← Back to Companies

Productivity Software

Canva

Watchful Central finding Level favourable, direction adverse Proposed pattern Allowance Compression Proposed Methodology CHI/CVI v2.0

Canva still delivers substantially more than it extracts. The concern is what changed after 2024.

Canva is the least customer-hostile and most customer-favourable company CHI has assessed to date. Roughly 234 million of its 265 million monthly users pay nothing, the free tier has grown on every measurable dimension, Canva Pro costs less in real terms than at its 2015 launch under both candidate US prices, and three acquired paid products — Affinity, Cavalry, Pexels/Pixabay — were made or kept free. That is the level, and it is genuinely favourable. The direction is not. Every adverse finding in the final research dossier is dated 2024 or later: a team-tier repricing of +316.7% delivered by private email and partially reversed, a new SKU that reached double the disputed rate thirteen months later without triggering any notice obligation, and a silent ~60% compression of included AI image and video capacity accompanied by the simultaneous appearance of paid overflow. Canva is not a prosecution case. It is a historically customer-friendly company in year two of a deliberate, IPO-motivated monetisation transition — and that transition, not the score, is the story.

CHI38/100Watchful
CVI89/100Exceptional
CFS+51Strongly Customer-Friendly
Level 38 — Watchful The lowest CHI yet published. Hostility is concentrated, not systemic.
Direction Adverse since 2024 A qualitative pattern trajectory, not a second numerical score.

CFS = CVI − CHI (89 − 38 = +51). Assessed under CHI/CVI Methodology v2.0: dimensions score the underlying customer harm, patterns describe the behavior that produced it. The direction indicator reflects the documented concentration of adverse findings from 2024 onward, not a calculated historical CHI time series — no prior CHI score for Canva exists. Methodology →

Pattern AnalysisAll patterns →
ESTABLISHED · SCOPED

Allowance Compression

ESTABLISHED

Reference-Price Erasure

MIXED / SEGMENTED

Price Creep

EMERGING

Grandfather Starvation

Pattern Heatmap
Teams/Business repricing (2024–25)
Silent AI allowance compression
Reference-Price Erasure
Marketing vs contract gap
Upgrade pressure on Free
Grandfather Starvation
Exit Resistance
Customer Lock-In*
Not established
Strongly supported
* Customer Lock-In is shown at minimum weight because the final dossier finds it contradicted: editable-format export, explicit ownership, perpetual content licences, local Affinity files, and an MCP server that lets rival AI assistants read and edit Canva content. Advertising Creep (unsupported: no in-product advertising in thirteen years) is omitted entirely.

The defining qualifier

The direction changed in 2024

Every adverse finding in the final dossier is dated 2024 or later. Nothing from 2013–2023 registers as hostility on any pattern. Fact Canva's COO stated on the record that the 2027 IPO was deferred so the shift "from feature-based SaaS subscription tiers to an AI credits system" could be "really bedded in so we're not having to explain ourselves to the market through a transitional period." Canva claim

The level is favourable. The direction is not. For this company, the direction is the more analytically important finding.

CHI dimension breakdown (higher = more hostile)
AI / metering conduct58/100
Transparency & promise integrity53/100
Pricing conduct48/100
Data & privacy conduct41/100
Subscription & exit conduct32/100
Free-tier conduct20/100
Lock-in / portability20/100
Advertising conduct12/100

Executive finding

Unusually customer-favourable. Watch the trajectory.

CHI credibility requires saying plainly when a company performs well. Canva performs well — and the same evidence base documents a monetisation transition that began in 2024. Both facts are held at once, deliberately.

The level — CHI 38, CVI 89, CFS +51

~234 million of 265 million monthly users pay nothing, and the free tier's contents are confirmed from Canva's own live pricing page: 1M+ templates, 4.7M+ assets, 1 Brand Kit, 5GB storage, access to Affinity, and up to 200 Standard or 20 Premium AI uses per month. Fact Canva Pro's 2015 annual price is US$168.47 in 2026 dollars against a current US$119.99–144 — a real-terms decline under both candidate figures. Fact Hostility is concentrated in the team/business segment and the AI metering layer, not systemic. Analytical inference

Canva delivers more value than it extracts, by a wide margin. The positive finding is not buried here; it leads.

