Why published prices are scarce
Performance work is hard to price from the outside. Two apps with the same number of pages can differ tenfold in complexity: one is a server-rendered catalogue with a few widgets, the other a client-rendered dashboard with real-time data and dozens of interactions. Providers therefore tend to quote after a scoping call, and public price lists are rare.
That makes comparison difficult for buyers. We publish starting prices on our own pricing page, and this guide tries to explain the market more broadly, while being clear that we sell these services ourselves.
What published prices tend to look like
The price lists that do exist online are mostly agency marketing pages, and they are hard to compare: some quote in other currencies, some bundle performance work with redesigns, and few say how many routes or interactions a price covers. We could not find an independent survey of Canadian rates for this kind of work, so we do not quote a market average. The useful pattern is qualitative: a findings-only audit is the smallest item on almost every list, implementation is priced well above it, and anything described as a rebuild sits at the top. When you see a figure, check what scope it describes before comparing it with anything else.
What drives the cost
Scope: how many routes and interactions
The single largest factor. A review of one landing page is a small job; a review of a signed-in app with search, filters, editors and dashboards is several times larger. Good quotes name the routes or route families and the interactions included.
Depth: score report or root causes
An automated report that lists Lighthouse findings takes little time and adds little beyond what free tools show. Tracing interactions on throttled devices, attributing long tasks to code, and writing framework-specific fixes takes skilled time. The price difference reflects that.
Access and environment
Work with source code, source maps and a staging environment goes faster and goes deeper than work from production bundles alone. Apps behind single sign-on or with complex test data take longer to set up.
Field data
If you already collect real-user metrics with attribution, the search is narrower. If not, a provider may need to set up collection first and wait for data, which adds time and cost but can pay for itself by focusing the work on what users hit.
Implementation
Findings and implementation are different jobs. Some providers quote only for findings; others include hands-on engineering or pull request reviews. Implementation is commonly priced by time or as a fixed block, because its size depends on what the findings reveal.
Retesting
A retest after changes ship confirms the effect. It is cheap compared with the original work and worth including.
Common pricing models
| Model | How it works | Suits | Watch for |
|---|---|---|---|
| Fixed-price audit or review | One price for a defined scope and deliverables | First engagements, clear scope | Vague deliverables, page-count padding |
| Time and materials | Hourly or daily rate, billed as used | Open-ended implementation | No cap or estimate |
| Fixed time block | A set number of hours at an agreed rate | Implementation after an audit | Unused hours, unclear priorities |
| Retainer | Monthly fee for ongoing monitoring and help | Teams shipping often | Paying for idle months |
How to compare quotes
Ask each provider the same questions:
- Which routes and interactions are included? A number of pages alone is not enough.
- Which devices and network profiles will be tested? Testing only on a fast laptop misses most problems.
- Will field data be used, and how? At least the Chrome UX Report should be checked. Core Web Vitals are assessed at the 75th percentile of page loads (source 1), and a provider should know that.
- What exactly will we receive? Traces, a ranked list of causes with evidence, framework-specific recommendations, a walkthrough.
- Is implementation included, and how is it priced?
- Is there a retest?
- What does the provider promise? Be wary of guaranteed scores or rankings. Nobody controls how your users' devices behave, and Google's ranking systems are not something a consultant can promise outcomes for.
Red flags in a proposal
- A fixed promise of a score. Lab scores vary between runs and devices; field values depend on your users. A promise of "95 or your money back" says more about sales than about method.
- No mention of field data. A proposal that only discusses lab scores is likely to miss interaction problems.
- Page counts instead of routes. Fifty pages built from the same template are one route family, not fifty units of work.
- Deliverables that are only automated reports. Free tools already produce these.
- Access requests by email. Asking for passwords in plain email is a security warning sign.
- Pressure to buy implementation before findings. It is reasonable to want to see the findings before committing to further work.
What a fair quote looks like
A fair quote names the routes and interactions in scope, the devices and network profiles, the use of field data, the deliverables, the timeline, the price and what happens if scope changes. It separates findings from implementation and states whether a retest is included. If any of these are missing, ask before you sign.
Hiring versus contracting
For a one-off problem, an outside review costs less than a hire and brings specialist experience. For an app that ships weekly and needs ongoing attention, building the skill in-house, with outside help for training or difficult problems, often makes more sense. A middle path is a review that includes a handover session so your engineers can repeat the method.
Estimating the value of the work
Price only makes sense next to value. For a commercial app, the value of performance work shows up in metrics your team already tracks: conversion on key flows, task completion in a signed-in tool, support tickets about slowness, or search visibility for public pages. We do not publish conversion figures from other companies, because results from one business rarely transfer to another. A better approach is to pick one flow that matters, measure its responsiveness and its business outcome today, make one change, and compare. That gives you your own number, which is more persuasive inside your company than any industry figure.
Taxes and invoicing in Canada
Consulting fees for Canadian clients generally attract GST or HST, depending on the province, and Quebec clients may also be charged QST. Check whether quotes include or exclude sales tax, and ask your accountant about the rules that apply to your business. This guide is not tax advice.
Questions to ask yourself first
Before asking for quotes, write down three things: the flows or routes that matter most to your business, what you already know from field data, and what decision the work should support. A team that knows it needs to decide between two rendering strategies needs a different engagement from one that needs to find why a checkout button responds late. Clear goals make quotes comparable and keep the work focused on what you will act on.
Our own prices
Our starting prices in Canadian dollars are on the pricing page, with what each service includes. We give a fixed written quote after a scoping conversation, and no deposit is requested before you accept it.
Sources
- Interaction to Next Paint (INP), web.dev Primary source (Google)