Is Your Canva Resume ATS-Friendly? We Ran It Through a Real Parser

Canva's resume templates look great. That part isn't in dispute: the designs are clean and modern, with typography that's genuinely pleasant to look at. The problem is that the first "reader" of your resume is almost never a person. It's a program that starts by extracting the file's text, in the order the file hands it over. And some of those templates, so good to the eye, fall apart at exactly that layer.

Since nobody had published the actual test, we ran it. We took 4 real Canva templates (from Canva's Spanish-market catalog; the findings are about layout, not language), two regular ones and two from the set Canva itself labels as ATS, exported each with the template's own placeholder content, and ran every PDF through text-extraction software: the same literal reading any ATS starts with. Nobody can claim we crafted the content to fail (it's the text Canva ships in the template). And the problem we found isn't missing words: it's that the text comes out in a different order.

The results: 4 Canva templates, read the way the machine reads them

The question this table answers isn't "is the resume good?". It's more basic and more decisive: when a program extracts the text from the PDF, does it come out in the order you see it?

The headline finding: Canva's ATS label doesn't guarantee the order

Canva already sells the idea: it has entire category pages of "ATS resume templates". That's why we included two templates from that set as the control, and the result became this test's headline.

Of the two labeled templates, one comes out nearly clean (Beige Negro Simple, and we name it below) and the other hands over other people's references before your own contact data and detaches the experience section from its heading. Two templates aren't the whole catalog, so the fair conclusion is not that the label always misleads. The conclusion the data does support is this: the label is no guarantee. You have to check the specific template you're going to use, because the label won't do it for you.

How we tested it: a text extractor, and the order things come out in

We ran each PDF through text-extraction software, the same kind of reading every ATS starts with: walk the file and pull out the text in the order the file delivers it, which doesn't have to be the order you see on screen. That's the central finding of this test: the problem is almost never missing text, it's text out of order.

You can approximate the same reading without installing anything: open your resume PDF, select all the text (Ctrl+A or Cmd+A), copy it, and paste it into a plain text editor like Notepad. What you paste is the file's text layer; if it comes out scrambled there, it comes out scrambled from the extractor too.

The blocks below are that extraction in full, line by line and untouched. They come collapsed for space; each block's button expands the text down to the last line, because if we're going to criticize, the whole proof should be on the table.

Template by template: what you see vs. what gets read

Minimalista Blanco y Negro: the name arrives 8th

On the left, the resume as you see it: the name huge at the top, everything in its place. On the right, the extracted text, which begins like this, verbatim:

Soy ingeniera de software y cuento con 5 años de experiencia en diseño, planeación y desarrollo completo de sistemas complejos.
Competente en tecnologías del lado del servidor, incluyendo el trabajo con varias plataformas y diversas herramientas y leguajes
de desarrollo para su realización.
ACERCA DE MÍ
EXPERIENCIA PROFESIONAL
EDUCACIÓN
PROGRAMADORA
MARCELA ANDRADE
hola@sitioincreible.com
(55) 1234-5678
Ciudad Increible
Trabajé como desarrolladora software, llevando a cabo varios proyectos desde el
concepto inicial hasta su realización concreta. A la vez que lideraba a mi equipo de
trabajo. Realicé tareas como la creación de soluciones de extremo a extremo
utilizando tecnologías de vanguardia para la web y asesoramiento a los miembros del
equipo, incluyendo revisiones diarias de código.
Tarea o logro a destacar en este rol
Tarea o logro a destacar en este rol
Tarea o logro a destacar en este rol
Tarea o logro a destacar en este rol
Tarea o logro a destacar en este rol
Logro o aprendizaje a destacar en este rol
Logro o aprendizaje a destacar en este rol
Logro o aprendizaje a destacar en este rol
Creadora de contenido
Larana, Inc. - 2020 a 2024
Como desarrolladora junior, me dediqué a la búsqueda de dar solución a fallos de
código en períodos limitados de tiempo. Aquí realicé tareas como la integración de
las herramientas y refactorización de código heredado, dando ayuda al equipo a
asegurar la corrección de errores en tiempo mínimo.
Desarrolladora de software Jr
Vandiere Inc. - 2019 a 2020
Formé parte de un equipo donde me dediqué a aprender y mejorar mis habilidades
cumpliendo con tareas simples. Aquí profundicé mi aprendizaje de lenguajes de
programación y aspectos de seguridad bajo la tutela de superiores.
Programadora Jr
Sociedad Martínez - 2017 a 2019
Universidad 1
Curso especializado 2
Grado en universidad 1
2016 - 2020
2021-2022
Grado en curso especializado
HABILIDADES
Lenguajes de programación.
Base de datos y caché.
Liderazgo.
Arquitectura de apps.
UX/UI.
Trabajo en equipo.
IDIOMAS
REFERENCIAS
Español ( NATIVO )
Inglés ( EXCELENTE )
Ricardo Soto / (55) 1234-5678
Ima Luque / (55) 1234-5678
PORTAFOLIOS
En los siguientes links se pueden ver mis portafolios y algunos de mis trabajos
Link 1
Link 2
Link 3
Link 4