The direction — adverse since 2024

Team pricing moved from US$119.99 for five people to US$500 (2024), then a successor SKU at US$1,200 (2025). Fact Included AI image and video capacity on paid individual tiers fell ~60% nominally between April and June 2026, unannounced, while paid overflow appeared in the same window. Fact A recurring gap separates what marketing promises (60 days' notice) from what the contract commits (30 days). Fact

Canva has begun extracting faster in specific parts of the business. The concentration of adverse findings after 2024 is the reason the status is Watchful.

The inflection

2013–2023: nothing. 2024 onward: everything.

Every meaningful adverse finding in the final dossier dates from 2024 or later. Chronological concentration is documented as a pattern trajectory; causation is not asserted from chronology alone.

2013–2023 Nothing from this period registers as hostility on the final pattern assessment. Fact
APR / SEP 2024Teams repricing: silent per-seat pricing for new customers, then legacy migration attempt by private email (US$119.99 → US$500/yr for five seats).
2024Pro trial standardised 45 → 30 days. Single secondary source
NOV 2024Nonprofit eligibility restricted: group-exemption subordinate organisations excluded from free access.
FEB–JUN 2025Template Links become paid-only to create — the one confirmed free-tier removal in thirteen years, narrower in scope than widely reported.
OCT 2025Affinity support forum made read-only, replaced by Discord. In the same month: Canva Business launches at double the 2024 per-seat rate and Canva Teams closes to new sign-ups and upgrades.
FEB 2026Nigeria: legacy pricing eliminated, Pro +97% with no grandfathering — the counter-example to the Teams grandfathering story.
MAR 2026Affinity Additional Terms add an "entitlement check" clause absent from the October 2025 version.
APR–JUN 2026AI allowance compression: pooled Standard/Premium/Ultra structure replaces per-feature allowances at ~60% less image/video headroom; AI Pass and AI top-ups appear in the same window. No announcement, no changelog.
H1 2026Probable quiet US pricing increase (Pro toward $144/yr, Business toward $250/seat/yr), unconfirmed from Canva and not asserted. Probable

Teams & Business pricing · Hero exhibit

The clearest traditional hostility event.

The same five-person team, US legacy pricing, before and after the 2024 migration attempt.

Legacy Canva Teams (pre-April 2024)

$119.99per year, up to 5 people

  • One paying account holder
  • "Up to four extra seats at no additional cost" — Canva's own phrasing
  • ≈ $2.00 effective per person per month

2024 migration attempt (Sep 2024)

$500per year, 5 people ($10/person/mo)

  • Flat team bundle
  • Per-seat pricing, 3-seat minimum (+316.7%)
  • Notified by private email (no public announcement)
Five-seat annual increase attempted +316.7%

The per-seat model had already been introduced quietly for new customers months earlier, in April 2024. Affected legacy subscribers were notified privately in September; backlash followed, and Canva reversed the migration for eligible legacy customers roughly five weeks later. That reversal was genuine and has remained honoured through the final dossier date — but it applied only to the legacy cohort. The new pricing structure remained in force for new customers throughout.

The +316.7% figure describes the US market, where the legacy price was anomalously low; the Australian equivalent was +69% and a UK customer report indicates +228%. Fact Customer report

Reference-Price Erasure · The 2025 sequel

The backlash killed a migration. It never killed the price.

Step 1 · Apr–Sep 2024

Silent rollout, private email

Per-seat pricing introduced quietly for new customers, then a migration attempt on the legacy base via private email. Canva's newsroom has never announced a price increase — only the reversal. Fact

Step 2 · Oct 2024

Partial reversal + Pricing Promise

Legacy customers keep their rate indefinitely, including free extra seats — honoured for 22 months and re-affirmed in writing. New customers keep paying the new rate. The list price never came down. Fact

Step 3 · Oct 2025

SKU Reset: Canva Business

Teams closes to new sign-ups and upgrades; Canva Business launches at US$20/person/month — exactly double the 2024 rate that triggered the backlash. No existing customer's price changed, so no notice obligation was engaged. Fact

SKU Reset — a named mechanism under Reference-Price Erasure, not a standalone pattern

The old reference product disappears from sale; a new tier arrives at materially different economics; existing legacy customers technically experience no price increase; therefore no price-change notice obligation is triggered. A prospective 2026 customer cannot purchase, price, or even see the tier that delivered comparable functionality at half the per-seat rate thirteen months earlier. Fact

The dossier establishes the effect, not the intent: Canva has never stated that avoiding the notice obligation was a purpose of the Business launch, and no document evidences it. A new SKU with a genuinely different feature set — and no seat minimum, which made solo and two-person teams strictly better off — is a defensible product decision on its own terms. Analytical inference

Five-seat arithmetic: $119.99 (legacy) → $500 (2024) → $1,200 (Business, 2025). The disputed economics were reached and passed, under a new name, with no comparable press cycle.

