You sent the same brief to three companies. One came back at 4,000, one at 15,000, one at 40,000, and all three swear they are quoting the same website. Your instinct is that somebody is either overcharging you or cutting corners, and usually one of those is true, but rarely in the way you expect. The spread is real, it is explainable, and once you know what each price band contains you can put all three onto one sheet and compare like with like. This article gives you that procedure, the questions to send back, and the arithmetic that shows why the cheapest quote is often not the cheapest project.
Three quotes, three different products
The single biggest reason for a 10x spread is that the word “website” covers at least four different purchases. The bidders are not disagreeing about price. They are answering different questions, and nobody told you the questions were different.
Template assembly
Someone buys a ready-made theme, which is a complete design sold to thousands of other buyers, swaps in your logo, your colours and your text, and hands it back. Nobody studies your business. Nobody decides what your homepage should say. The design thinking was done once, years ago, for a generic buyer who is not you. Realistically this is 15 to 40 hours of work, which is why it can be sold for four figures and still be profitable. It is a legitimate product. It is not the same product as the one at the top of your pile.
A builder site, configured for you
Webflow, Framer, Squarespace, Wix or WordPress with a page builder. The layout is arranged for your content rather than inherited wholesale, but it lives inside the platform’s components and constraints. You get a content management system, which is the admin area where your team edits pages without a developer. Call it 60 to 150 hours depending on how much is bespoke. This band is where most small business websites should sit, and there is no shame in it.
A custom design and build
The design is drawn for your content, your customers and your sales process, then built to match that design rather than to whatever a template already did. Every state gets designed: what the form looks like while it is sending, what the search results page shows when there are no results, what happens on a 360 pixel wide phone. Add performance work, accessibility, and a CMS configured around how your marketing team actually publishes. Call it 200 to 400 hours. Our breakdown of what a custom website costs walks through the drivers inside this band.
An agency engagement with strategy and research
On top of the custom build you are buying the work that happens before anyone opens a design tool: customer interviews, a review of your current analytics and search performance, a competitor teardown, a messaging framework, professional copywriting, art direction of new photography, a design system so future pages stay consistent, a measurement plan, and a project manager whose job is to keep all of it moving. Add the agency’s own overhead: offices, account management, a QA person, insurance, and the cost of the pitches they lost. That is how a project reaches five and six figures.
None of these four is a scam. What is dishonest is a bidder allowing you to believe they are quoting the fourth thing while quoting the first.
A quote prices hours and risk, not pages
Most buyers count pages and divide. It feels rigorous and it tells you almost nothing.
Pages are the wrong unit
What costs money is the number of unique page types, not the number of pages. A 40 page site built from five templates (home, service, case study, blog post, contact) is cheaper than a nine page site where every page is a bespoke layout, because a template is designed once, built once and tested once, then filled with different content. Ask every bidder how many unique templates they are pricing and which pages share one. If they cannot answer that, they did not scope your project, they guessed at it.
Risk is a real line item even when it is invisible
Your content is not written yet. Two stakeholders have not seen the brief. The old site runs on a CMS nobody has logged into for three years. Your booking tool has an API that may or may not be documented. Every unknown is a chance that a two week task becomes five, and a bidder who gives you a fixed price is absorbing that chance. Sensible bidders price a contingency into the number. The cheap quote often has none, which means the risk has not disappeared; it will come back to you as a change request, which is the polite name for an invoice you did not budget for.
Where the rate comes from
Price is a rate multiplied by hours, and both halves move. A studio charging 150 an hour can be cheaper than one charging 60 if it takes a third of the time and gets the structure right the first time. It can also be more expensive for exactly the reasons you fear: a layer of account managers between you and the person doing the work. You cannot tell from the number. You can tell from the answer to one question: who, by name, is doing the design and the build, and what else are they on during your project.