Three observable facts: the profile paragraph arrives first, before any heading, as orphan content. The three section headings arrive fused in a row, detached from their content. And the candidate's name is the 8th element of the stream, behind the body text and the headings. Expand the block and it stays that way: each job's description arrives separated from its title (the first description starts at line 12; the first job title comes much later). A parser that expects name and contact details at the top of the document ends up misfiling them, and one that assigns each piece of text to the nearest heading has no way to tell which paragraph belongs to which section.

The two-column marketing resume: six headings fused, no content

The classic sidebar case. The extracted text opens like this:

Ingles
Alemán
Portugues
WEB DESIGNER & MARKETING
Me recibí de Diseñador Web en el año 2020, y estoy en busca de un nuevo trabajo.
Actualmente estoy cursando en la carrera Marketing, especializadome en redes
sociales. Me considero una persona responsable y ordenada. Buscando mi nueva
experiencia laboral.
CONTACTO
IDIOMAS
REFERENCIAS
EDUCACIÓN
EXPERIENCIA
CURSOS Y MASTERCLASS
• Graduado con honores académicos
• Jefa en Investigación del "Mercado del Diseño", 2020
• Graduado con honores académicos
• Vicepresidente del Club de Arte, 2015
Este curso es especializado en la administración de
oficinas, pero además en la administración y
liderazgo de equipos.
Este curso esta orientado para impulsar las redes sociales
de los emprendedores, para conocer el mercado y
expandir mis conocimientos.
Teléfono:
(55) 1234-5678
Mail:
Dirección:
hola@sitioincreible.com
Calle Cualquiera 123, Cualquier lugar.
40%
50%
74%
SANDRA HARO
Encargada de Personal
Mail:
Telefono:
hola@sitioincreible.com
(55) 1234-5678
FELIPE SAENZ
Universidad Borcelle
Escuela Secundaria Fauget
Colegio de Marketing
• Actualmente estudiando
• Vicepresidente del Club de Curso, 2018
Asistente de Gerencia
Pasante Administrativo
Atención al Público
 Seguimiento de Agenda. Revisión de Idoneidad
de documentos y control de archivo. Creación
de Presentaciones mensuales.
Recepción de Clientes. Manejo de Conmutador.
Organización de libro de entrada y salida.
Mantenimiento de archivo.
Atención a Clientes personalmente
 Mantenimiento y seguimiento de Stock y logística
Curso de Administración
Curso de Manejo de Redes
RICARDO SALINAS
Encargado de Personal
Mail:
Telefono:
hola@sitioincreible.com
(55) 1234-5678

The document starts with the sidebar's language list, before the job title and before the name. Then six section headings arrive in a row, all fused together and all without their content; the bullets that follow can't be attributed to any of them. The contact labels come out separated from their values ("Mail:" and "Dirección:" back to back, with the email after both): a field extractor that expects label-value pairs matches them wrong or not at all. And the skill-level bars arrive as loose numbers (40%, 50%, 74%) with no skill next to them.