Price Creep · Final verdict is segmented

One verdict would be false. The index records the split.

Averaging these five relationships into a single "Canva has Price Creep" claim would misstate the evidence in both directions. The final dossier verdict is MIXED / SEGMENT-DEPENDENT.

Canva Free — CONTRADICTED Price has been $0 since 2013 and delivered value expanded on every measurable dimension — storage 5×, assets ~17×, templates ~4×, Brand Kit partially de-paywalled, AI caps moved from lifetime to monthly refresh. No mechanism for price creep exists.
Canva Pro / individual — CONTRADICTED (lifetime) The 2015 annual-billed price (US$119.40) is US$168.47 in 2026 dollars. Whether the current US price is $119.99 or $144, the lifetime real-terms change is negative: −28.8% or −14.5%. The 2023→2026 window flips sign depending on the unresolved US figure and is not asserted.
Legacy Canva Teams — CONTRADICTED (applied price) The grandfathered rate has been preserved since October 2024, re-affirmed in writing in October 2025 including the free-extra-seats mechanic. No price increase has been applied to this cohort. The adverse finding here is Grandfather Starvation, not Price Creep.
New Teams / Canva Business — ESTABLISHED, severe $30/seat/yr (Jul 2023) → $120/seat/yr (Apr 2024) → $240/seat/yr (Oct 2025) → probably $250 (2026). Real-terms +264% on the first step alone. This is where the significant price escalation exists, and the directly observed Czech Business price independently confirms the tier's current economics.
Regional — MIXED, net favourable Canva's ~5× purchasing-power localization (Nigerian Pro at ~US$29/yr against a probable US$144 US price) is genuine, steep, pro-customer pricing that most global SaaS vendors do not attempt. But Nigeria's legacy cohort was repriced +97% in February 2026 with no grandfathering — an important adverse exception — and the AI Pass add-on is not localized at all.

Allowance Compression · Proposed pattern · Hero exhibit

The subscription price didn't move. The capacity inside it did.

Canva Pro's expensive-inference AI allowance, before and after the ~April–June 2026 window.

Documented as current on 11 Apr 2026

500generations / month

  • Per-feature monthly allowances
  • "Canva offers no overflow option" — verbatim, independent source
  • Headline subscription price unchanged

Canva's help centre, by June 2026

2,000 / 200 / 20Standard OR Premium OR Ultra uses — one pooled budget

  • Comparable image headroom: 200 Premium/mo (−60% nominal)
  • Video: 20 Ultra/mo, drawn from the same pool (−60% nominal)
  • No paid overflow — AI Pass (US$100/person/mo) and AI top-ups appeared in the same window
Image/video capacity reduction, compounded by pooling ~60%

The labels above are not one-to-one equivalents: the new system is pooled and tiered, with Standard, Premium and Ultra uses drawing on one shared budget at roughly 1× / 10× / 100× weightings. That pooling compounds the cut — a Pro user who spends 20 Ultra video generations has consumed the entire budget and has zero left for images. 200 images and 20 videos in one month is no longer possible. No announcement, changelog, effective date or grandfathering was found; Canva's help page presents the current numbers as though they had always applied.

A conventional price tracker sees "subscription price unchanged." The customer receives less usable capacity, and the shortfall is monetised through paid overflow — so the economic price rises while the headline price does not. Analytical inference

Allowance Compression · Definition and scope

A new taxonomy candidate — precisely scoped.

Proposed pattern N1 — ESTABLISHED for Canva, scoped

Allowance Compression

Same subscription → less included capacity → paid overflow

A vendor reduces the metered capacity included in an unchanged subscription after customers have integrated that capacity into normal workflows, then monetises the resulting shortfall through paid overflow. Because the headline price does not move, no price-change notice obligation is triggered and no price tracker registers a movement. Proposed definition

Canva's own 60-day Pricing Promise covers price changes and is silent on allowance changes — a loophole in a commitment Canva itself authored. Fact Customer reports add that failed generations still consume allowance, with no refund mechanism. Customer report

The economic effect is a price increase; the disclosure obligations are not.

