Screenshot-based visual testing: a visual regression detection method that captures a full image of a web page or component, then compares it pixel by pixel against a reference image — any difference beyond a configurable threshold is flagged as a potential regression — and these false positives are a well-known pain point.
Percy has long been the first name that came to mind when talking about automated visual testing. Acquired by BrowserStack in 2020, the tool established itself as the "obvious" solution for teams wanting to add a visual comparison layer to their test pipeline. And for a long time, it was indeed the best available option — more by default than by excellence, but the best nonetheless.
The landscape has changed. And the question every team should ask today is no longer "how do I integrate Percy into my tests?" but "is Percy's approach still the most relevant for my needs?"
Delta-QA offers a radically different answer to the same problem. No screenshots, no cloud, no SDK, no per-screenshot billing. A deterministic, local, no-code, and free approach. Let's compare the two head-on.
What if testing every page never raised your bill? Delta-QA does visual testing locally, free in its Desktop edition, with no sign-up and no SDK to integrate. Try Delta-QA for free →
Percy: Visual testing as an extension of your existing tests
Percy works on a principle that seems simple: at strategic points in your automated tests, you trigger a screenshot capture. That capture is sent to BrowserStack's servers, where it's compared pixel by pixel against a reference image (the baseline). Differences are displayed in a web dashboard where your team can approve or reject each change.
The idea is elegant. The execution, less so.
To use Percy, you must first have automated tests — Cypress, Playwright, Selenium, Puppeteer, or one of the many supported frameworks. You then install the Percy SDK corresponding to your framework. You modify your tests to add capture calls at the relevant points. You configure an authentication token to communicate with Percy's servers. And you hope your CI pipeline has a stable enough network connection to send high-resolution screenshots on every run.
If you're a developer comfortable with testing tools, all of this is doable in a few hours. If you're not a developer — and a significant proportion of the people responsible for visual quality aren't — you depend entirely on someone else to set up and maintain your visual testing tool.
Delta-QA: Visual testing as a standalone activity
Delta-QA rests on the opposite premise: visual testing shouldn't be a graft on your functional tests. It's a standalone activity, with its own needs, its own users, and its own constraints.
In practice, you provide Delta-QA with two versions of a page (two URLs, two environments, or two points in time), and the engine reports the differences that matter. No screenshots sent to the cloud. No naive pixel diff. Delta-QA renders both pages in a real browser and compares the visual output, calibrated on human perception, to identify what actually changed.
The result is a clear report of the differences a human eye would notice: elements added, removed, modified, moved, with the property that shifted. No fuzzy interpretation, no "this pixel changed color by an imperceptible shade." Verifiable, explainable facts.
And all of this without writing a single line of code. If you can use a web browser, you can use Delta-QA. That sentence isn't a marketing slogan — it's literally the reality of the interface.
Code vs no-code: Why it changes everything
The most fundamental difference between Percy and Delta-QA isn't technical — it's organizational.
Percy is a developer tool. It lives in code, it's installed via npm or pip, it's configured in configuration files, and it runs in CI/CD pipelines. That's perfectly logical when your testers are developers. It's a wall when they're not.
Think about your team. Who cares about the visual quality of your product? Developers, certainly. But also the designers who created the mockups. The product owners who validated the user journeys. The QA managers who define acceptance criteria. The marketing leads who manage content pages. None of these people should need to write code to verify that the site looks correct.
Percy structurally excludes them from the visual testing process. They can view the results in the dashboard — approve or reject changes — but they can't configure new tests, add new pages to monitor, or adjust detection parameters. They are passive consumers of a tool designed for others.
Delta-QA includes them. Anyone on the team can launch a visual comparison, add pages to monitor, configure sensitivity thresholds, and interpret results. Visual testing becomes a shared responsibility instead of a delegated technical task.
Per-screenshot billing: A model that punishes rigor
Percy's pricing model deserves particular attention because it creates a perverse incentive.
Percy charges per number of screenshots per month. Plans start with an included screenshot quota, and each additional screenshot beyond the quota is billed. The more you test, the more you pay. The more rigorous your visual coverage, the higher the bill.
Think for a moment about what this means in practice. You have an e-commerce site with 500 product pages. You want to visually verify each page after every deployment. That's 500 screenshots per deployment. If you deploy daily, that's 15,000 screenshots per month. If you test on three resolutions (desktop, tablet, mobile), that's 45,000 screenshots per month — far beyond the 5,000 free screenshots, so you're on a usage-based paid plan whose cost grows with your volume.