Beige Negro Simple, ATS-labeled: the near-pass

Let's say this as clearly as the rest: this template does pretty well. The main column reads in the correct order: role, company, dates, achievements, job after job. If you want to use Canva, this is the kind of template to pick, and we'll recommend it by name: Beige Negro Simple CV de Profesional en Marketing Digital.

With two caveats. First: the extracted text opens like this:

Hacé clic acá para ver mi
portfolio
Experiencia laboral
Especialista en marketing digital
Gran Arca Publicitaria, una agencia publicitaria
global que se especializa en llegarle al público en
línea
Agosto de 2022 a mayo de 2025 (2 años, 9 meses)
Logré un aumento del 30 % en seguidores en
redes sociales mediante el desarrollo de una
fuerte estrategia de contenido que condujo a un
crecimiento del 5 % en las ventas en seis meses
Aumenté las asociaciones B2B y B2C en un 5 % al
supervisar las campañas de marketing desde la
planificación hasta la optimización
Impulsé el reconocimiento de la marca de los
clientes en un 10 % mediante el desarrollo de
nuevas ideas y estrategias de marketing de
contenidos
Especialista en ventas y marketing
Grupo de Medios Digitales Schuester
Noviembre de 2020 a enero de 2022 (1 año, 2 meses)
Aumentó el tráfico orgánico del sitio en un 79 %
al desarrollar y ejecutar estrategias de contenido
de marketing digital
Aumentó nuestra base de clientes en un 5 %
mediante el establecimiento de asociaciones con
anunciantes
Trabajé en propuestas publicitarias y campañas
de marketing en línea
Pasante de Marketing
Grupo de Medios Digitales Schuester
Marzo de 2020 a junio de 2020 (3 meses)
Aprendí sobre las operaciones del departamento
de Marketing de la empresa
Coordiné ferias comerciales y eventos
Resumen
Soy una gestora de marketing
digital creativa y colaborativa
con años de experiencia. Me
especializo en marketing entre
canales, comunicación digital y
gestión de redes sociales.
Habilidades y
competencias
Optimización de motores de
búsqueda (SEO)
Gestión de redes sociales
Redacción de contenido web
Datos de contacto
Perfil de LinkedIn:
linkedin.com/in/nombre
hola@sitioincreible.com.ar
(54) 1234-5678
www.sitioincreible.com.ar
Carla Iglesias
Ejecutiva de marketing digital

The experience reads in order, which is exactly what a template should do. But the stream opens with the portfolio link ("click here to see my portfolio"), and the candidate's name and title are the last two elements of page 1: expand the block and the last two lines are "Carla Iglesias" and her title. A parser that looks for the identity at the top finds it at the bottom.

The second caveat: the exported PDF includes a third page with the template's internal instructions, Canva's text inside "your" resume. Delete it before exporting; whoever doesn't sends it to every employer.

Modern light green, ATS-labeled: the references before your own data

The second template from Canva's ATS set. The extracted text starts well and goes wrong immediately:

Alonso Ibáñez
Desarrollador de
software
REFERENCIAS
Manuel Dorantes Gil
Gerente de Software
Software de Estrategia
123 456 789
Javier Acosta Ramos
Desarrollador de software
sénior Estrategia de software
123 456 789
DATOS DE CONTACTO
HABILIDADES
Python
Java
JavaScript
C++
SQL
EXPERIENCIA LABORAL
PERFIL PERSONAL
Soy un desarrollador de
software con tres años
de experiencia
trabajando con bases
de datos, servicios en la
nube y tecnologías
front-end.
EDUCACIÓN
Teléfono:
123 456 789
Correo:
hola@unsitiogenial.es
LinkedIn:
www.unsitiogenial.es
Dirección:
Calle Cualquiera, 123.
Cualquier Ciudad.
Desarrollador de software
Desarrollador de software
Agosto. 2031 - Sept. 2034
Colaboré en la mejora
de una plataforma de
e-commerce,
aumentando un 40 %
las ventas online en el
primer año.
 Optimicé el código,
reduciendo un 30 % el
tiempo de carga y
mejorando la
experiencia de usuario.
 Entregué 10 proyectos
exitosos a tiempo y
presupuesto,
contribuyendo a un
20 % más de
satisfacción del cliente.
Liceo “Jacinto Lara”
Máster en Programación
de Automatismos
Cursando desde mayo
de 2031 hasta la fecha
Cursos de
especialización en
ciencia de datos
Optativas en
ciberseguridad
Universidad Europea Edil
Grado en Ciencias de
Programación
2026-2030
Delegado de clase
Fundador del Club de
Programación Blanca
Egea
Campeón de mus de
la Universidad

