | 6 to 8 people, 8 to 12 months | 8,300 to 16,600 | $274k to $548k | $448k to $896k | $778k to $1.56m |
High-stakes assessment: proctoring, item banking, audit trails | 8 to 10 people, 12 to 18 months | 16,600 to 31,100 | $548k to $1.03m | $896k to $1.68m | $1.56m to $2.92m |
The rates in that table are survey data. The team sizes and durations are a planning model, which makes them the half you should argue with, and also the half any vendor's estimate is made of. Change them to match your product and rerun the multiplication.
The table leaves out design research, content production, third-party licenses, cloud infrastructure, app store and payment fees, legal review, and everything after launch.
Three quotes differ for three reasons, and the hourly rate is rarely the main one.
The first is the definition of done. One vendor is pricing a working demo. Another is pricing a system that survives a security review, an accessibility audit, and a Monday morning in September when 40,000 students sign in at once.
The second is how much compliance sits inside the number. Accessibility conformance, student data handling, and interoperability certification get quoted as "as required", or left out silently, then billed later as change requests.
The third is who covers the gap when scope moves. That is the engagement model, and it is the second half of this article.
A fourth possibility is worth naming: the estimate may simply be wrong. McKinsey and the BT Centre for Major Programme Management at the University of Oxford studied more than 5,400 IT projects and found that large ones ran 45% over budget and 7% over time while delivering 56% less value than predicted. That work was published in 2012 and covered projects with budgets above $15 million, so it does not transfer cleanly to a $200,000 build. It supports a plainer point. A single number with no range attached is a guess wearing a suit.
The EdTech surcharge is the set of line items that separate a $150,000 quote from a $400,000 one for the same feature list. Ordinary business software carries some of them. Education software carries all of them.
Education software gets bought by institutions, and institutions are regulated. The ADA Title II web rule names WCAG 2.1 Level AA as the technical standard for state and local government entities, which covers public schools, community colleges and state universities. The Department of Justice published that rule in April 2024, then on April 20, 2026 issued an interim final rule moving compliance to April 26, 2027 for entities serving 50,000 people or more, and April 26, 2028 for smaller entities and special districts. The obligation sits with the institution. It reaches you through procurement, usually as a request for a VPAT, the Voluntary Product Accessibility Template a buyer reads before signing.
In Europe the deadline has already passed. The European Accessibility Act came into force on June 28, 2025 across all 27 member states and applies to non-EU companies selling into the EU. Anything placed on the market after that date has to comply immediately. Services already on the market have until June 28, 2030.
Retrofitting is the expensive route. WebAIM's February 2026 scan of the top one million home pages detected WCAG 2 failures on 95.9% of them, up from 94.8% a year earlier, averaging 56.1 errors per page. Deque, which sells accessibility tooling, puts the cost of fixing a defect after release at 60 to 100 times the cost of fixing it during design, drawing on older IBM defect research rather than accessibility-specific data. Read the multiplier as an order of magnitude. The direction of it is not in dispute.
The FTC's amended COPPA Rule took effect on June 23, 2025, with full compliance required by April 22, 2026. It adds obligations that turn into engineering tickets: a written children's personal information security program, data retention limited to as long as is reasonably necessary under a written policy you disclose, and separate verifiable parental consent before you share a child's personal information with a third party. Biometric identifiers, including voiceprints and iris patterns, now count as personal information, which matters if your product does speech scoring or remote proctoring.
One point here belongs to your lawyer rather than to me. In these amendments the FTC declined to adopt a school authorization exception and declined ed-tech-specific provisions, citing expected updates to FERPA, the federal law covering student education records. Older FTC guidance about schools consenting on behalf of parents still circulates widely. Get your consent model confirmed by counsel before anyone builds it.
A learning tool that cannot be launched from the platform a customer already runs does not get bought. LTI 1.3 and LTI Advantage, maintained by 1EdTech, handle that launch along with roster provisioning, deep linking and grade passback.Certification requires paid 1EdTech membership, priced by organization size and revenue and not published publicly, so ask them directly. SCORM and xAPI cover content portability, QTI covers assessment items, OneRoster covers class rosters. Each is a working implementation plus a certification cycle.
Connecting to a school's student information system is the integration buyers underestimate most. Building it yourself means connection adapters, sync scheduling, data validation, referential integrity checks and file transfer infrastructure, repeated per system. The industry newsletter EdTech Insiders describes the two main platform routes as economically different for vendors: Clever is free for districts and charges EdTech vendors for integration access, while ClassLink charges districts and gives vendors access through the open OneRoster standard. Confirm current terms with each provider, because this choice creates a recurring cost you do not control.
Education traffic is seasonal in a way most consumer products are not. Term start and exam windows compress demand into a few days. The architecture is sized for the peak and paid for across the whole year, which moves autoscaling and caching out of the backlog and into scoped work.
Somebody has to build the tools your instructional designers will use, and somebody has to move the existing courses in. Migration is slow work, and almost never in the headline quote. Ask where it sits.
Serving many institutions from one system means per-tenant branding, roles, data isolation and reporting. Decided at the start, it shapes the data model. Decided in year two, it is a rewrite.
The build route moves the number more than any single feature does.
| Route | Upfront cost | Time to first users | Control | Run rate |
|---|---|---|---|---|
| Build from scratch | Highest | Slowest | Total | Yours entirely |
| Build on Open edX or Moodle | Middle | Faster | High, within the platform's model | Hosting plus upgrade work |
| Configure commercial SaaS | Lowest | Fastest | Limited to what the vendor exposes | Per-seat license that grows with you |
The middle route gets misread most often. Starting from an existing platform removes the foundations, not the engineering. Arbisoft has worked on Open edX with edX since 2013, with over 150 people contributing to development, customization, course authoring, testing and support for a platform now used by more than 20 million learners and 140 partners. That is the shape of platform work at scale. The starting point is further along, and the work does not stop.
