Most buyers approach this backwards. They collect three quotes, compare the bottom line, and pick the one in the middle. Searching for a website company Singapore businesses actually recommend is not a procurement exercise you can win on price comparison, because the three quotes in front of you are almost never quoting the same job. One includes copywriting. One assumes you supply it. One has quietly scoped four templated page layouts where you assumed fourteen unique ones. The numbers look comparable. The deliverables are not.
This post is about the decision process itself: how to build a shortlist, what to score, which questions separate a capable partner from a confident salesperson, and when to walk away. It is deliberately not a list of features. Features are easy to claim. Process is harder to fake.
Start by writing down what the site has to do
Before you speak to anyone, write a single page that answers three questions. What business outcome does this site need to produce. Who has to be able to use it after launch. What has to integrate with it.
That page is your control document. Every vendor conversation gets measured against it. Without it, you end up evaluating vendors on the things they choose to talk about, which is always the things they are good at.
The outcome question matters most and gets skipped most often. "A modern site" is not an outcome. "Enquiry forms from qualified B2B prospects, at a rate we can measure against our current eleven a month" is an outcome. "Reduce the time our sales team spends answering the same six pre sales questions" is an outcome. In our experience the buyers who arrive with a written outcome get better proposals, because good agencies can then propose against it and mediocre ones visibly cannot.
The second question, who uses it after launch, is the one that determines your total cost over three years. If your marketing coordinator needs to publish a case study without raising a support ticket, that is a build decision, not a training decision. It changes how the page builder is configured, how content types are structured, and how much of the site is hard coded. Say it out loud at the briefing stage or you will pay for it later.
The third question, integrations, is where scope tends to explode. A CRM handoff, a booking system, a payment gateway, a shipping rate calculator, a customer portal, a multilingual switch: each of these is a project inside a project. Vendors price very differently depending on whether they have built that integration before. Listing them up front lets you ask who has, rather than discovering in week six that nobody had.
Build the long list properly
A long list of three is not a long list. Aim for six to eight before you cut.
Sources that actually produce good candidates, roughly in order of usefulness:
Referrals from businesses structurally similar to yours. Not similar in industry, similar in shape. A 40 person B2B services firm and a 40 person B2B manufacturer have far more in common as web clients than two retailers of wildly different sizes. Ask specifically what went wrong during the project, not whether they were happy. Everyone is happy the week after launch.
Portfolio archaeology. Find sites you admire in adjacent industries and look at who built them. Footer credits still exist on a surprising number of Singapore sites, and where they do not, the agency usually features the work in its own portfolio. This finds you vendors whose aesthetic and technical range you have already verified.
Specialist directories and industry bodies, treated with caution. Listings are frequently paid. Use them to generate names, then verify independently.
Cold search, treated with more caution. The firms ranking for competitive commercial terms are demonstrably good at their own marketing. That is genuinely informative: an agency that cannot rank its own site is an odd choice for a project where organic visibility matters. It is not, however, evidence of build quality. Treat the ranking as a signal about their SEO capability and nothing else.
What should not be on your long list: anyone who cannot show you a live site they built in the last eighteen months, and anyone whose portfolio is entirely screenshots rather than URLs. Screenshots are the oldest tell in this industry. If the live URL is not supplied, assume the site no longer exists in that form or was never launched.
The scoring framework
Score each shortlisted vendor across six dimensions. Weight them for your own situation, but do not collapse them into a single gut feel at the end. The point of scoring is to make you notice when you are about to hire someone because you liked them.
| Dimension | What you are actually testing | Weak signal | Strong signal |
|---|---|---|---|
| Relevant evidence | Have they solved a problem shaped like yours | Portfolio of unrelated sites, no live URLs | Two or three live sites with comparable complexity and a named contact you may call |
| Discovery depth | Do they understand before they design | Quote issued within 24 hours of first email | Paid or structured discovery, written requirements doc before pricing is fixed |
| Technical ownership | Who owns the stack after launch | Proprietary CMS, no code export, hosting locked to them | Standard CMS, code in a repository you own, documented handover |
| Content realism | Have they planned for the hardest input, your words | Copy listed as client supplied with no deadline | Content plan with owners, dates and a fallback if you slip |
| Post launch model | What happens in month four | Support quoted only as ad hoc hourly | Defined maintenance scope, response times, update cadence |
| Commercial clarity | Can you predict the final invoice | Vague inclusions, undefined revision rounds | Itemised scope, named revision limits, written change request process |
Two notes on using this. First, do the scoring immediately after each meeting, before the next one. Recency bias in vendor selection is enormous and almost entirely unconscious. Second, if a vendor scores well on five dimensions and catastrophically on one, that is usually a worse outcome than a consistent middle scorer, because the catastrophic dimension is the one that will consume your project.