The name comes first, which is good. But the next thing is the full references block, two people with their phone numbers, before the candidate's own contact data. "DATOS DE CONTACTO" and "HABILIDADES" arrive fused. And look at lines 20 and 21: the "EXPERIENCIA LABORAL" heading is immediately followed by the personal profile, while the actual jobs appear much later in the stream, far from their heading. A parser that assigns content to the nearest heading ends up with the profile filed as experience and no jobs under any heading. As a bonus: the role line appears duplicated, and the contact labels ("Teléfono:", "Correo:", "LinkedIn:") once again arrive separated from their values.

The filter doesn't use AI to rebuild your resume

There's an idea worth dismantling here: "the system is probably smart, it'll figure it out". The traditional ATS that filters applications doesn't work that way. Its reading is the one you just saw: extract the text in the order it comes out, split it into fields with rules (known section headings, label-value pairs, dates), and compare the result against the posting. There is no AI on the other side rebuilding the document.

We say this knowing exactly what the alternative looks like, because we built it: Dopplio's reader does use AI to interpret the document and reassemble it before evaluating it. But the filter at the company you're applying to isn't ours, and you can't count on it doing that work. For a literal reader, a resume that comes out scrambled is a fatal flaw, because nothing downstream repairs it: if the profile got attached to the experience heading, that's how it gets filed; if the skills came out as loose numbers, there are no skills to match against the posting.

Why it happens: the failure modes a parser doesn't forgive

None of this is a secret defect in Canva. These are known formatting failure modes, documented by sources that don't sell templates; our excerpts just show them live.

UConn's Center for Career Readiness names Canva directly: "Digital tools like Canva, LaTeX, and other résumé/CV builders may use formatting that is unreadable by Applicant Tracking Systems", and elements like images, text boxes, columns, and tables can make it difficult for the ATS to decipher your content.

The University of Minnesota Duluth's career center describes the mechanics without naming any tool: an "ATS reads left to right across the page even with tables and columns"; from graphics, graphs, logos, or images it "is not able to pull information"; and headers and footers "may get dropped completely". That's exactly what you saw above: the sidebar interleaved into the reading and the headings detached from their sections.

And The Muse explains the underlying mechanism, quoting career coach Leto Papadopoulos: "Every bit of text on a Canva resume template is placed using text boxes... it won't help in getting your content read properly by an ATS." The text is there, but it floats in loose boxes, and the order those boxes dump into the text has no reason to match the order you see them in.

The practical lesson isn't "don't use Canva". It's: any template with columns, icons, level bars, or text boxes needs the paste test before you send it. The full method for getting your resume ready against a posting is in how to pass the ATS filter.

Boring on purpose: the alternative that can't fail this way

Here's the house position, stated openly. Dopplio's templates look boring next to Canva's, and that's a deliberate decision.

Our resumes are built programmatically: a single column, real text top to bottom, standard section headings. No text boxes, no sidebar, no icons, no headers or footers. Go back over the failure list UConn, UMN Duluth, and The Muse warn about: every element they flag is an element these templates structurally cannot contain. They don't dodge the mistakes; the mistakes have nowhere to happen. Run the paste test on one of our PDFs: it comes out in the exact order you see it.

That's the honest trade-off: pretty for the eye or readable for the machine. And since the first read of your resume is done by a program, we chose the machine. You can build your resume in the builder free, or copy the structure by hand from the ATS template that does pass.

If you already have your Canva resume, don't rebuild it blindly: run the paste test first. If it comes out in order, congratulations, you picked the right template. If it comes out scrambled, you now know exactly what reaches the parser and why.