I’ve been looking back through the Lovable projects we’ve built at Made Grand, partly because I wanted a better answer to a question businesses have quite reasonably been asking me: how much effort will it take to build something of their own?
The Lovable credit system is easy enough to understand, but working out how many credits you’ll need, or how much of your own time you’ll spend getting the result you want, is more challenging.
Our projects range from simple websites to grander projects, like Eat With Friends, our newly released group dining app. We’ve also built tools that we use daily within Made Grand, including a project management workspace and a time tracking and reporting app. Working across these different projects gives us a better understanding of the effort involved and what can be achieved when discussing a new idea with a business.
I wanted to share some of my own experience, including the estimated credit figures. They need some explanation, though, because the numbers on their own leave out quite a lot of the work.
Every build is different
When somebody says they want to build a website, that could mean a few pages explaining their services. It could also mean a website where staff can manage content, customers can sign in and information moves between several business systems.
These are some of the types of project we’ve been working on:
| Type of project | Project name | What the work involves |
|---|---|---|
| Service website | KR Counselling | Helping visitors understand the service and feel comfortable making contact |
| Website with its own management tools | Mindful Moments Kids Yoga | Giving families information about classes while allowing the owner to manage activities through a private administration area |
| Website connected to an existing CMS | Made Grand and Rachael Blackmore Counselling | Building a new frontend while keeping WordPress for content management |
| Customer-facing application | Eat With Friends | Connecting meal organisers with guest replies through shared invitations |
| AI-connected business application | Project Command Centre | Organising private project information and making it available to supported AI tools through an MCP connection |
KR Counselling is a good example of a focused website where the design and wording deserved proper attention, helping visitors feel comfortable about making contact.
Mindful Moments Kids Yoga had an established social media brand to work from, which helped us produce the initial working version in a morning. It also needed tools for managing classes and events, so there is more behind the website than parents see when browsing. The morning refers to that first working version; content development and subsequent improvements continue beyond it.
With Eat With Friends, the host can create a meal and share an invitation, then guests reply without having to create accounts. Keeping that part of the experience straightforward meant working on a well-defined customer journey and clear UX, as well as how the underlying records relate to each other and who is allowed to see or change them. It is live and currently in beta, which gives us the opportunity to develop it around real use.
Project Command Centre serves an entirely different purpose as it’s for internal use only. We use it to keep track of projects, business concepts and ideas so we can decide where to focus our attention. Its AI connections allow supported tools to work with those records and projects, for instance by viewing sign-ups and waiting lists in the live apps. That introduces questions about secure access and what an AI tool should be allowed to update, alongside the work of building the application itself.
At the more involved end, applications may combine extensive database work with AI agents and workflows that use connected services to carry out tasks. The scope then includes what happens when a service fails or an agent takes an unexpected route. Those behaviours need designing and testing as carefully as the screens people use.
A look at the credits
| Project | Recorded credits | What the figure covers |
|---|---|---|
| Mindful Moments Kids Yoga | 72.19 | Reported all-time usage |
| Eat With Friends | 203.59 | Reported all-time usage |
| Rachael Blackmore Counselling | 212.79 | Reported all-time usage |
| Made Grand | Approximately 311 | Build usage shown in the project dashboard |
| Project Command Centre | 333.44 | Reported all-time usage |
The all-time figures come from our initial usage report generated in Lovable and still need checking against the same reporting filters. Those totals may include running costs as well as development. The Made Grand figure comes specifically from the Build category in the project dashboard. Treat this as a provisional snapshot of the records available, rather than a comparison of build-only costs.
Even with a consistent set of figures, there would still be more to explain. These projects have evolved after their first versions, so the totals include work undertaken at different stages. They also reflect our approach as experienced developers, including changes made outside Lovable in a local development environment.
You’ll notice that the recorded totals for Rachael Blackmore Counselling and Eat With Friends are fairly close, despite being different types of projects. Before drawing conclusions, we’d need to separate the work behind those totals and check the reporting. A more complicated product doesn’t give us a reliable multiplier to apply to another business’s idea.
For budgeting, I’d begin with the features needed for the first release and review usage as those are implemented. As the project develops, that gives you evidence from your own build to work with.