The questions that actually separate vendors
Most buyer question lists are too polite. These are the ones we would ask.
"Show me a project that went wrong and what you changed afterwards." The answer tells you whether the firm has a retrospective culture or a sales culture. A vendor who says nothing has ever gone wrong is either inexperienced or managing you. What you want is specificity: a scope that was mis estimated, a stakeholder who was not in the room, a content deadline that slipped by two months, and the concrete process change that followed.
"Who exactly will be working on this, and what else are they on?" Pitch teams and delivery teams diverge constantly. Ask for names and rough allocation. Then ask what happens if that person leaves mid project. The honest answer involves documentation and overlap, not reassurance.
"Walk me through your last handover." You are testing whether handover is a real phase or an afterthought. Good answers involve a documented admin walkthrough, a recorded training session, access credentials transferred to accounts you own, and a defined support window. Weak answers involve sending you a login.
"What would you cut from this scope if my budget were 30 per cent lower?" This is the single most revealing question we know. A vendor who understands value will tell you immediately which elements drive the outcome and which are nice to have. A vendor who says everything is essential either has not thought about your business or is protecting margin.
"How do you handle content that arrives late?" Content is the number one cause of delay on website projects, and every experienced firm knows it. The answer you want includes staged deadlines, placeholder strategy, and a written consequence: what moves in the timeline when content slips. If your own team is likely to be the bottleneck, look at whether the vendor offers copywriting or has a content partner, and price it in rather than pretending you will write it.
"What does the site look like on the tenth of never, when nobody has touched it for a year?" Slightly facetious phrasing, serious question. You are asking about plugin dependency, framework upgrade paths, security patching and website maintenance. A site built on twenty third party plugins is a site with twenty future failure points.
What a portfolio cannot tell you
Here is the contrarian part, and it is the section buyers most often resist.
A beautiful portfolio is weak evidence of anything you care about. It proves the agency employed a good designer at some point, on some project, possibly with an unusually cooperative client and a large budget. It tells you nothing about whether the site converted, whether it shipped on time, whether the client could update it afterwards, or whether the design survived contact with real content.
A common mistake we see is buyers falling in love with a hero image. The hero image is the cheapest part of the build. What costs money and determines whether the project succeeds is the unglamorous middle: information architecture, form logic, template flexibility, load performance, the fifteenth product page, the FAQ accordion nobody photographs for a case study.
So invert the portfolio review. Instead of asking to see their best work, ask to see a site they built two years ago that the client has been maintaining themselves ever since. Open it. Look at the pages the client added after launch. Do those pages look like the launch pages, or do they look like a different website wearing the same logo. That gap is the single best available proxy for whether the build was designed for its owner or for the agency's own showreel.
The same logic applies to awards. Design awards are judged on design. They are not judged on whether the enquiry rate went up. We have audited award winning Singapore sites that converted worse than the template driven site they replaced, usually because someone put a full bleed video where the value proposition should have been. Good UI and UX design is not the same thing as decorative design, and buyers conflate them constantly.
None of this means aesthetics do not matter. They matter enormously for credibility, particularly for professional services firms where the website is the only evidence a prospect has before the first call. The point is narrower: aesthetics are necessary and not sufficient, and they are the one dimension every vendor optimises their sales material for. Which means they carry almost no discriminating information at the shortlist stage.
What a shortlist looked like for one Singapore retailer
One Singapore precision engineering firm we worked with came to us with a shortlist of three vendors and a deadlock. Roughly SGD 45,000 separated the cheapest and most expensive quotes, and the internal team had split.
We ran their own scoring exercise with them. Three things surfaced within an afternoon.
The cheapest quote, at around SGD 22,000, had scoped eight page templates. The firm had 63 product pages across four product families with different specification tables. Eight templates meant either heavy compromise on the product pages, which were the entire commercial point of the site, or a change request in month two. The quote was not cheap, it was incomplete.
The most expensive quote, at roughly SGD 67,000, included a bespoke product configurator nobody in the business had asked for. It had appeared because a junior stakeholder mentioned a competitor had one. When we asked the sales team how many enquiries would actually start with a configurator, the answer was close to none: their buyers send a drawing and ask for a quotation. That single line item was around SGD 18,000 of the difference.
