Who this case is for
Foundations, non-profits, educational bodies, and any organisation that is writing a TOR (terms of reference) or opening a tender for a website.
This case explains, before we won the tender, what we analysed, what we proposed, and how we plan to deliver the project. If you are about to open a tender, you will see what a good proposal should answer beyond the items in the contract.
We do not name the organisation, the project, the prototype site, or the contract value.
1. The brief in the TOR
The Thailand office of an international conservation foundation opened a tender for the Thai-language website of an online volunteer project, building on a prototype site made by the organisation’s office in another country.
The website has 2 goals
- Educate young people, teachers and the general public about the wildlife trade online
- Invite people to sign up as volunteers to help report illegal wildlife sale listings
The TOR asked for the language, content, case examples and illustrations to be adapted to the Thai context, and specified WordPress.
While reading the TOR alongside the prototype site, we set ourselves one criterion: the project only succeeds when Thai people actually read the content and actually sign up, not just when there is a website that ticks every item in the contract. The whole proposal was written from that angle.
2. What we analysed before submitting the proposal
The prototype site looks good and has strong content, but is very slow on mobile
We measured the prototype site with Google Lighthouse, simulating a phone on a mid-speed connection, which is close to what most people actually use.
Picture a secondary school student who taps a link a friend shared in a group chat and gets a frozen white screen for nearly ten seconds. Most will go back before it loads. The great content the team worked hard on never reaches its readers.
| Finding | Cause | Impact on the Thai version |
|---|---|---|
| Slow pages that do not respond to taps | Several layers of Page Builder and plugins stacked on top of each other | Building it the same way would carry over the same speed limits |
| 6.1 MB page weight | 4 autoplay videos and large PNG images not yet converted to WebP | Hard to access for people with weak signal or limited data |
| Sign-up button leads to an external form | A third-party form service | No way to track how many people are interested or where they drop off; applicant data is not in the organisation’s own system |
| All content on a single page | Designed as a one-page site | Google has no separate page for each topic, so the site relies on social sharing alone |
| Metadata and Sitemap | Already done well, Lighthouse SEO 100 | Keep the existing standard and build on it |
The organisation is highly trusted, but nobody can find the project through search yet
| Domain | Domain Rating | Referring Domains | Keywords ranking in Thailand |
|---|---|---|---|
| The project’s prototype site | 67 | 12 | 0 |
| Organisation site, Thailand office | 52 | 809 | 0 |
| Organisation site, prototype office | 67 | 1,847 | 2 |
The right-hand column shows that no Thai search term yet brings anyone to the project. People who already care about this issue cannot find the project on their own unless they happen to see someone share it.
But another column is much better news. A Domain Rating of 52–67 is trust the organisation has built up over a decade or more. A newly founded organisation would need years to reach that point. The asset is already there; it just is not being used yet.
What words Thai people use to search about wildlife
What people searching “สัตว์ป่าคุ้มครอง” (protected wildlife) read today
The difficulty scores in the chart are only estimates, so we opened the live search results for this keyword, which is searched 1,700 times a month.
| Position | What holds it | Notes |
|---|---|---|
| Top | Google’s AI Overview | Summarises the answer; many people read only this and leave |
| 2 | A small DR 22 site | Holds the top organic result and gets around 739 visits a month |
| 3 | A Thai-language online encyclopedia | A general article, not designed to teach or prompt action |
| 5, 9 | Social media posts from government agencies | Not web pages with a clear content structure |
| 6, 7, 10 | Government websites | Almost nobody links to them, 0–2 Referring Domains |
What we saw: most people searching this term end up on social posts and government notices, which are accurate but hard to read, and do not tell the reader what to do next. Meanwhile the page getting the most visits is a small site with a DR of just 22. If an organisation with a DR of 52 builds a strong content page on this topic, it has a very good chance of becoming the main reference for this term, without buying ads.
And if the organisation wants its accurate information to be what AI picks up in its answers, the content has to be structured so machines can understand it from the moment it is built, not patched later.
3. What we proposed
Keep the look, change how it is built behind the scenes
The TOR specified WordPress, and we agree with that for content management, because the organisation’s team needs to edit content themselves. But the prototype site is slow because WordPress does two jobs at once: it stores the content and assembles each page live every time someone visits. So we proposed two options.
| Option | Details | Speed target |
|---|---|---|
| A · Full WordPress, following the TOR to the letter | A licensed theme with custom code, no stacked Page Builder layers, page-level caching, images converted to WebP, unnecessary JavaScript removed | Performance 85 or higher |
| B · Astro + Headless WordPress (our recommended option) | The team still writes and edits content in the WordPress dashboard as before, but the pages people see are pre-built as static files, with almost no JavaScript running on the user’s device | Performance 95 or higher · LCP under 2 seconds |
In plain terms, option B is like a book that has already been printed and is sitting on the shelf. Anyone who walks in can pick it up and read it straight away. The old way waits for someone to walk into the shop and then typesets the whole book again. For the organisation’s team, the screens they work in stay exactly the same.
Split the content into separate pages
We mapped out the whole structure of the prototype site: the sign-up button, the statistics bar, the how-to-join section, the reporting timeline, the clickable cards on trafficked animals, and the section covering media reports.
We then proposed moving the content with educational value onto pages with their own URLs. Each topic gets a chance to rank for the terms Thai people actually search, while the continuous-scroll experience on the homepage stays in place.