Mandatory scope limit — what this does NOT say

Not all Canva AI got worse

The pattern is established only for expensive-inference capacity — image, video and agentic generation — on relevant paid tiers. The dossier explicitly finds the opposite movement elsewhere: Magic Write went from 500/month metered to unmetered fair use; Background Remover was never converted into an AI credit sink; and Canva Free improved structurally, from lifetime AI caps to a monthly refresh. Fact

Unmetering the cheap inference while tightening the expensive inference is coherent marginal-cost behaviour. The pattern captures the disclosure and notice failure, not the metering itself. Analytical inference

Exhaustion throttles rather than blocks — "short pauses between generations," not a cutoff. Stated fairly.

The AI Pass regional asymmetry

Canva materially localizes its core subscription prices — Pro annual runs from ~US$29 in Nigeria to a probable US$144 in the United States, a ~5× spread. Fact Canva states that "AI Pass is USD $100 globally per person, per month." Canva claim The add-on can therefore represent radically different economic burdens across markets: roughly 8–10× a US Pro subscription, and roughly 41× the monthly Pro subscription in Nigeria. Analytical inference

Regional pricing on the way in; globally flat capacity pricing once you're there. The tension between the two is the finding.

Grandfather Starvation · Proposed pattern · Emerging

The price is honoured. The roadmap is the question.

Proposed pattern N2 — EMERGING: mechanism established, harm not yet established

Grandfather Starvation

Legacy price preserved → feature roadmap frozen or narrowed → grandfathering value decays

A vendor preserves a legacy price while narrowing or freezing the feature roadmap available to that legacy cohort, causing the economic value of grandfathering to decay over time. Proposed definition

What is established for Canva: legacy Teams customers retain their pricing, including free extra seats; Teams is closed to new sign-ups and upgrades; Canva says Teams receives new product updates "as they roll out to Canva Pro" — the individual track; and new team-oriented capabilities (Grow Insights, higher AI limits, approvals, 100 Brand Kits, 500GB storage, enterprise integrations) are shipping to Business. Fact

What is not established: that any customer has been forced to migrate. No legacy Teams customer has been forcibly moved, no complaint of forced migration was found, and the price commitment has been kept for 22 months. The inference is only that this arrangement may eventually create voluntary migration pressure. Analytical inference

This is a monitoring classification, not completed harm. The trigger to watch: the first documented case of a legacy Teams customer needing a Business-only capability.

Affinity · The strongest falsifier

Evidence against a high-hostility conclusion.

Canva acquired Serif/Affinity in March 2024 and made the professional suite free on 30 October 2025. This is presented as a genuine customer-positive case in its own right, not as mitigation inside a hostility section.

Delivered value — verified through v3.2.3, July 2026

A ~£145 perpetual-licence professional suite became free, and in the 21 months to the dossier date no previously free core non-AI feature had been clawed back — across five releases that shipped substantial free functionality. Fact Files remain local by default; export includes PSD, AI, PDF, SVG, TIFF and IDML; Capture One and DaVinci Resolve have adopted the .af format. Fact Canva then made Cavalry free as well (April 2026), and Pexels and Pixabay have remained free for seven years after acquisition. Fact

Three separate opportunities for enclosure. Three declined. This is what a hostility thesis most has to explain away.

The qualification — risk is not harm

The Affinity Additional Terms effective 12 March 2026 added an "entitlement check" clause absent from the 30 October 2025 version, with no interval, frequency or grace period specified; the same document reserves Canva's right to disable access "at any time without notice." Fact The no-AI-training promise lives in help-centre prose and executive statements, while the binding terms are silent. Fact That creates structural dependency risk.

But the value is delivered today. The dependency risk is real, and the feared clawback has not happened — Canva is not penalised on this page for hypothetical future enclosure. Analytical inference

A documented contractual change in the customer's disfavour, with no realised harm. A monitoring trigger, not a finding of hostility.

The free tier · What the evidence rejects

"Canva keeps paywalling things that used to be free" is not supported.