The line items missing from the cheap quote
Here is the mechanism that turns a 4,000 quote into a 13,000 project. The cheap proposal is not lying about the 4,000. It is quoting a narrower job, and the rest of the job still exists. It arrives later, in your inbox, as a change request or as work your own team absorbs at night. Below is each commonly missing item and what it is worth, so you can price the gap yourself.
Discovery and structure
Discovery is the phase where the team works out what the site has to do, who it is for, what must be on it and what must go. Information architecture is the output: the sitemap, how content is grouped, and what each thing is called in the navigation. On a 14 page site this is genuinely three to five days of work. Skip it and you get your old sitemap re-skinned, including the four pages nobody has read since 2019 and the navigation label that makes sense only to people who work at your company. Our guide to getting the structure right covers what this phase should produce.
Real copywriting
“Client to supply all copy” is the most expensive sentence in a cheap proposal, for two reasons. First, the money: assume three hours per unique page for someone in your team who is not a writer, or 200 to 400 per page from a freelance copywriter. Second, and worse, the schedule. The single most common cause of a stalled website project is copy that was going to be written last weekend and still has not been. A studio that includes writing has removed your biggest delivery risk, which is worth more than the line item costs. If you do write it yourself, our copywriting guide will save you a round of revisions.
Photography and imagery
Stock photos are cheap and they look it, and on a services or manufacturing site they quietly undermine the claim that you do the thing you say you do. A half day shoot with a local photographer, plus editing, is a real number you should put on the sheet. If a quote includes “imagery” without saying whether that means a licensed stock subscription, a shoot, or you emailing over phone pictures, it is unstated, not included. Sizing, cropping and compressing those images is separate work again, and it is the difference between a fast site and a slow one (our notes on images on the web explain why).
Content population and migration
Somebody has to type it all in. Fourteen pages, thirty product entries, sixty blog posts with their images, their formatting, their internal links and their published dates. At ten to fifteen minutes per blog post that is a solid day and a half before anyone has proofread anything. Ask explicitly whether population is included, whether migration of existing posts is included, and who checks the result. Content migration is boring, it is expensive, and it is the line most often left out.
The omissions that bite at launch
The first group of missing items costs you money. This group can cost you revenue, which is a different and larger problem.
Redirects and protecting your rankings
When a page moves to a new address, a 301 redirect is the instruction that sends both visitors and Google from the old address to the new one. Without a redirect map covering every old URL, every page you currently rank for becomes a dead end, and the search traffic you spent years earning goes with it. This is a few hours of careful work plus testing, and skipping it is the single most damaging omission in a cheap website quote. Ask for the map as a deliverable, not as a promise, and check it against our redesign SEO checklist.
Accessibility
Accessibility means the site works for people using screen readers, keyboard navigation, or simply needing larger text. The common standard is WCAG 2.2 at level AA. In the US and across Europe this is increasingly a legal exposure and not only a courtesy, and it is dramatically cheaper to build in than to retrofit, because it touches colour contrast, heading structure, focus states, form labels and image descriptions, all of which are design decisions. A quote that says nothing about accessibility has not priced it. Our accessibility checklist is short enough to send to a bidder and ask which items they cover.
Browser and device QA
QA means somebody deliberately tries to break the site on real devices: an old iPhone, a mid-range Android, Safari, a 1280 pixel laptop, a widescreen monitor. Budget eight to sixteen hours on a small site. If it is not in the quote, the tester is you, after launch, when your sales director emails to say the contact form looks wrong on his phone.
CMS setup and training
Two quotes can both say “CMS included” and mean opposite things. In one, your marketing person can add a case study, change the homepage headline and publish a job posting. In the other, the text is baked into the build and every change is a support ticket at an hourly rate. Ask which specific parts of each page you can edit, and whether training and a written handover are included. A CMS you cannot operate is a retainer in disguise.
The omissions that show up after launch
Analytics and tracking
Analytics is not the little snippet of code. It is deciding what counts as success, then wiring it up: form submissions, quote requests, phone taps, downloads, the pricing page visit that precedes most enquiries, plus consent handling so the whole thing is lawful. Without it, six months from now you cannot say whether the redesign worked, which means the next budget conversation is an argument about taste. Decide what you will measure before the build starts, because some of it changes how pages are built.