Features that can be measured, not just delivered
| Feature | What we deliver |
|---|---|
| Educational quiz | A quiz the team can add to and edit themselves, with answers and explanations, plus a button to share results on social media, which spreads it naturally |
| Interactive learning modules | Animal cards that open in-depth information, a reporting timeline, and an exercise in spotting suspicious listings |
| Downloadable learning materials | A categorised, searchable file library that counts downloads per file, so the organisation knows which materials are actually used |
| Volunteer sign-up form on the organisation’s own site | Spam protection without an intrusive CAPTCHA, email notifications, export to a spreadsheet file, and data kept in a system the organisation manages itself |
| Measurement | GA4 via Google Tag Manager, tracking sign-up button clicks, successful form submissions, completed quizzes, material downloads and reading depth |
Adding work the TOR did not ask for, because it is cheapest to do during the build
The TOR did not ask for SEO or AI Search work. We included it in the same proposal, because this work takes only a few days during the build, but doing it later means going back and reworking almost the entire page structure.
- Technical SEO: Sitemap, canonical, Thai-English hreflang, readable URLs, and passing all three Core Web Vitals
- On-Page: page titles, meta descriptions and headings on every page based on real search terms, plus an internal linking plan
- Structured Data: Organization, WebSite, Article, BreadcrumbList, FAQPage and Quiz
- AI Search: an llms.txt file, access for the major AI bots, and content organised as questions and answers that AI can easily cite
- Accessibility: following WCAG 2.2 level AA, because the audience includes schools and colleges
We also extended post-launch support to 60 days, against the TOR’s minimum of 30 days, and delivered a content plan for the first 6 months so the team can keep writing on their own. The reason is that projects like this run on donations and grant funding, so what we deliver should keep working after the contract ends, not stop on the day the site goes live.
4. Now that we have won, how we plan to deliver the project
Our proposal was selected. The project plan in the proposal takes 12 weeks.
| Phase | Work | What the organisation will see |
|---|---|---|
| Week 1 | Kick-off meeting | Scope summary, image files, content and brand guidelines received, choice of option A or B confirmed |
| Weeks 2–3 | UX/UI design | Screen designs for every page, on both mobile and desktop |
| Weeks 3–4 | System architecture | CMS, page structure and a URL map with target keywords for each page |
| Weeks 4–8 | Website development | A staging site where progress can be viewed at any time |
| Weeks 6–9 | Content upload | Thai and English content, quizzes and downloadable materials ready to use |
| Weeks 9–10 | Testing and UAT | A test report with evidence, and a 5-working-day trial period for the team |
| Week 11 | Go-live | Sitemap submitted to Google and user data collection begins |
| Week 12 | Handover | A Thai-language user manual with screenshots for every step, the full source code, and the 60-day support period begins |
We hold to 3 principles throughout the project
- Measure success by readers and sign-ups, not just by delivering every item on the list. That is why measurement must be ready from launch day.
- The organisation’s team must be able to carry on alone. The manual is written as “to change this, go here, then click this”, and all source code belongs to the organisation.
- Fast on mobile everywhere, because the main readers are young people opening links from chats.
5. Expected results
This table shows the criteria we proposed must be met before handover. These are targets, not results that have already happened. The website is still in development.
| Pre-handover criterion | Target |
|---|---|
| Lighthouse mobile Performance | 90 or higher (option B) or 85 or higher (option A) · SEO 100 |
| Core Web Vitals | Pass all three |
| Accessibility | No critical errors in axe testing; every function usable by keyboard |
| Links and forms | No broken links; forms and quizzes work correctly on every path |
| Cross-device testing | Chrome, Safari, Firefox and Edge on desktop, iOS and Android |
The long-term outcome we want to see is Thai people searching about wildlife finding the project on their own, instead of only hearing about it through shares. We will update this case once the site is live and we have real numbers.
Checklist for organisations writing a website TOR
Before opening the tender
- Set measurable goals, such as the number of sign-ups or quiz completions, not just a list of pages
- Ask bidders to measure the speed of the current or prototype site and state their targets as numbers
- Specify that applicant data must stay in a system the organisation manages itself
- Require a Thai-language user manual and full ownership of all source code
When reading proposals
- Has the bidder looked at what words people use to search for this topic?
- Are SEO and measurement included in the build, not added later?
- Are there measurable acceptance criteria, such as Lighthouse scores and accessibility test results?
- Is the post-launch support period long enough for the team to become confident using the site?
Questions non-profits often ask us
Can a WordPress site be fast? Yes, if you avoid stacking page builders and plugins in layers. For the fastest result, use WordPress only as the content store and serve the pages people see as static files. The team still edits content in the WordPress dashboard as before.
What is wrong with a one-page website? It looks good and tells a story well, but Google has no separate page for each topic to rank, so people searching for that topic cannot find it. The best approach is to keep the continuous-scroll homepage and move the educational content onto its own pages.
Our organisation is already well known. Do we still need SEO? Yes. An organisation’s reputation is trust that Google gives a lot of weight to, but without pages that answer what ordinary people search for, that trust goes unused. In this case the domains have a DR of 52–67, yet no Thai search term brought anyone to the project.
Why think about AI Search now? Because many informational searches already show an AI Overview summarising the answer at the top. If you want your organisation’s accurate information to be what AI uses in its answers, the content must be structured for machines to understand from the day it is built.
We build websites for organisations like this
If your organisation is about to build a website or write a TOR, start with a free website audit. We will measure your current site first and tell you plainly where it is worth investing. Or see our website development service.