The free tier has increased in absolute value across thirteen years. One narrow removal is confirmed — Template Links, June 2025 — and it must not be generalized.

Canva Free as directly observed, 10 Aug 2026 (Czech pricing page)

0 Kč · 1M+ templates · 4.7M+ photos, videos, graphics and audio · 1 Brand Kit · 5GB storage · access to Affinity · up to 200 Standard or 20 Premium AI uses per month. Fact — directly observed

Historical increases include storage expansion (1GB → 5GB), asset expansion (~17×), template expansion (~4×), partial Brand Kit access, improved AI availability (lifetime caps → monthly refresh; Magic Write to unmetered), Canva Code 2.0 shipped to the free tier, and Affinity access. Fact

On every measurable dimension, Canva Free became more generous between 2013 and 2026.

Why customers still feel a squeeze

Canva ships new premium capability even faster than it expands Free, so Free may represent a shrinking share of total capability even as its absolute value rises. That is Paywall Creep by accretion, not Feature Erosion — the distinction matters, and the final dossier scores them differently. Analytical inference

The one confirmed removal: from 2 June 2025, only paid plans can create Template Links. Pre-existing links keep working, and free users on a different email domain from the template owner can still open and edit shared templates — the practical bite falls on same-domain free users inside an organisation. Fact

One genuine free-tier removal in thirteen years — narrow, confirmed, and not a general practice.

Upgrade pressure — real, and quantified

Free-only content filtering is not the default and premium and free assets appear together in search. Fact Premium assets vastly outnumber free ones: free templates are roughly 28% of Pro's template count, and free assets roughly 3.3% of the premium-scale library — both figures now directly observed on Canva's own pricing page. Fact Designs created during a trial using premium assets can later encounter watermark pressure after trial expiry unless the user upgrades or strips the premium elements. Fact

This is why Canva can feel increasingly paywalled while free value rises in absolute terms. It is not evidence that free functionality is being systematically removed.

Promise integrity · Exhibit

Marketing commitment > contractual commitment.

Pricing Promise (marketing)

60 days

"We'll always provide at least 60 days' notice of any price changes." Canva claim

Terms of Use (binding)

30 days

Price increases apply at "your next renewal or thirty (30) days after notice, whichever is later." The enforceable commitment is half the advertised one. Fact

Enterprise MSA (binding)

45 days

A third notice standard, in a third Canva document. Fact

Framed precisely

This is a gap between marketing and contract — not a documented breach of the 60-day promise. No specific event is established in which Canva gave less notice of a price change than promised; the Business launch structurally never engaged the promise, because no existing customer's price changed. Analytical inference

The more consequential omission: the Pricing Promise covers headline price and is silent on allowance reductions. Reducing a Pro user's monthly image generations from 500 to 200 is economically a price increase, and it triggers no notice obligation under any of the five principles. That gap is load-bearing for the Allowance Compression story. Analytical inference

The pattern is not broken promises. It is a systematic gap between promotional and contractual language.

Pattern verdicts · Final dossier

What the evidence actually supports.

Major adverse patterns

Allowance Compression Proposed ESTABLISHED — scoped to expensive-inference capacity (image, video, agentic) on paid individual tiers. Capacity fell ~60%, monetised overflow appeared, the headline price was unchanged, and nothing was announced. Not established for text generation, on-device tools, or the free tier, where capacity increased.
Reference-Price Erasure ESTABLISHED — four mechanisms: SKU retirement (Teams withdrawn from sale), rename-repricing (Business at 2× the disputed rate with no notice obligation engaged), retroactive archive editing (the 2015 "Canva for Work" announcement now reads "Canva Pro"), and unpublished prices. The cumulative effect: Canva's price history is not independently auditable.
Price Creep → MIXED / SEGMENT-DEPENDENT — established and severe for new Teams and Canva Business; contradicted for Canva Free and, on a lifetime basis, for Canva Pro; neutral-to-favourable for low-income regional markets with a documented exception in Nigeria. The segmentation is preserved deliberately; averaging it would be false.
Subscription Creep ESTABLISHED — three stacked layers on one product: subscription → AI Pass at US$100/person/month (8–10× the subscription it attaches to, unlocking no new features) → AI top-ups (30-day expiry, non-cancellable, eight countries, price unpublished).
Feature Fragmentation → ESTABLISHED — one "AI" capability split across seven commercial surfaces: unmetered fair-use tools, Standard, Premium, Ultra, AI Pass, AI top-ups and Canva Credits. Privacy controls span four surfaces. Upgrade Escalation → ESTABLISHED — $10 → $20/seat/month in thirteen months; a 3-seat minimum that forced solo users to buy three seats (Apr 2024–Oct 2025); AI Pass as the escalation path above the plan; premium content mixed into default search; the post-trial watermark trap. Offset: the Business launch removed the seat minimum.
Data Creep ESTABLISHED — AI-training rights asserted in June 2023 with no opt-out for six months; "Memories" makes transient analysis a persistent profile; connectors to Gmail, Drive, Calendar, Slack, Notion and Zoom with undocumented governance; data-broker enrichment and LiveRamp identity matching.
Grandfather Starvation Proposed EMERGING — mechanism established, harm not yet established. Legacy Teams pricing is honoured while the cohort's roadmap is pinned to the Pro track and new team functionality ships to Business. No forced migration has occurred or is claimed.