Post-launch support
Every site has a fortnight of small surprises after launch. What you want in writing is a defined window (30 days is normal, 90 is generous), a response time, and a clear line between a bug, which is the site not doing what was agreed, and a new request, which is billable. Without that line the studio and you will each quietly believe the other is being unreasonable.
Hosting, domains and the small recurring bills
Hosting, a CDN, the CMS plan, a form service, a font licence, SSL, the domain itself. None of these is large and together they are a real monthly number that many quotes never mention. Ask what the running cost is, whose accounts they live in, and what happens if you leave.
Fixed price, day rate, or phased: who carries the risk
Every pricing model is a decision about who absorbs the cost of being wrong about scope. There is no free option; there is only a choice about where the risk sits.
Fixed price
You transfer the risk to the studio and pay a premium for it, because they price a contingency into the number. The incentive it creates is honest to acknowledge: once the price is fixed, the fastest route to profit is finishing early, which is fine when scope is genuinely nailed down and dangerous when it is not. Fixed price works when the deliverables list is specific enough that both sides could tell a stranger whether the work is done.
Day rate or time and materials
You carry the risk. The studio bills for the time it takes. This is the right model for work where nobody can honestly predict the effort, such as integrating with a system you inherited. Only accept it with three protections: a not-to-exceed cap, a weekly report of hours burned against the estimate, and the right to stop at any milestone. Without those you have written a blank cheque.
Phased, with a paid discovery
A small paid first phase produces the thing that makes an accurate fixed price possible: the sitemap, the page inventory, the technical constraints, a wireframe set. Then the build is quoted properly. This splits the risk, and it is what we recommend most often, which is exactly why you should be sceptical of it. The test is simple: the discovery must have a standalone deliverable you could hand to a different studio and get a competitive quote from. If it does not, it is a deposit wearing a costume. The same logic applies to phasing the build itself rather than launching everything at once.
How to normalise three proposals onto one sheet
You cannot compare three documents written to three different templates by reading them. You have to force them into a common shape. This takes about ninety minutes and it is the single highest-value thing you will do in the whole buying process.
Write the line item list before you reread anything
Open a spreadsheet and put these rows down the left, in this order, before you look at a single proposal again: discovery, sitemap and information architecture, wireframes, visual design, number of unique templates, responsive design, copywriting, photography and imagery, content population, migration of existing content, redirect map, forms and integrations, accessibility standard, performance targets, browser and device QA, CMS setup, training and documentation, analytics implementation, post-launch support window, hosting and running costs, ownership and handover, timeline, named team.
Mark every row included, excluded, or unstated
Three columns, one per bidder. Included means the proposal names it. Excluded means the proposal says it is not included, which is honest and useful. Unstated is the dangerous value, and it will be the most common one on the cheapest proposal. Do not resolve unstated by assuming. Resolve it by asking.
Price every gap yourself, at your own real cost
For each excluded or unstated row, write what it will cost you to fill it: cash if you will buy it, and your own team’s hours at their loaded hourly cost if you will absorb it. Internal hours are not free, and pretending they are is how a cheap quote wins on a spreadsheet and loses in reality. Total each column. That total, not the number on the front page, is the price of that proposal.
The questions to send every bidder
Send the identical email to all three, with a 48 hour deadline, and tell them you are asking everyone the same thing. How they answer is as informative as what they answer: the good ones will thank you for the clarity, the weak ones will send back adjectives.
- How many unique page templates are you pricing, and which pages share a template?
- Who writes the words? If it is us, how many words per page do you need, and by when?
- Who supplies photography, and is a licence for the images included?
- Who types the content in, and does that include migrating our existing posts?
- Will you deliver a redirect map from every old URL, and who tests it after launch?