The middle quote had the weakest visual work and the strongest technical answer. Their proposed structure treated product specifications as structured content rather than hand built pages, which meant the engineering team could add a new variant in ten minutes without touching layout.
The firm went with the middle vendor and redirected part of the saved budget into product photography and a proper specification content audit. Eleven months after launch, the site had grown from 63 to 91 product pages, all added in house, none of them requiring agency time. That is the outcome the scoring framework was protecting: not the prettiest launch, the lowest friction second year. If your catalogue is the commercial core of the site, the same logic applies to ecommerce website design decisions, and it is worth reading our breakdown of what separates the best ecommerce website design companies in Singapore before you scope.
Red flags that should end a conversation
Some signals are not scoring dimensions. They are exits.
A quote issued before anyone has asked what your business does. This happens more than buyers expect, particularly on inbound enquiries. A number produced in under a day was produced from a template, which means it is priced for an average project, which means it is wrong for yours in one direction or the other.
Ownership ambiguity. If you cannot get a straight written answer to who owns the domain, the hosting account, the code and the design files after final payment, stop. This is the most expensive thing to discover late. Every legitimate firm answers this in one sentence.
Pressure on the timeline for pricing reasons. Discounts that expire are a sales technique, not a commercial reality. The cost of building your website does not fall because you signed on a Friday.
No written change process. Scope will change. It always changes. What matters is whether there is an agreed mechanism: a change request form, a rate, an approval step, a timeline impact statement. Without it, every change becomes a negotiation and the relationship degrades.
Refusal to talk about performance or search implications. A rebuild is the highest risk moment in a site's organic life. Redirect mapping, URL structure decisions, template level metadata and page speed are all build time decisions. A firm that says search is someone else's problem is telling you they will hand you a site that needs remedial on page work before it recovers. Ask what their redirect and migration checklist looks like, and if you do not have a baseline, run an SEO audit before the build starts, not after.
What a fair evaluation timeline looks like
Buyers frequently compress this and then wonder why the decision felt rushed.
A realistic run, from writing your one page brief to signing, is six to eight weeks. Week one to two, brief and long list. Week three, first conversations, typically 45 minutes each, purely qualifying. Week four to five, detailed briefings with the three vendors still standing, ideally with your actual delivery stakeholders in the room rather than only the decision maker. Week six, proposals in. Week seven, clarification round and reference calls. Week eight, decision and contract.
The step most often skipped is the reference call, and it is the highest value hour in the entire process. Ask the reference three things: what the agency was like when something went wrong, whether the project finished on the original timeline, and whether they would hire them again for a different type of project. That third question is the useful one, because the answer is rarely a flat yes and the qualifications are where the truth lives.
Field Notes
Across the website projects we reviewed for Singapore clients this year, the pattern that came up most was scope drift on content rather than design. In roughly two out of three of the delayed builds we looked at, the critical path was a client side content deliverable, not an agency side development task. Average slip attributable to content alone was in the region of five weeks.
Second observation: of the proposals we helped clients compare, the spread between lowest and highest quote for what was the same brief was typically between 2.5 and 3 times. That spread is not mostly margin. It is mostly unstated assumptions about template count, revision rounds, content supply and post launch support. Once we normalised those four variables, the spread usually compressed to something closer to 1.4 times.
Third, and the number we quote most often internally: the median time we see between launch and the first significant client side content update is about seven months. Sites that cannot be updated easily do not get updated slowly. They get abandoned.
Making the final call
When you reach the end with two credible vendors, and you usually will, the tie breaker is not price and it is not portfolio. It is this: which team asked you the better questions.
Over a twelve to sixteen week build you will spend more time in conversation with this firm than with most of your own suppliers. The team that probed your assumptions during the sales process will probe them during delivery, and that is precisely what you are paying for. The team that agreed with everything you said will agree with everything you say during the build too, including the things you are wrong about.
If you want a second opinion on a shortlist you have already built, or you would rather start from a written brief than from a quote, that is a conversation worth having early. You can see how we approach builds on our website design and web design in Singapore pages, get a sense of how we work on our about page, and see the wider service set on our services overview. If visual assets are likely to be your bottleneck, our production team handles that side. When you are ready to talk it through, get in touch.
Natasha Tan is the founder of Digital Marketing Singapore, a full-service SEO and digital marketing agency based in Singapore. With hands-on experience across SEO, paid media, and content strategy, she works directly with Singapore businesses to build organic visibility and generate consistent leads. Natasha specialises in the Singapore market — including local search behaviour, PDPA compliance, and government grant navigation for SMEs.