Probable patterns

Choice Illusion → PROBABLE — free-only filter default-OFF and re-applied per search; "up to X" allowance language demonstrated uninformative; failed AI generations consume allowance with no refund; opt-outs scattered across four-plus surfaces, two partly email-gated; the contributor AI opt-out forfeits payment eligibility. Offset: output ownership is unconditional and exhaustion throttles rather than blocks. Paywall Creep → PROBABLE — by accretion, not withdrawal. Free holds ~28% of templates and ~3.3% of assets; every substantial 2015–2026 capability shipped into paid tiers. Offset: folders and Brand Kit moved to free, Canva Code 2.0 shipped to free, free assets grew ~17×. Access Downgrading → PROBABLE — legacy Teams receives updates "as they roll out to Canva Pro" while new team functionality ships to Business; Canva Create 2024 moved admin/SSO/SCIM into a tier above Teams in the same period Teams was repriced. No existing paid tier has had a feature actively removed at unchanged price. Promise Reversal → PROBABLE — 60 days advertised vs 30 contracted; Nigeria's legacy cohort terminated at +97%; the Affinity Pledge's perpetual-licence commitment retired without acknowledgement; marketing-only AI carve-outs. Substantial offset: the single biggest promise — the October 2024 Teams grandfathering — has been honoured for 22 months and re-affirmed in writing.
Surveillance Creep PROBABLE — FullStory session replay listed with no disclosure of scope, masking or retention; "Living Memory" with no published retention period or deletion mechanism; uploaded content analysed for "predicting the most relevant subscription or product offerings." Offset: the enumerated product-functional ML purposes are legitimate.
Tracking Creep PROBABLE — Facebook pixel with advanced matching; LiveRamp identity matching with a partly email-gated opt-out; OneTrust deployed on acquired properties but not canva.com. Offset: the stack is unremarkable for consumer SaaS at this scale, and Education students are carved out in binding text.

Falsifiers and negatives — patterns the evidence does not support

Customer Lock-In Contradicted Editable-format export (PPTX, DOCX, XLSX, CSV, SVG, PDF, HTML bundle); explicit ownership; perpetual content licences; Affinity files local by default; .af adopted by Capture One and DaVinci Resolve; an MCP server that lets ChatGPT, Claude, Gemini, Perplexity and Copilot read and edit Canva content. Caveats: a Canva account is now contractually required for Affinity and Cavalry, and bulk export delivers flattened images with the editable path capped at 500 exports/day.
Exit Resistance → Possible — not established The Terms of Use guarantee self-service cancellation. No regulator, consumer body, class action or Hacker News thread raises cancellation friction. The authenticated UI was not tested and nothing adverse is inferred from that gap; the verdict rests solely on the documented seat-billing asymmetry and the absence of consumer-law acknowledgement in the contract.
Feature Erosion Possible — narrow Exactly one confirmed free-tier withdrawal in thirteen years (Template Links, June 2025), narrower in scope than widely reported. Not presented as broadly established; the substantive capacity reduction is captured by Allowance Compression instead.
Advertising Creep → Unsupported No third-party advertising served to Canva users, on any tier, in thirteen years. Education students are explicitly carved out of advertising in binding text.

