Comparing website proposals is hard because they rarely speak the same language. One charges per page, another per hour, a third sends a PDF with three sentences and a number. And you have to decide with real money.
This article covers what a serious proposal contains, seven warning signs that show up constantly, and the questions that separate a partner from a liability.
What a serious proposal contains
It does not need to be thirty pages. It needs to be verifiable. Seven blocks are enough:
- Understanding of the problem. A paragraph proving the vendor listened: what you sell, to whom, and what the site has to trigger. If the proposal could be sent to any company, nobody listened.
- Itemised scope. How many unique design screens, which pages, which features, how many languages, how many revision rounds.
- What is excluded. As important as what is included. Photography, logo design, copy, ad management, translation, migration of old content. Writing this down prevents the month-two argument.
- A dated schedule. Stages, deliverables, and what depends on you. If you take ten days to approve a layout, the deadline moves ten days.
- Technology and where it runs. Which platform, which hosting, who administers it.
- Commercial terms. Payment schedule, recurring fees, post-launch warranty, hourly rate for out-of-scope work.
- Ownership. A clause stating that domain, code, content and accounts are yours once payment clears. More on that in who actually owns your website.
Without those blocks you are not comparing prices, you are comparing promises. Reference ranges for judging whether a number makes sense are in how much a professional website costs.
Red flag 1: a price with no scope
"Complete business website: R$ 4,900." That is the whole proposal.
The problem is not the number, it is the absence of definition. Complete means how many pages? Custom design or template? Who writes? Does the blog count? How many revisions? With no written scope, any future disagreement becomes one word against another, and the person who paid loses.
What to do: ask for the same proposal with an itemised scope. If the price then moves a lot, you have discovered why it was low.
Red flag 2: "unlimited pages"
Nobody delivers unlimited work for a fixed fee. The promise only holds one way: every page reuses the same layout with different text. In that case unlimited is true and meaningless, because duplicating a template costs almost nothing.
What costs money is the unique screen: a new layout, designed and built. An honest proposal counts unique screens, not URLs.
What to do: ask how many distinct layouts are included and what an additional one costs.
Red flag 3: no defined delivery date
"We deliver fast" is not a deadline. "Around six weeks" is not one either, because it has no start date and no milestones.
A real schedule has stages: discovery, architecture, screen design, approval, development, review, launch. And it says what blocks each stage. Without that, the project drags for months and you have nothing to point at.
What to do: ask for dated milestones and a clause covering delays on both sides.
Red flag 4: proprietary code with no export clause
Some companies build on their own closed platform. That can be legitimate, as long as the proposal answers one question: if I leave, what do I take?
The bad version: you pay monthly fees for years, decide to switch vendors, and discover there is no way to export the site. You leave with the domain and the copy you wrote yourself. You start over, and search rankings go with it.
What to do: require in writing the right to export content in an open format and, where custom code exists, access to the repository or the files. Even on a closed platform, content export is negotiable.
Red flag 5: mandatory hosting with the agency
Hosting with the team that built the site is often good: they know the environment and fix incidents faster. The problem is the word mandatory.
If the contract says the site only runs on the agency infrastructure and no exit path exists, the monthly fee can rise with no alternative available to you. That stops being a service and becomes dependency.
What to do: accept the hosting, but require a portability clause: on termination the agency delivers files, database and migration instructions within a defined window. To understand which environment you need, see shared hosting, VPS and cloud.
Red flag 6: guaranteed first place on Google
Nobody controls Google rankings, not even people who have done this for twenty years. What can be guaranteed is technical work: an indexable site, correct architecture, speed within the Core Web Vitals thresholds, structured data, and content published regularly.
There is a confusing exception: it is easy to "guarantee first place" for the company name itself, because there is almost no competition for it. That brings no new customers.
What to do: ask for the promise to be rewritten as a verifiable deliverable. Instead of "first place", something like "site indexed, sitemap submitted, structured data validated, monthly Search Console report". The technical SEO checklist is a good reference for what to demand.
Red flag 7: no written contract
A proposal accepted over chat with no contract is common, and it almost always fails at the moment something goes off script: someone gets sick, scope grows, a payment slips, work stops.
A simple two or three page contract solves it: scope as an annex, deadlines, amounts, intellectual property, confidentiality, termination terms, and what happens to work already produced.
What to do: if the vendor avoids a contract, that is the one warning sign on this list that on its own justifies ending the conversation.
Eight questions to ask before signing
Send them by email and keep the answers. Email is documentation.
- How many unique design screens are included, and what does an extra one cost?
- Who writes the copy, and how many revision rounds are in the price?
- Which technology will you use, and who else on the market can maintain it?
- Can I publish content myself through the admin panel without calling you?
- Will the domain and the Analytics and Search Console accounts be in my company name?
- What is the post-launch warranty and what does it cover?
- Are backups automated, and has a restore actually been tested? Covered in backups and continuity.
- What is the monthly cost after delivery, and what does it include? Compare it against the expected maintenance routine.
A good vendor answers all eight without hedging, because they thought about them before you asked. A vendor who bristles at question five or question seven is telling you something useful.
One last test: look at what they already shipped
A portfolio is the cheapest test available. Open two or three sites the company built, on a phone, on mobile data. See whether they load fast, whether the menu works, whether the form sends, whether the copy says anything. Then search those client names on Google and see whether the site shows up.
That five-minute test is worth more than any slide deck.
ALB Seven writes proposals with itemised scope, dated milestones and full ownership in your name. See real projects in the portfolio and get in touch for a proposal that survives being compared.