Why counting prompts only tells us so much
The original reason for reviewing this data was to understand how many prompts different projects had taken. Some of the available counts appear to be incomplete due to the way Lovable presented the data, so I’m holding back on publishing those until we can establish what they cover.
There is also a practical problem with using prompt count as a measure of effort. A request to alter a heading and a request to implement an invitation system each count as a prompt, even though the second can involve complex changes and multiple actions by the AI agent throughout an application.
Lovable’s Build mode usage depends on the complexity of the request and the work completed. The dashboard separates Build usage from Run usage, which includes Cloud, AI gateway and connector activity. You can inspect an individual response’s credit cost through its response menu.
Credits spent developing a feature are separate from the usage involved when people run that feature in the published app. Storage, compute, network activity and connected AI can continue to use credits after launch, depending on how the application is hosted and used.
Lovable’s credit guidance explains the current reporting in more detail.
A prompt count would still be interesting alongside a clear account of what was built. I just wouldn’t use it to promise that another business could achieve the same result in the same number of requests.
Getting more out of the credits you use
The work you do before opening Lovable has a considerable bearing on what you ask it to build. For a service website, that might mean getting the content into shape and deciding what visitors should do after reading it. For an application, it could mean understanding the process well enough to explain what should happen.
Eat With Friends gave us a useful way to approach this. The wider idea had room for much more functionality, but we focused the beta around organising a meal and receiving replies. That gave us something we could put in front of people without needing every future feature in place.
I’d take the same approach with an internal business tool. Choose a piece of work that would make a difference to the person using it and get that working properly before expanding it. It is easier to judge whether a tool is useful when its first job is clear.
As you build, give Lovable enough information to understand the outcome you need. Asking it to add a form leaves quite a few decisions open. Explaining where the submission should go, what information is required and what the visitor should see if it fails gives you a much better basis for checking the result.
When something goes wrong, it is worth stopping development to clearly understand what failed. Repeatedly asking Lovable to fix the same problem can use more credits without resolving the underlying cause. Being able to reproduce the problem and provide the actual error is often more useful than another broad request to try again.
The same applies to testing. Follow the journey yourself, including the awkward parts. For example, an invitation working for the person who created it doesn’t necessarily mean it works for somebody opening it on their phone. Check each feature works, no matter how small, and make sure any changes you make to the project don’t break it. You can add ‘Don’t change any other functionality’ to your prompts to make the scope clear, but that instruction does not guarantee other features will remain unaffected. Then test, test, test.
We also make changes locally, working directly with the code on our own computers. That includes design refinements and copy updates, alongside development work and local build checks. Those changes don’t use Lovable credits, although they still require our time.
For us, working this way keeps us close to the project. We can understand how the website is put together and make a direct change when that is the appropriate way to handle it. It also means that a credit total will always be an incomplete account of what went into one of our builds.
Do you have the time to build it yourself?
This is the question I’d encourage a business owner to ask themselves before jumping straight into building their own project.
Lovable gives you the ability to get involved in development very easily, but you still need somebody who can take responsibility for the result and spend time building and checking it. That time has to come from somewhere, particularly if the person building it is also running the business.
A website might receive updates over several months while being live throughout that period. Equally, a first version produced in a morning may still need further work before it covers everything the business has in mind.
If you’re interested in learning and have room to experiment, building something yourself can be worthwhile. If the project needs to be ready for a particular date, or will handle important business information, consider how you will get help when you reach something you cannot confidently assess.
Where Made Grand can help
Made Grand is a Certified Lovable Partner and we can work with your business at different points in the process. We can plan and build the project for you. If you have already made some progress, we can step in to finish a difficult feature or review what you have built before launch.
We also offer guided training using your own project, so you can develop your understanding while working on something relevant to your business. That can be a useful option if you want to remain involved but would benefit from help making the technical decisions.
If you’re considering Lovable, or have a build that needs attention, tell us what you’re working on. You don’t need a finished brief to start the conversation.
If you're planning an MVP or early-stage product and want to make sure the foundations are right before you build, we're happy to talk it through.
Start a conversation