Allowance Compression and Grandfather Starvation are Proposed taxonomy candidates produced by this investigation, not yet formal entries in the sitewide CHI Lexicon; they are shown unlinked until dedicated pattern pages exist. SKU Reset is a named mechanism under Reference-Price Erasure, not a standalone pattern.

Evidence against hostility

The case for the defense.

This section carries equal weight to the patterns above, not a footnote. It is the evidence behind the CVI 89 — and several of these items are independently sufficient to defeat a high hostility score.

01

An extremely large functioning free user base: ~234 million of 265 million monthly users pay nothing, and the free tier's contents are directly verified on Canva's own pricing page rather than relayed from marketing.

02

Substantial free-tier expansion over thirteen years: storage 5×, assets ~17×, templates ~4×, Brand Kit partially de-paywalled, Magic Write unmetered, AI caps moved from lifetime to monthly refresh, and Canva Code 2.0 shipped to the free tier.

03

Long-term real-price decline for Canva Pro: the 2015 annual price is US$168.47 in 2026 dollars against a current US$119.99–144 — negative under both candidate figures, while the product gained at least eight product categories that did not exist in 2019.

04

A genuine price reversal for legacy Teams customers within five weeks of the 2024 backlash — and grandfathering honoured for 22 months since, re-affirmed in writing, including the free-extra-seats mechanic.

05

Affinity made free (21 months, five releases, zero core clawbacks), Cavalry made free, and Pexels/Pixabay still free seven years after acquisition — three declined opportunities for enclosure.

06

Editable-format export (PPTX, DOCX, XLSX, CSV, SVG, PDF, HTML bundle), explicit user ownership of content and AI output on all tiers, and perpetual content licences.

07

Strong interoperability, including an MCP server that lets ChatGPT, Claude, Gemini, Perplexity and Microsoft Copilot read and edit a user's Canva content — the structural opposite of enclosure.

08

Substantive accessibility work: WCAG 2.1 AA, a published VPAT, third-party audits with Intopia, disability research with Fable, JAWS/NVDA/VoiceOver testing, and 100+ languages.

09

No evidence of in-product third-party advertising on any tier in thirteen years, with Education students carved out of advertising in binding policy text.

10

No established customer lock-in and no established exit resistance: self-service cancellation is contractually guaranteed, price increases are deferred to renewal or 30 days, and a pro-rata refund is owed if Canva discontinues a paid service.

Canva exhibits several genuine customer-hostile behaviors — concentrated since 2024 — while still delivering substantially more customer value than measured hostility overall. CHI and CVI are scored separately, on purpose: CVI 89/100 sits far above CHI 38/100, and CFS +51 reflects that gap directly.

What remains unresolved

Open research questions, not published claims.

The 2026 US list priceProbable but unconfirmed: five 2026 sources and three triangulation checks favour Pro $144/yr and Business $250/seat/yr, but no Canva announcement or confirmation exists. Not asserted — and no longer material: Pro's lifetime real-terms decline holds under both candidate figures.
The cancellation and downgrade UINot tested from an authenticated account. Per the dossier's rule, no adverse weight is assigned from the absence of a test; Exit Resistance is not inferred from an untested UI.
AI-training toggle defaultsPublic sources directly contradict each other on whether the toggles default on or off for Free/Pro users. Preserved as unresolved and scored in neither direction.
Affinity's 365-day revalidation questionWhether the entitlement check runs periodically is unresolved: the annual-expiry claim traces to one hedged clock-forward self-experiment, and Canva's own help article on offline authentication could not be read. What is resolved: the contractual right to check was added in March 2026.

Claims explicitly rejected

CHI is not simply compiling complaints.