Pick the route by what you are competing on. If your differentiator is pedagogy or content, a platform base is usually the cheaper path to the same outcome. If your differentiator is the learning mechanic itself, the platform's data model will eventually fight you.
Launch converts a project cost into a run rate, and the run rate is what your board will ask about in year two.
Hosting and video delivery scale with usage rather than with headcount, so a product that succeeds costs more to run. Support and content operations need real people. Standards and regulations move, so conformance is a recurring exercise rather than a one-time certificate; the compliance dates in this article have already shifted once. AI features add a per-use inference cost that grows with engagement, which changes the unit economics of a product priced per seat.
A common planning convention puts annual maintenance at 15% to 25% of the original build cost. That figure is repeated widely and I could not trace it to an original study, so treat it as a budgeting habit rather than evidence. The honest version is to cost the run rate from its parts: infrastructure, a named support rotation, content operations, and a standing allocation for compliance work.
A fixed price does not remove risk. It transfers risk to the vendor, and the premium for that transfer is the contingency sitting inside the number you were quoted.
Fixed price works when the scope is written down and stable, and you genuinely will not change it. The vendor carries the overrun. In exchange, every clarification becomes a change request, and the incentive on both sides is to argue about the specification rather than improve the product. It suits a defined piece of work such as an integration, a migration, or an accessibility remediation.
Time and materials means you pay for hours worked. Budget risk sits with you, along with full visibility to stop, cut, or redirect. It fits work where the problem is understood and the solution is not, which describes most discovery and most first versions. Cap it by phase rather than leaving it open.
A dedicated team is a monthly retainer for named people who work only on your product while you own the roadmap. Unit cost is usually the lowest of the three because there is no scope contingency priced in, and continuity compounds: a team that has already been through one exam season knows what breaks in the next one. It needs product leadership on your side to be worth it.
Outcome-linked or co-investment deals, where part of the fee depends on a delivery or commercial milestone, exist but are rare and slow to negotiate. They need a metric both parties trust and a relationship with history behind it.
| What you are certain about | Model that fits | Who carries the overrun |
|---|---|---|
| Scope is written, stable, and will not change | Fixed price | The vendor, and you paid a premium for that |
| The problem is clear, the solution is not | Time and materials, capped by phase | You, with visibility to stop |
| Long-running product, you own the roadmap | Dedicated team | You, at the lowest unit cost |
| You need evidence before committing runway | Fixed-price discovery, then a dedicated team | Split by stage |
Most EdTech products that reach institutional customers end up on the last row. A short fixed-price discovery produces a specification worth pricing, and the build that follows runs on a team that stays.
Differently shaped quotes become comparable once you force the same answers out of every vendor. Seven questions do most of the work, and how a vendor answers tells you as much as what they answer.
| Ask every vendor | What a vague answer is telling you |
|---|---|
| What is your written definition of done for a completed feature? | Demo-quality work is being priced as production work |
| Is WCAG 2.1 AA conformance in this price, and will you produce a VPAT? | Accessibility will return later as a change request |
| Which integrations are named in scope, and which say "as required"? | The rostering and student information system work has not been estimated |
| What counts as a change request, who decides, and what does one cost? | You will be negotiating every clarification |
| What is the year-two run rate for infrastructure, support and compliance? | Nobody has modeled the product after launch |
| Who owns the code, the pipelines and the cloud accounts? | Leaving this vendor will be expensive |
| What happens to the team and the price if we pause for two months? | A pause will cost you the people who know your product |
Two quotes that answer these differently are not two prices for the same thing. They are two different products.
A learning management system with courses, enrollment, assessment, reporting and institutional single sign-on generally falls in the production platform band: roughly $274,000 to $896,000 depending on your team's location, using the rates and planning assumptions above. Building on Open edX or Moodle lowers the starting cost. Multi-tenancy, LTI certification and accessibility conformance raise it.
Education software carries obligations that most business software does not: accessibility conformance to a named standard with legal deadlines, children's privacy rules under COPPA and student record rules under FERPA, interoperability certification through 1EdTech, rostering integration with school systems, and seasonal traffic peaks that set the architecture. Each one is scoped engineering work.
Yes, and it is a common sequence. A fixed-price discovery or MVP produces a specification and a working relationship, then a dedicated team takes over the ongoing build. Agree the transition terms at the start, including what happens to the code, the documentation and the people if you decide not to continue.
Under the amended Rule, with full compliance required by April 22, 2026: a written children's personal information security program, data retention limits with a published policy, separate verifiable parental consent before disclosing a child's data to third parties, and treatment of biometric identifiers as personal information. Have counsel review your consent model before building it.
Enough that a change in scope is a decision rather than a crisis. Base it on how much of your scope is genuinely fixed, and remember that the contingency in a fixed-price quote is already there, priced by the vendor and invisible to you.
Price the compliance surface before the feature list. Accessibility conformance, student data handling and interoperability certification are the items that move the number most and that vendors quote most differently, so pinning them down first turns three incomparable quotes into three comparable ones. Send the seven questions above to every vendor on your shortlist, and read the answers before you read the totals.
Trusted by top platforms for our transformative solutions and exceptional results:






