Ga naar inhoud
CODT Technologies logoCODTTECHNOLOGIES
CODT Technologies research

How long real software takes to ship.

Most delivery timelines you can find online are marketing ranges — invented for the click, attached to no real project. This report takes the other route: anonymized benchmarks compiled from CODT Technologies' own delivery records for shipped client products — how many weeks kickoff to v1, what team actually built it, and where the time went.

The short answer
Data pending

One sentence, written from the dataset: the median kickoff-to-v1 timeline across CODT Technologies' shipped products, with the range around it. Published only once the underlying records are compiled and verified.

Median kickoff → v1
Data pending

Median weeks from signed kickoff to a v1 in real users' hands, across the full dataset.

Products in dataset
Data pending

Count of shipped products whose delivery records meet the inclusion rules below.

Period covered
Data pending

First and last kickoff dates in the dataset, stated to the year.

Draft — dataset in preparation

This report publishes its methodology first. Every figure slot on this page is marked "Data pending" until the underlying delivery records are compiled and verified — nothing here is an estimate, and no slot will ever be filled with one. The numbers get published once, correctly, or not at all.

Methodology

Where the numbers will come from.

A benchmark is only worth citing if you can see how it was made. These are the rules this report is built under — fixed before any figure is published, so the data cannot be quietly bent to flatter the result.

  1. Source of record

    Every figure is taken from CODT Technologies' internal delivery artefacts — project plans, milestone sign-offs, release history and billing records — for products that actually shipped. Nothing is reconstructed from memory, collected by survey or borrowed from third-party studies. If a number cannot be traced to a delivery artefact, it does not enter the dataset.

  2. What counts as kickoff

    The clock starts at the signed engagement's first working sprint — not at first contact, not at the proposal. Pre-sales conversations and unpaid scoping are excluded, so the timeline measures delivery, not sales cycles.

  3. What counts as v1

    The clock stops at the first production release in real users' hands — a release the client's business actually operates on. Internal betas, demos and staging milestones do not count. This is a deliberately strict line: it is the one buyers care about.

  4. Anonymization

    Client names are never published in the dataset. Products appear as anonymized labels plus a platform type, dates are stated only to the year, and any figure that would identify a client on its own is published as an aggregate instead. Engagements under NDA are excluded from row-level data entirely.

  5. Inclusion and exclusions

    The dataset covers shipped client products meeting the definitions above. Projects that were paused, cancelled or absorbed into other builds are excluded — and the published report will say how many were excluded and why, because a benchmark that hides its exclusions is an advertisement.

  6. Honest precision

    A single firm's dataset supports medians and ranges — not decimal-point precision or percentile curves. Figures are published at the precision the sample honestly supports, and every aggregate is printed alongside the n and date range it was computed from.

n — products included
Data pending

Final count after the inclusion rules are applied to the delivery records.

Date range
Data pending

Kickoff years covered by the dataset, earliest to latest.

Products excluded
Data pending

Count of excluded projects, with the reasons stated in the published report.

Findings

What the delivery records will show.

Four questions, each answered from the same dataset. Until the records are compiled and verified, every slot below states exactly what it will hold — and holds nothing else.

Kickoff to v1: the overall timeline

The question

How many weeks pass between a signed kickoff and a v1 in real users' hands?

This is the number every founder asks for and almost no agency publishes from real records. The published figures here will be the median, fastest and slowest kickoff-to-v1 timelines across the dataset — under the strict v1 definition above, so a demo never masquerades as a launch.

Median weeks to v1
Data pending

Median kickoff → v1 across all products in the dataset.

Fastest v1
Data pending

Shortest kickoff → v1 in the dataset, in weeks.

Longest v1
Data pending

Longest kickoff → v1 in the dataset, in weeks.

Kickoff → v1, product by product
ProductPlatform typeWeeks to v1Core team size
Data pending

One anonymized row per shipped product: label (e.g. P01), platform type, kickoff → v1 in weeks, and core team size at delivery.

Narrative pending data

The written read of the timeline data: where the median lands, what separated the fastest projects from the slowest, and what that means for scoping a v1.

Timeline by product type

The question

Does a mobile app really ship faster than a SaaS platform or a voice agent?

"How long does an app take" has a different honest answer than "how long does a multi-tenant platform take" — and mixing them is how misleading averages get made. This finding will split the timeline by platform type, with the n for each group printed beside it so a thin group is never mistaken for a trend.

Fastest category (median)
Data pending

Platform type with the shortest median kickoff → v1, and that median in weeks.

Slowest category (median)
Data pending

Platform type with the longest median kickoff → v1, and that median in weeks.

Widest spread
Data pending

Platform type with the widest fastest-to-slowest range in the dataset.

Median timeline by platform type
Platform typeProducts (n)Median weeks to v1Range (weeks)
Data pending

One row per platform type — mobile, SaaS / web platform, AI & voice agents, IoT — with group size, median timeline and range. Groups too small to aggregate honestly will be merged and labeled as such.

Narrative pending data

The written comparison across product types: which categories genuinely take longer, by how much, and where the spread inside a category outweighs the difference between categories.

Team size and composition

The question

How many people does it actually take to ship a v1?

Team size is where delivery myths run wildest — from the solo-genius story to the forty-person programme. The dataset will record the core team that actually built each shipped product, and how that headcount split across engineering, design and product roles.

Median core team
Data pending

Median core team size across all products in the dataset.

Smallest shipping team
Data pending

Smallest core team that took a product in the dataset to v1.

Largest core team
Data pending

Largest core team recorded for a single product in the dataset.

Core-team composition across the dataset
RoleTypical headcountShare of products with the role
Data pending

One row per role — engineering, design, product/delivery, QA — with the typical headcount on a shipped v1 and how many products staffed the role at all.

Narrative pending data

The written read of the staffing data: the typical core team behind a shipped v1, how composition shifted with product type, and what that implies for budgeting a team.

Where the weeks go: phase breakdown

The question

How much of a delivery is discovery, design, build and hardening?

Buyers picture a timeline as one long "build" — then meet the parts nobody advertised: scoping, design, integration hardening, launch. The dataset will break each product's timeline into phases so the published shape of a delivery matches the real one, including the unglamorous stretch between feature-complete and production.

Discovery + design share
Data pending

Median share of the timeline spent in scoping and design, across the dataset.

Build share
Data pending

Median share of the timeline spent in core implementation.

Hardening + launch share
Data pending

Median share of the timeline spent in QA, integration hardening and launch.

Median phase share of the delivery timeline
PhaseMedian share of timelineWhat the phase covers
Data pending

One row per phase — discovery & scoping, design, build, hardening & QA, launch — with its median share of the kickoff → v1 timeline and a one-line definition, summing to the whole.

Narrative pending data

The written read of the phase data: which phase dominates the schedule, which one is most underestimated, and how the shape differs on integration-heavy builds.

Limitations

What this report can and cannot claim.

Every dataset has edges. Naming them is what separates research from marketing — these are the ones this report carries by construction.

  • One studio's data

    This is CODT Technologies' delivery record, not an industry sample. A senior-led, fixed-scope engagement model shapes every number in it — the figures describe how we ship, and generalize to teams that work the same way, not to all software everywhere.

  • Shipped products only

    The dataset measures products that reached v1. Projects that were paused or cancelled are excluded from the timelines — the published report will state how many and why, but survivorship still tilts what a dataset of finishers can say about starting.

  • Definitions move numbers

    Kickoff-to-v1 depends entirely on where those two lines are drawn, and ours are strict. Comparing these figures against other published timelines inherits every mismatch in definition — comparisons are only honest between studies that draw the same lines.

  • A small dataset, deliberately

    A single firm's shipped portfolio supports medians and ranges, not distribution curves. Where a group is too small to aggregate honestly, the report will merge or omit it and say so — thin data presented confidently is the failure mode this page exists to avoid.

  • Long engagements weight the data

    Some products in the record belong to long multi-product platform engagements, where later products ship on infrastructure earlier ones paid for. Where that materially shapes a figure, the report will flag it rather than let compounding advantages read as raw speed.

Citation

How to cite this report.

The report exists to be cited — with attribution and a link. Figures may be revised when the dataset is extended, so citations should name the edition; every revision will be noted on this page.

CODT Technologies — Software Delivery Benchmarks. [data pending: edition]. https://codttech.com/research/software-delivery-benchmarks

Edition and publication date are set when the verified dataset is published.

Figures and tables from this report may be reproduced with attribution to CODT Technologies and a link to this page.

FAQ

Questions about this research

Iets gemist?

Zet het in een briefing. Een senior engineer — geen verkoper — reageert binnen één werkdag.

Zet het in een briefing[email protected]
Where does the data in this report come from?

From CODT Technologies' own internal delivery records for shipped client products — project plans, milestone sign-offs, release history and billing records. No surveys, no third-party studies, no estimates: if a figure cannot be traced to a delivery artefact, it does not enter the dataset.

Why is every figure marked "data pending"?

Because the honest order is methodology first, numbers second. This page fixes the definitions, inclusion rules and anonymization standards before any figure is published, and the data slots stay visibly empty until the underlying records are compiled and verified. Nothing on this page is an estimate — an empty slot is more honest than a plausible one.

How do you define a v1?

The first production release in real users' hands — a release the client's business actually operates on. Internal betas, demos and staging milestones do not count, and the clock starts at the signed engagement's first working sprint, not at first contact. Strict lines keep the benchmark comparable and hard to game.

Are client names published?

No. Products appear as anonymized labels plus a platform type, dates are stated only to the year, and any figure that would identify a client on its own is published as an aggregate instead. Engagements under NDA are excluded from row-level data entirely.

Can I cite or reproduce the findings?

Yes — that is the point of publishing them. Figures and tables may be reproduced with attribution to CODT Technologies and a link to this page. Cite the edition, since figures may be revised when the dataset is extended; every revision will be noted here.

Will the dataset be updated?

Yes, by edition: each published edition states its n and date range, and newly shipped products enter the dataset under the same inclusion rules. Revisions change the edition, never silently overwrite it — a citation to a past edition stays checkable.

Why would an agency publish its own delivery timelines?

Because the alternative is the status quo: invented ranges nobody stands behind. Real records, published with their definitions and limitations, are checkable — and a benchmark we are willing to be held to says more about how we work than any marketing range could. The same records inform how we scope fixed quotes in our discovery sprints.

Keep exploring
App development costSaaS development costHire a dedicated development teamOur portfolio — the shipped productsTrust center
Klaar om te bouwen

Een probleem dat het waard is om
goed op te lossen?

Vertel ons over uw product, uw planning en uw randvoorwaarden. We reageren binnen één werkdag met een eerlijke inschatting van fit, scope en het juiste team ervoor.

Plan een kennismakingGesprek inplannenChat via WhatsAppMail ons direct