How to choose a website development partner: 12 questions before signing
A practical checklist for selecting a web studio or developer: compare estimates, validate case studies and team, security, code ownership, and post-launch support.
Why choosing by the proposal total fails
Two proposals called 'website development' may describe different products. One includes discovery, content, responsive states, CMS, analytics, QA, and launch; another covers a few designed screens and implementation only. The lower total proves nothing until you compare boundaries, assumptions, and work left to the client.
Before searching, write a one-page request: business goal, audience, key actions, mandatory integrations, languages, constraints, target date, and decision owner. It is not a complete specification. It gives each candidate the same foundation for a comparable approach.
Freelancer, studio, or product team
The right format follows risk and connectedness. A freelancer can be ideal for local work with a clear deliverable and few dependencies. A studio is useful when design, content, engineering, and launch must behave as one system. A dedicated product team fits an ongoing roadmap with experiments and deep operational integration.
There is no winner by label. Validate the actual people, availability, and accountability. A large agency can allocate a random crew; a small team can bring exceptional expertise. Ask who makes architecture, design, and product decisions and how continuity is protected when someone is unavailable.
12 questions to ask before signing
A useful meeting reduces uncertainty. Ask for an explanation of process and concrete artifacts rather than promises: how the team understands the problem, makes decisions, shows work, and proves release readiness.
- What business problem did you hear, and what should we measure?
- How much discovery is needed before a reliable estimate?
- Who is personally on the team and how much availability do they have?
- Which comparable live product can we open and test?
- What is included in the estimate and explicitly excluded?
- Which client decisions are needed and by when?
- How do prototypes, the design system, and responsive states work?
- How are forms, integrations, accessibility, and performance tested?
- Which security requirements are included in development and acceptance?
- Who owns the domain, analytics, repository, design files, and source code?
- How are scope changes approved and the schedule recalculated?
- What follows launch: warranty, support, monitoring, and knowledge transfer?
How to inspect a portfolio rather than a portfolio presentation
Open the product on a phone and desktop, complete the primary journey, and test the clarity and speed of its forms. Ask which work this exact team delivered: strategy, design, engineering, content, 3D, or one isolated screen. A polished case page without boundaries does not prove repeatable capability.
Look for similarity of complexity, not only sector. A marketplace tests catalog and roles; a brand tests a system across touchpoints; an advertising product tests visual production. In the DUO MESH portfolio, ART KEY demonstrates a digital platform, Dikaya Tish connects brand, packaging, and site, and A Brand in Motion demonstrates a motion system. They help evaluate separate capabilities rather than promise the same outcome for every company.
Normalize different estimates
Move every proposal into one deliverables table. Instead of 'UX/UI — 300 hours,' ask which journeys, states, and viewports are included. Instead of 'development,' identify templates, integrations, roles, CMS functions, tests, and publication environment. List licenses, hosting, paid services, taxes, and support separately.
For unknown functionality, a two-stage model is more honest: paid discovery with a prototype and technical decision, followed by a refined implementation estimate. A premature fixed price either carries a large risk premium or invites later change fees.
| Criterion | Weight | Evidence |
|---|---|---|
| Business understanding | 20% | Goal, audience, metrics, constraints |
| Relevant live work | 15% | Team's role and product quality |
| Estimate transparency | 15% | Stage outputs, exclusions, assumptions |
| Team and process | 15% | Owners, review rhythm, risk control |
| QA and security | 15% | Validation plan and acceptance criteria |
| Ownership and handoff | 10% | Accounts, code, designs, data, documentation |
| Post-launch support | 10% | SLA, warranty, monitoring, roadmap |
Security is evaluated before development
Security cannot remain a line that says 'industry standard.' NIST SSDF recommends integrating secure practices into the lifecycle and creating common language between purchasers and suppliers. OWASP ASVS provides verifiable web-application requirements and explicitly supports procurement and contract use.
A standard website baseline covers secrets and access, dependency updates, backups, form-abuse protection, safe input handling, and a vulnerability-response process. Accounts, payments, and sensitive data need deeper requirements connected to a threat model.
What belongs in the contract and schedules
Acceptance should not depend on whether someone likes the result. Each stage needs an artifact, feedback window, iteration or change mechanism, and testable criteria. Define result ownership, third-party licensing, confidentiality, data processing, and the point at which access is handed over.
The client's organization should usually own the domain, analytics accounts, hosting, and repository while granting managed access to the team. This is not distrust. It prevents the product from becoming hostage to one supplier and makes planned transfer possible.
- scope and measurable output of every stage;
- assumptions, exclusions, and change-request process;
- acceptance, remediation, and launch criteria;
- intellectual-property terms and third-party licenses;
- ownership of domain, accounts, code, designs, and data;
- backups, incidents, warranty, and support;
- termination and the complete handoff package.
Red flags before work begins
Be cautious when a supplier quotes precisely before asking questions, guarantees revenue growth, refuses to show live work, hides the delivery team, or treats source-code transfer as a surprise extra. Starting design without a goal, content, or client decision owner is another warning.
A red flag is not always malicious; it can indicate an immature process. Funding a supplier's learning curve on a business-critical product should be a conscious decision with a reduced pilot scope and a clear stop point.
Run a safe paid pilot
For a large engagement, begin with limited discovery: interviews, a journey map, one-scenario prototype, technical outline, risks, and a refined plan. The output should retain independent value and belong to the client. Both parties can then decide whether to continue and understand the estimate's basis.
A pilot is not a free design contest. A free visual tests willingness to spend sales time but says little about collaboration, engineering quality, or difficult decision-making.
Common questions
How many suppliers should enter a shortlist?
Three to five relevant candidates are usually enough. A larger list creates more shallow meetings. Give everyone the same request and score against a matrix chosen in advance.
Should we request a free test assignment?
For complex work, choose a small paid discovery engagement. It tests research, communication, and decisions. A free mockup rarely represents the future process and can turn selection into a taste contest.
Fixed Price or Time & Material?
Fixed Price fits a well-defined scope and acceptance criteria. Time & Material fits a product with changing priorities. A hybrid is often stronger: fixed discovery and milestones, then transparent delivery under a budget ceiling.
Who should own the source code?
Terms depend on the contract, but commissioned development generally requires the client to receive the agreed rights and repository access. Libraries, fonts, and other third-party assets keep their own licenses, which should be listed in advance.