- What accessibility standard will the site meet, and how will you demonstrate it?
- Which browsers and devices will you test on, and how many hours are budgeted for QA?
- Which parts of each page can we edit ourselves, and which are locked in the build?
- Is analytics implementation included, and which specific events will be tracked?
- What does post-launch support cover, for how long, and what is your response time?
- Who hosts it, what is the monthly running cost, and can we move it elsewhere?
- Who owns the design files, the code and the accounts, and when does ownership transfer?
- What is your process and hourly rate for a change after design sign-off?
- Who exactly is doing the design and the build, and what else are they on that month?
A worked example: 4,000 against 15,000 against 40,000
Every number here is an assumption, chosen to be realistic and to make the arithmetic visible. Currency does not matter; swap in yours. The point is the method.
The brief and the three proposals
A B2B industrial supplier replacing an eleven year old website. Fourteen pages across six unique layouts, thirty product entries, sixty blog posts to bring across, a contact form and a quote request form, and a CMS so the marketing lead can publish without help. Assume the marketing lead’s loaded cost to the business is 45 an hour, and that any specialist contractor you hire afterwards charges 90 an hour.
Proposal A, 4,000. Design and development of a 14 page responsive website, SEO friendly, CMS included, two rounds of revisions, four weeks, client to supply all text and images, 100 percent payable upfront.
Proposal B, 15,000. Discovery session, sitemap and wireframes, design of six templates, build with a CMS, migration of the sixty posts and thirty products, redirect map, cross-browser QA, a training session, thirty days of post-launch support. Explicitly excludes copywriting, photography, analytics implementation and hosting. Nine weeks.
Proposal C, 40,000. Everything in B plus eight customer interviews, an audit of current analytics and search performance, a messaging framework, full copywriting, art direction of a photo shoot with production billed at cost, a small design system, WCAG 2.2 AA conformance with an external audit, a measurement plan, ninety days of support and a named project manager. Fourteen weeks.
Pricing the gaps in the 4,000 proposal
Copywriting: nine of the fourteen pages need original writing, three hours each at 45 an hour is 1,215. Thirty product descriptions at twenty minutes each is ten hours, another 450. Photography: a half day shoot plus editing, 900. Content population and migration: sixty posts at twelve minutes, thirty products at eight minutes, fourteen pages at forty-five minutes, which is 26.5 hours at 45, so 1,193. Redirect map for 104 URLs: five hours at 90, so 450. Accessibility retrofit after the fact: 1,500. Device QA: twelve hours at 90, so 1,080. CMS training and written handover: six hours at 90, so 540. Analytics implementation: five hours at 90, so 450. A month of post-launch fixes: six hours at 90, so 540. Hosting, CMS plan and form service for year one at 45 a month: 540.
Those gaps total 8,858. Add the 4,000 sticker and Proposal A is a 12,858 project. Of that, 2,858 is your own team’s time rather than cash, so you will write cheques for about 10,000 and absorb the rest in evenings and weekends that were supposed to be spent elsewhere.
The other two, normalised
Proposal B is missing copywriting (1,665 on the same assumptions), photography (900), analytics (450) and hosting (540), so 3,555 of gaps. Normalised, B is 18,555.
Proposal C is missing only the shoot production it manages but bills at cost, say 1,800, plus hosting at 540. Normalised, C is 42,340.
What the sheet actually says
The sticker prices were 1 to 3.75 to 10. Normalised, they are 12,858, 18,555 and 42,340, which is 1 to 1.44 to 3.29. The cheap quote was not a tenth of the expensive one, it was closer to a third, and the gap between the two realistic options is 5,700, not 11,000. That changes the conversation completely, because 5,700 buys discovery, structure, migration, redirects, QA and training, and you were going to pay for those anyway, just later and worse.