"Canva keeps taking things that used to be free and putting them behind Pro"Rejected as a general pattern. The free tier expanded on every measurable dimension; most frequently-cited "paywalled" features (Magic Resize, Background Remover, Content Planner, custom fonts) were born paid and were never free.
"All Canva AI got worse"Rejected. Magic Write became more generous and unmetered, Background Remover was never turned into a credit sink, and the free tier's AI structure improved from lifetime caps to monthly refresh. The compression finding is scoped to expensive-inference capacity on paid tiers.
"Legacy Teams customers have been forced onto Business"Not established. No forced migration has been documented; the grandfathered price has been honoured for 22 months. Grandfather Starvation is classified as emerging — mechanism established, harm not established.
"Canva broke its 60-day notice promise"Not established as an event. The documented finding is a standing gap between the marketed 60 days and the contracted 30 — and that the promise covers price, not allowances.
"The whole 2024 price rise was reversed"Rejected. Only the retroactive migration of the legacy base was reversed. The new per-seat pricing remained for new customers, and the list price never came down.
"Canva intended the Business launch to evade its notice obligation"Not asserted. The dossier establishes the effect — the disputed economics were reached with no notice obligation engaged — not the intent. No document evidences intent.
"Affinity free is a bait-and-switch"Not supported to date. No previously free core non-AI feature has been clawed back in 21 months. The entitlement-check clause is a documented dependency risk, and risk is not harm.
"Cancellation is a gauntlet" / "Canva watermarks work you already downloaded"Unproven and not published as findings. No independent public evidence of manipulative cancellation friction exists, and no contractual basis or documented practice of retroactive watermarking of downloaded work was found.

What Canva didn't do alone

  • The AI metering has a documented economic basis: a US$1.25bn GPU cloud commitment, 100+ AI researchers, a growth forecast cut from 30% to 20%, and Canva's co-founder on record that the cost of serving an AI task was "too high." The metering is not invented scarcity — the scored hostility is in the disclosure practice, the top-up mechanics, and the refusal to localize AI Pass, not in the existence of limits.
  • Every AI-era subscription business now prices metered inference inside flat subscriptions. Allowance Compression is proposed as a general taxonomy entry precisely because the mechanism will generalise beyond Canva; Canva is its first documented case here, not its only plausible one.
  • Canva's ~5× regional price localization is deeper than most global SaaS vendors attempt and materially widens access in lower-income markets; the adverse findings (Nigeria's legacy termination, the non-localized AI Pass) sit inside a structure that is otherwise pro-customer.
  • The 2024 Teams increase corrected a US legacy price that was anomalously low — Australian customers had been paying roughly four times the US rate for the identical bundle. That does not excuse the method; it qualifies the headline percentage.

These concessions strengthen rather than weaken the assessment: CHI scores Canva's particular implementation choices — the silent rollout, the private-email migration, the unannounced allowance cut, the SKU reset — against the realistic alternative of the same commercial strategy executed with honest disclosure, not against an idealized company with no commercial pressure at all.

Methodology note

This page is built from the completed Canva Customer Hostility Index final research dossier (10 August 2026) under CHI Methodology v2.0, which passed its publication gate. Evidence labels — Fact, Canva claim, Customer report, Analytical inference — preserve the dossier's evidentiary distinctions.

The dossier ran nine primary adversarial research streams plus a targeted closeout pass that resolved or partially resolved 22 of 28 open items, including corrections that reduced findings adverse to Canva. Source priority favours directly observed evidence (Canva's live pricing page, Czechia, 10 August 2026), Canva's binding policy documents and their public version archive, Canva newsroom and help-centre text quoted verbatim, regulatory registers, and dated contemporaneous journalism. Four items remain genuinely unresolved and are listed above; none is material to the score.

CHI 38/100 (Watchful), CVI 89/100 (Exceptional) and CFS +51 (Strongly Customer-Friendly) are the settled analytical values for this assessment and are not presented as a range. The trajectory qualifier — every adverse finding dated 2024 or later — is a documented pattern concentration, not a numerical time series. Current pricing, allowances and plan details should be reconfirmed immediately before any republication, since Canva's price stack has been shown to change without announcement.

Final finding

CHI38/100Watchful
CVI89/100Exceptional
CFS+51Strongly Customer-Friendly

Canva is one of the strongest examples yet that the Customer Hostility Index is not a blacklist. Canva remains strongly customer-favourable overall: its free product is unusually substantial, Canva Pro has become cheaper in real terms over its lifetime, and several acquisitions expanded customer value rather than enclosing it.

But the company's behavior since 2024 deserves attention. Team pricing escalated sharply, AI capacity was quietly compressed while paid overflow appeared, and older pricing commitments are increasingly routed around through packaging and feature-roadmap changes rather than broken outright. Canva's score is low. Its trajectory is the reason to keep watching.

Canva delivers far more value than it extracts.
And it has been extracting faster, in specific places, since 2024.