This model creates an unhealthy tension between test coverage and budget. Teams end up limiting the pages tested, reducing test frequency, or testing on fewer resolutions — not because it's the right technical decision, but because the budget demands it. The tool's pricing model actively degrades test quality.
To put real 2026 numbers on it: Percy's free tier is genuinely generous on the surface — 5,000 snapshots per month with unlimited users. The trap is in how a snapshot is counted: each viewport counts as a separate snapshot. Test 10 pages across 3 resolutions (desktop, tablet, mobile) and that's already 30 snapshots per run. At two runs per day, you exhaust the free tier in 2–3 weeks. Beyond that, Percy applies usage-based paid plans (see the grid on browserstack.com/percy) whose bill climbs quickly as your volume grows — so the "free" tier is, in practice, an evaluation tier rather than something a serious test cadence can live within.
Delta-QA charges nothing. Not per screenshot, not per page, not per month. You test as many pages as you want, as often as you want, on as many resolutions as you want. The rigor of your test coverage is limited by your ambition, not your wallet.
The Desktop edition is free with no snapshot counter, no time limit, and no signup. For teams that need shared workflows, the Team and Business editions are priced at a fixed rate — not indexed to your snapshot volume — so the bill never grows just because you decided to test more. There's no metered counter to watch and no surprise overage at the end of the month: you know exactly what you pay regardless of how thorough your coverage becomes.
Cloud vs local: Your data sovereignty
Every screenshot sent to Percy passes through BrowserStack's servers and is stored there. For most public websites, that's not a problem. But public websites are only a fraction of what teams test.
What about your admin backoffice? Your internal dashboard with customer data? Your staging environment with anonymized production data (or not always perfectly anonymized)? Your medical portal, your banking interface, your HR management system?
Sending screenshots of these pages to a third party's servers, even a SOC 2-certified trusted third party, raises questions that some organizations cannot ignore. Industry regulations (GDPR, HIPAA, PCI-DSS), internal security policies, and common sense around data protection impose limits on what can pass through external servers.
Delta-QA eliminates the question. The tool runs locally. Your pages are analyzed locally. Results stay local. No data leaves your environment. This isn't a premium feature or an enterprise deployment option — it's the tool's base architecture. Percy, by contrast, is exclusively cloud (BrowserStack): there is no on-premise option at all. Delta-QA offers a local Desktop version for individuals and an On-Premise version that runs on your own enterprise servers — so even regulated teams that must keep everything inside their own perimeter have a compliant path.
Result stability: Capricious pixels vs reliable structure
Anyone who has used a screenshot-based visual testing tool knows that false positives are the bane of the approach. Percy has made significant progress with smart comparison algorithms, configurable sensitivity thresholds, and exclusion zones. But the fundamental problem remains: comparing pixels is inherently unstable.
Font rendering varies across operating systems. Anti-aliasing differs depending on the graphics card. Animations captured at slightly different moments produce different images. Dynamic content (dates, counters, ads) changes between captures. Images loaded from a CDN may arrive in a different order. A scrollbar that appears or disappears shifts all elements by a few pixels.
Each of these cases generates a "diff" in Percy that isn't a real bug. And each false positive consumes time: someone has to look at the diff, decide it's not a real issue, and approve it. Multiply by hundreds of screenshots and dozens of deployments, and you get hours of work wasted sorting false positives. There are more exciting things in life — virtually all of them, in fact.
Percy has invested in mitigating this. Since late 2025, a "Visual Review Agent" powered by AI automatically filters roughly 40% of false positives tied to anti-aliasing and font-rendering variations. It's real progress — but the framing matters: filtering 40% of the noise means around 60% of it still lands in front of a human to sort by hand. The agent reduces the symptom; it doesn't remove the underlying cause, because the comparison is still pixel-based.
Delta-QA's engine is immune to these problems by construction. Font rendering may vary, anti-aliasing may differ, text content may update — none of it is something a human eye would flag as a regression, so none of it generates an alert. Because the engine is deterministic and calibrated on human perception rather than running a naive pixel diff, it reports zero false positives across 429 validated test cases. It doesn't filter noise after the fact — it never generates it in the first place.
Tired of cloud screenshots flagging false positives? Catch real visual regressions on your own machine with Delta-QA's free Desktop app, no account and no data leaving your computer. Try Delta-QA for free →
CI/CD integration: Two approaches, same destination
Percy integrates into CI/CD pipelines via its SDK. It's a mature integration that works well with major CI systems (GitHub Actions, GitLab CI, Jenkins, CircleCI). But it adds a network dependency: if Percy's servers are slow or unavailable, your pipeline is impacted.
