Case Study
The words are BondPro’s. The numbers are in code.
A landing page that was already live, a rebrand in flight, and one thing that could not be bought at a price worth paying: a bond calculator that felt right.
Marnus van Staden
Product & Process
TL;DR
BondPro had a working WordPress landing page and a rebrand in progress. What they could not get was the calculator: every bond calculator on offer was either expensive or did not feel like something a homebuyer would trust. The build put the words and the pages into a content system BondPro edit themselves, and put the three calculators behind an API where the lookup tables and the arithmetic sit in code and can be tested. It was billed by milestone, the ordinary way. It has been live since March with no support requests. This was the first piece of work ReadyRun did, and it is the smallest study on this site.
The Starting Point
BondPro is a financial services provider working in South Africa’s property market. They help homebuyers through bond applications, affordability, and the transfer process, which is most people’s largest financial decision handled at the least confident moment of it.
There was already a site. A WordPress landing page, live and doing what a landing page does, with a brand refresh in progress that it did not yet carry. So this did not start with anything broken.
It started with the calculators. A homebuyer works out affordability before they are ready to speak to anyone, and what they are really doing is deciding whether to be hopeful. Every bond calculator BondPro could put on the page was either priced as though it were the whole product, or worked but felt wrong in a way that is easy to notice and hard to write down. Strauss Smit, BondPro’s director, was not willing to put either in front of someone at that moment.
The engagement was billed by milestone. Scoped, delivered, invoiced, the ordinary way. The other arrangements on the pricing page came later and suit larger builds; this one did not need any of them.
Where the line got drawn
“You will be able to change it yourself” is the most common promise in software, and it usually arrives attached to configuration screens. By the time those screens have been bent into doing what the business actually asked for, they are code wearing a different hat: conditions, branches and rules that nobody on the client side can safely touch, and that the developer had to fight to produce. The promise gets kept on paper and broken in the room.
So on this build the line went somewhere plainer. Words on one side, arithmetic on the other.
In BondPro’s hands
- Page content and service copy. Wagtail, unmodified, sitting behind the site rather than serving it directly. BondPro edit blocks of content, not settings that pretend to be content. When the market moves and the messaging has to move with it, that is an afternoon rather than a ticket.
- New pages, when they want one. Assembled from the same blocks, so a new page cannot arrive off-brand or half-styled.
In code, deliberately
- The three calculators. Affordability, Repayment, and Bond & Transfer Cost. The page sends what the visitor typed to an API; the server holds the lookup tables, does the arithmetic, and sends back the figure.
- The tables behind them. Transfer duty and bond costs are set by legislation and revised on somebody else’s schedule. Nobody should be typing those into a form field at half past four. In code they can be tested, and a wrong one is a failing test rather than a number a stranger relied on.
- The front end itself. A Svelte renderer, written for this, which is where the refreshed brand actually landed and where the calculators got the feel that none of the bought ones had.
Rented, because running it would be worse
- Brevo, for transactional email. A form on the site has to reach a person. Deliverability is somebody’s full-time problem and it should stay theirs.
- PostHog, for analytics. What people do on the site, rather than how many of them arrived.
Three tiers, and the rule behind them is short: open source where it makes sense, a subscription where one is genuinely better than anything worth running, and something built only for the part that is the business’s own. Here that last part was the calculators, which is why they exist at all.
The editable half and the coded half are not a concession either way. It is the same question asked twice: does changing this thing call for judgement about words, or does it call for being right? BondPro get the first one, on their own, whenever they like. The second one stays where it can be checked.
“They built the BondPro website exactly the way we wanted it, including our custom bond and transfer cost calculators. Friendly team, quick to assist, and very easy to work with throughout the whole process.”
Strauss Smit
Director at BondPro
What was specified and not built
The original specification went further than the site. It carried full application workflows: tracking a buyer through the process, building up an application as the documents and the decisions arrived, and giving the BondPro team a view of where each one stood.
None of that got built. It was parked for the first release, because it was not needed to go live, and because designing an application workflow before anybody had used the calculators would have been guessing at how people arrive. The structure it would attach to is there. Whether it gets built is a conversation for when BondPro want it.
“I really appreciated how you made me feel that you genuinely dedicated time and effort to my project. It was clear that you went the extra mile, and that level of commitment really stood out. It gave me confidence that my project was in good hands and that you cared about delivering quality results.”
Strauss Smit
Director at BondPro
Where it stands
- Live since March, with no support requests. BondPro run the content themselves and have not needed to ask for anything. For a handover, that is the result. There is nothing to report here because there has been nothing to report.
- A calculator session can become an enquiry. Pre-Approval and Apply Now sit where someone has just worked out a number and is deciding what to do about it, which is the only moment that question gets a genuine answer.
- Running on a plain server. The front end and the API on one bare GoFast server in South Africa, close to the people using it. No vendor between BondPro and their own site.
Recognise the situation?
Every business has a version of the BondPro story: a site that mostly works, one thing on it that nothing off the shelf does well enough, and a brand that has moved on without it. If that sounds familiar, we would like to hear from you.