The threshold test for the expensive one
You still have to decide whether C is worth 23,785 more than B. Do not argue about it; calculate what it would have to earn. Assume your profit on a won contract, after you deliver the work, is 9,000, you close one in four qualified enquiries, the site currently produces thirty qualified enquiries a year, and you will keep this site for three years. C has to produce 23,785 divided by 9,000, so 2.64 extra contracts. That needs 2.64 divided by 0.25, so about 10.6 extra qualified enquiries across three years, which is 3.5 a year, an 11.7 percent lift on thirty.
Now the question is answerable. If your current site is a decade old, your messaging has never been tested on a customer, and buyers genuinely research you online before calling, a 12 percent lift is a reasonable bet and C is defensible. If your enquiries come from a sales team, a trade show and referrals, and the website is a formality people check after they have already decided, then no research programme is going to move that number and B is plainly the right purchase. This is the calculation we run with clients even when it argues against the bigger project.
Why the lowest bid is often the most expensive outcome
Run the failure case. Proposal A ships, and eighteen months later it is slow, the rankings never recovered from the missing redirects, and the marketing lead cannot edit anything. You then buy Proposal B properly. Total spend: 12,858 plus 18,555, so 31,413, for one working website, plus eighteen months of enquiries you did not get. Buying B at the start costs 18,555. That is the whole argument, and it is the same arithmetic as the real cost of a cheap website: the cheap option is only cheap if it does not have to be replaced.
Red flags in how the work is described
No discovery, straight to design
A proposal that jumps to visual design with no phase in between is telling you it will re-skin whatever you already have. That is a valid product at the template price. It is a serious problem at any price above that.
Deliverables described as adjectives
“Modern, clean, user-friendly design.” “Fully optimised.” “Engaging user experience.” None of these is a deliverable, because neither side can ever tell whether it has been delivered. Every line in a proposal should be a noun you could point at: a sitemap, six page designs, a redirect map, a training session, a Figma file. Adjectives in a scope are how disputes start.
Unlimited revisions
This sounds generous and it is the opposite. Nobody can afford unlimited work, so one of two things is true: the price already includes a fat cushion you are paying for, or the studio will start slowing down the moment your requests exceed what they budgeted, and you will feel it as unreturned emails. A specific number of rounds at a defined stage, plus an hourly rate for anything beyond, is more honest and usually cheaper.
“SEO included”
Ask what it means. If the answer is proper page titles, clean URLs, a sensible heading structure, fast loading, structured data and a redirect map, that is real and it should be in the base price of any professional build. If the answer is vague, or if it involves promises about rankings, treat the line as worth nothing. Nobody can promise a position in a search result, and anybody who does is either inexperienced or counting on you not checking.
Red flags in the commercial terms
No named team and no timeline
A proposal without dates is not a plan. You want a start date, the milestone where you give feedback, the design sign-off, the build, the review window, and the launch. You also want the names of the people doing the work. A studio that will not name them is either subcontracting without saying so or does not know yet, and both are things you would rather learn now.
Ownership left unstated
Get this in writing before you sign: you own the design files, you own the code, the domain is in your account, the hosting and analytics accounts are in your name with you as owner, and it all transfers on final payment. Where a page builder or theme licence cannot be transferred, the proposal should say so plainly. Proprietary platforms are not automatically bad, but you should choose one knowingly rather than discover in year three that leaving means rebuilding.
100 percent upfront
For a small project a deposit of 40 or 50 percent is normal and fair, because the studio is committing calendar time it cannot resell. The full amount before any work exists leaves you with no leverage and no recourse, and any studio confident in its process will not need it. The mirror image is also a red flag: a bidder who asks for nothing upfront on a large project is either desperate or has no cash discipline, and neither is a good partner for a three month engagement.
Payment schedules that are fair to both sides
What a fair schedule looks like
Tie money to milestones that both sides can verify. On a project of this size: 30 percent to start and hold the dates, 30 percent at design sign-off, 30 percent at the point the build is on a staging site for your review, 10 percent at launch. Nobody is ever more than one phase out of pocket, and each payment attaches to something you can look at.