Delta-QA also integrates into CI/CD pipelines, but without an SDK or external dependency. The tool runs locally on your CI runner. No network latency, no timeouts if a third-party service is down. Your pipeline depends only on your own infrastructure.
When Percy remains the right choice
Percy is a good tool in certain specific contexts.
You want visual testing as a complement to your existing functional tests. If you already have a robust Cypress or Playwright test suite, adding Percy to that suite is the path of least resistance. You add captures at key moments in your tests, and your visual coverage is immediately operational.
You need cross-browser visual testing. Percy can capture screenshots on different browser/resolution combinations via BrowserStack's infrastructure. If verifying visual rendering across Chrome, Firefox, Safari, and Edge simultaneously is a requirement, Percy offers this capability.
Your team is composed exclusively of developers. If the only people who will do visual testing are developers comfortable with SDKs and CI/CD pipelines, Percy's barrier to entry is negligible.
When Delta-QA is the better option
Visual testing is a need in itself, not an extension of your functional tests. If you want visual testing independently from your automated tests — or if you simply don't have automated tests — Delta-QA is self-sufficient.
Non-developers need to participate in visual testing. QA, designers, product owners, marketing — if these profiles need to be able to launch and interpret visual tests without going through a developer, Delta-QA is the only realistic choice.
Budget is limited or you refuse volume-based billing. If you don't want your test coverage dictated by your screenshot budget, Delta-QA's free model eliminates this constraint.
Data confidentiality is non-negotiable. If your pages contain sensitive data or your security policies prohibit sending screenshots to third-party servers, Delta-QA's local approach is the only compliant option.
You're tired of false positives. If you've already experimented with screenshot-based visual testing and sorting false positives made you give up, Delta-QA's perception-calibrated engine will reconcile you with visual testing.
The unfiltered verdict
Percy is a mature tool, well integrated into the BrowserStack ecosystem, and effective for development teams that want to add a visual layer to their automated tests. It's a good tool for its specific use case.
But that use case is narrower than Percy's marketing suggests. The majority of teams don't need an SDK, a cloud service, and per-screenshot billing to detect visual regressions. They need a simple, fast tool accessible to everyone that doesn't punish testing rigor with an escalating bill.
Delta-QA is that tool. Not because it's more sophisticated than Percy — it isn't, and it doesn't claim to be. But because it solves the visual testing problem with less friction, less cost, and more accessibility. And in the real world, the tool that gets adopted is always more useful than the tool that impresses.
Ready to test without counting your screenshots? Run your first comparison with Delta-QA, free and with no sign-up. Try Delta-QA for free →
FAQ
Is Percy free?
Percy offers a free plan limited to a certain number of screenshots per month (the threshold has varied over time). Beyond that, Percy applies usage-based paid plans (see the grid on browserstack.com/percy) whose cost grows with screenshot volume. For serious professional use with significant test coverage, the free plan is insufficient. Delta-QA is free with no volume limitations.
Can you use Percy without writing code?
No. Percy requires integrating an SDK into an automated testing framework. Even the simplest integration solutions (Percy CLI with URLs) require using a terminal and command-line instructions. Delta-QA is entirely no-code — the web interface is sufficient to configure and run visual tests.
Does Delta-QA detect the same bugs as screenshots?
It flags the vast majority of visual regressions — colors, sizes, positions, fonts, margins — that a human eye would notice. It does not chase pure graphical rendering issues (image corruption, webfont problems, browser-specific rendering bugs) that require pixel-level visual comparison. For most teams, these regressions represent over 90% of visual bugs encountered in practice.
Does Percy handle pages with authentication?
Yes, but only within the context of your automated tests. Your test script must handle authentication (login, cookies, tokens), and then Percy captures screenshots of the authenticated pages. With Delta-QA, access to authenticated pages is direct since the tool runs in your local environment where you already have access to your applications.
How much time is saved by eliminating false positives?
Teams using screenshot-based visual testing tools regularly report that sorting false positives represents between 30% and 60% of the time spent on visual testing. Over a typical week with 20 deployments and hundreds of screenshots, that's several hours of human work spent approving differences that aren't bugs. Delta-QA's perception-calibrated engine drastically reduces this ratio.
Can Percy and Delta-QA coexist?
Absolutely. You can use Percy in your automated tests for pixel comparison of critical components, and Delta-QA in parallel for daily coverage of all your pages. The two tools are independent and don't step on each other's toes. It's even a smart combination if you want the best of both approaches.