Website speed optimization is priced badly across the whole market because "make my site faster" describes anything from a 90-minute image cleanup to a six-week rebuild. Two quotes ten times apart can both be honest — they are quoting different jobs.
This is what the bands actually mean in 2026, based on quotes we see clients bring us and the work we deliver ourselves.
The 2026 price ranges
| Provider | Typical price | Turnaround | Best for |
|---|---|---|---|
| Marketplace gig (Fiverr / Upwork, low end) | $50–$150 | 1–3 days | Basic image and cache work only |
| Experienced freelancer | $300–$900 | 3–7 days | Most small business sites |
| Specialist studio (flat fee) | $800–$2,500 | 1–2 weeks | E-commerce, paid-traffic sites |
| Full-service agency | $3,000–$12,000 | 3–6 weeks | Complex or custom platforms |
| Ongoing retainer | $250–$1,500/mo | Continuous | Sites that ship changes weekly |
| Enterprise engagement | $15,000+ | 2–4 months | Multi-region, multi-team platforms |
For reference, Softshelf works in the flat-fee specialist band: a scoped speed sprint from $450, a full performance and conversion pass from $1,200, with the number agreed before anything starts.
What each tier includes
Under $150 — the mechanical tier
Bulk image compression, a caching plugin, minification toggled on, and a screenshot of a higher Lighthouse score. This is real work and it does help, but it is the work you could do yourself in a weekend with the checklist from our nine-causes guide. What it does not include is diagnosis. Nobody is asking why the server takes 1.4 seconds to respond.
Watch for: aggressive optimisation plugins configured once and never tested against your checkout or contact form. The commonest repair job we get is a store whose "Add to cart" broke after a $99 speed gig.
$300–$900 — the diagnostic tier
Someone actually reads the waterfall. You should get: a measured baseline, image and asset work, script deferral, third-party audit, caching and CDN configured properly, font loading fixed, layout shift resolved, and a before/after report using the same tool on the same URLs.
This is the right band for the large majority of small business and small e-commerce sites, and it is where the cost-per-second-saved is lowest.
$800–$2,500 — the revenue tier
Everything above, plus work inside the theme or application: template-level changes, critical CSS, removing framework weight, and — the part that separates this band — conversion work alongside speed. Faster pages that still bury the call to action do not make you money.
Expect a staging environment, a rollback plan, and someone who will tell you when the honest answer is "your theme is the ceiling."
$3,000+ — the rebuild tier
At this point you are usually not buying optimisation, you are buying a new front end. Legitimate when the platform genuinely cannot get there — a page builder generating 400 KB of CSS per page, or a 2016 custom build with no asset pipeline. Ask for the reasoning in writing. "It needs a rebuild" is sometimes true and sometimes just a bigger invoice.
Hourly vs flat fee vs retainer
| Model | You pay for | Risk sits with | Good when |
|---|---|---|---|
| Hourly ($40–$150/hr) | Time spent | You | Scope is genuinely unknown |
| Flat fee | A defined outcome | The provider | Scope can be defined — most cases |
| Retainer | Ongoing capacity | Shared | You ship changes weekly |
| Performance-based | A result | The provider | Rare, and attribution is hard |
Hourly billing quietly puts every risk on the client. If the developer is slow, inexperienced, or spends three hours reading your theme, you pay for it. Worse, it creates an incentive structure nobody wants to think about: efficiency reduces the invoice.
Flat fee inverts that. The provider carries the estimate risk, which is the right place for it — they know the work better than you do. It also means you can approve the spend without a range and a shrug.
Flat fee only works with a scoped deliverable. "Make the site fast" is not scopable. "LCP under 2.5s on these five URLs, mobile, measured in PSI" is. Insist on the second.
See your own numbers in 30 seconds
The free Softshelf audit runs your URL through Google's own PageSpeed data and returns your performance, SEO and accessibility scores plus the specific elements slowing the page down. No signup, no sales call.
Working out what it is worth to you
Speed work is one of the few marketing expenses you can size before you spend. Run these three numbers:
1. Your current revenue per month from the site. Sessions × conversion rate × average order value.
2. The same with conversion up 10%. A conservative figure for a site going from 5 s to under 2.5 s on mobile. Deloitte's retail research found a 0.1 s improvement associated with an 8.4% conversion lift; Google's field data puts bounce probability up 32% between 1 s and 3 s.
3. Divide the fee by the monthly difference. That is your payback period in months.
12,000 sessions/month, 1.8% conversion, $85 average order = $18,360/month. At 1.98% conversion (a 10% lift) it is $20,196 — a difference of $1,836/month. A $1,200 fixed fee pays back in about three weeks, then compounds.
If you buy traffic, the case is stronger still: you are already paying for every visitor who bounces on a blank screen. Speed work is the only optimisation that improves both your organic and your paid economics at once.
Why the $95 quote is expensive
Three ways cheap speed work costs more than it saves:
- Something breaks and nobody tells you. Aggressive JS minification breaking a checkout step is invisible on a Lighthouse score and obvious in your revenue — two weeks later.
- The score improves and the experience does not. Lab scores can be gamed. Deferring everything gets a 95 and an interface that does nothing for four seconds. Ask for field data, not a screenshot.
- You pay twice. The commonest job we quote is undoing a previous optimisation attempt. Repair work costs more than the original job because the first step is untangling what was done.
Seven questions before you pay anyone
- What is the target, in numbers? A specific metric on specific URLs, mobile, or the engagement has no definition of done.
- Will you show me before/after field data, not just a Lighthouse screenshot?
- Do you work on a staging copy? The only acceptable answer is yes.
- What is your rollback plan if conversion drops?
- Are you removing weight or adding a caching layer on top of it? The first is optimisation; the second is a plugin.
- What happens if you miss the target? A serious provider has an answer ready.
- Who owns the changes afterwards? You should have the code, the config and the documentation.
Any provider who cannot answer all seven in a short reply is quoting a hope, not a job.
Want this done for you this week?
Softshelf fixes performance and conversion problems on a flat fee — you approve a fixed number before anything starts, and you get a before/after report when it ships. Most speed sprints are delivered in 3–5 working days.
Frequently asked questions
How much does it cost to speed up a website?
In 2026, expect $50 to $150 for basic image and caching work, $300 to $900 for a properly diagnosed freelance engagement, $800 to $2,500 for specialist flat-fee work including conversion improvements, and $3,000 or more for agency-level or rebuild work. Most small business sites get their best return in the $300 to $900 band.
Is website speed optimization a one-time cost?
The core work is one-off, but sites regress. Every new app, plugin, tracking pixel or unoptimised image added afterwards costs some of it back. Sites that ship changes weekly benefit from a light retainer or a quarterly re-audit; sites that rarely change usually do not need one.
Should I pay hourly or a flat fee for speed work?
Flat fee, in almost every case. It puts the estimate risk on the provider who understands the work, and it lets you approve a known number. Hourly only makes sense when the scope is genuinely unknowable, which is rare for performance work after an audit.
How do I know if speed optimization will pay for itself?
Multiply monthly sessions by conversion rate by average order value, then recalculate with the conversion rate 10% higher. Divide the quoted fee by that monthly difference to get a payback period in months. Under three months is a straightforward yes.