What to push back on, in both directions
Push back on a final payment that is 5 percent, because it gives the studio nothing to come back for. Push back equally on holding 40 percent until launch, because launch dates slip for reasons that are frequently yours: the legal review, the product photos, the director on holiday. A common and fair compromise is that the final invoice becomes due at launch or thirty days after the site is ready for launch, whichever comes first. It protects the studio from your delays and protects you from theirs.
What to negotiate, and what to leave alone
Negotiate scope, never craft
The right way to reduce a quote is to remove work. Cut two page templates. Launch with the blog migrated but the case study section coming in phase two. Use stock photography now and shoot in the spring. Every one of those is a real, honest reduction and a good studio will help you find them. The wrong way is to ask for the same scope for less, because the hours do not vanish, they get taken out of the parts you cannot see.
The three things never to cut
QA, testing and content. Cut QA and you ship a site that is broken on a device you do not own. Cut content work and you launch with placeholder text, which is the most common reason a finished site sits unlaunched for two months. Cut testing of forms and redirects and you can lose both your enquiries and your search traffic on the same afternoon, silently, for weeks. These three are the cheapest insurance in the entire project and they are always the first things a nervous bidder offers to drop.
What you can genuinely cut
Custom animation. A bespoke icon set. A second language at launch. Anything on a page that gets 2 percent of your traffic. The blog design, if you publish twice a year. A design system, if you have no plans to add pages. Ask each bidder directly: if my budget were 20 percent lower, what would you remove and what would that cost me later? The quality of that answer tells you more about a studio than their portfolio does, and it is one of the questions we suggest in our guide to choosing an agency.
When the cheap quote is genuinely the right purchase
This article is written by people who sell the expensive option, so here is the part that costs us sales. Sometimes the small template site is the correct, disciplined buy, and a studio that tries to talk you out of it is selling, not advising.
Three situations where the small site wins
You are testing whether anyone wants the thing at all. Spend as little as possible until you have evidence, and put the money into finding customers instead. You need a presence, not a channel: your work comes from referrals, a trade body listing, or a sales team, and the website exists so people can confirm you are real. Or you are pre-revenue and the choice is between a modest site now and no site for a year, in which case ship the modest one.
How to buy cheap well
Buy cheap deliberately, with the omissions priced. Insist on three things even at the bottom of the market: you own the domain and the accounts, the redirect map exists if you are replacing an existing site, and the content is yours to export. Then set a review date twelve months out. The mistake is never buying the cheap site. The mistake is buying it as a permanent solution for something your revenue depends on, budgeting 4,000 for it, and being surprised in month seven.
The decision procedure to run tonight
Print the three proposals and highlight every deliverable that is a noun. Open a spreadsheet and write the line item list from the section above down the left, with one column per bidder. Mark each cell included, excluded or unstated. Send the same fourteen questions to all three bidders with a 48 hour deadline. When the answers come back, fill in the unstated cells, then price every gap at your own real cost, counting your team’s hours at their loaded rate. Total each column, and write the normalised number at the top of each proposal in pen.
Now check the non-negotiables: is there a redirect map, is QA in the scope, is content work priced by somebody, and is ownership stated. Any proposal missing two or more of those comes off the table regardless of price. For the two that remain, run the threshold test: divide the gap between them by your profit per won contract, then by your close rate, and ask honestly whether the better proposal can plausibly produce that many more enquiries over three years. If the normalised numbers land within about 10 percent of each other, stop comparing money and choose on the people: who asked the sharpest questions about your business, who told you something you did not want to hear, and who you would rather be on a call with in week eight when something breaks.
Then hold the winner to their own words. Every answer they sent you goes into the contract as scope. A studio that meant what it wrote will not blink.
If you are sitting on two or three proposals right now and cannot tell which one is honest, send them to us at hello@beconfidency.agency and we will tell you what we would ask each bidder, even when the answer is that the cheapest one is fine and you do not need us.
That discipline is built into how we quote our web design service.