# Choosing a rating engine for a new MGA > A decision guide for new MGAs choosing a rating engine: what capacity providers expect, speed to market, broker and aggregator distribution, audit trail and the cost model that suits a book with no premium yet. By Ptally Pricing Team, 4 November 2026. Category: Distribution. Source: https://ptally.ai/blog/rating-engine-for-a-new-mga When choosing a rating engine for a new MGA, five things matter most: whether it gives your capacity provider the control and evidence they ask for, how quickly it can get your first product quoting, whether it can serve the brokers and aggregators you plan to sell through, whether every quote has an audit trail, and whether its cost model fits a business with no premium yet. Feature lists matter less than these five. A new MGA needs an engine that is quick to launch, easy to change and cheap until the business is writing. This is a decision guide. For the steps between a rating guide and a first live quote, see [launching an MGA](https://ptally.ai/blog/launching-an-mga-pricing). For a full feature checklist, see our [buyer's checklist](https://ptally.ai/blog/insurance-pricing-platform-buyers-checklist). ## 1. Your capacity provider's requirements come first Your insurer, syndicate or reinsurer is the first audience for your pricing. Before you look at any engine, find out what they need. Ask them directly: - Do they set the rates, or approve rates you propose? - What changes can you make within the binding authority without their sign-off? - What evidence do they want before launch and after each rate change? - What reporting do they expect on quotes, declines and referrals? Then test each engine against the answers. A good fit lets you show the partner the model itself, the test results for each change and a record of who approved what. If the engine stores the model in a form only its developers can read, every review becomes a meeting rather than a document. On Ptally a model is a human readable JSON file of ordered steps, and every change runs against its regression test suite before it can be approved. Your partner can read the rules, see the tests pass and see the version history. ## 2. Speed to market A new MGA has a launch window. Capacity is agreed, underwriters are hired and distribution partners are waiting. Every month without a live quote is a month of cost with no premium. Ask each vendor for the time from a signed-off rating guide to the first live quote, and ask who does the work. The answer often hides a separate implementation project. Questions that expose the real timeline: - Can the engine build from what you have, whether a rating guide, an Excel rater or a US rate filing? - Who builds the first model, and is that included? - How long does a typical rate change take once you are live? Rate changes matter as much as launch. Early trading nearly always shows that some rates need adjusting. If a change takes weeks, you trade on the wrong rates for weeks. ## 3. Broker and aggregator distribution Decide where you will sell before you choose an engine, because channels set technical requirements. - **Broker platforms and trading partners** usually call a pricing API. The engine needs a clear, documented quote endpoint. - **Your own website** needs an embedded quote form. - **Price comparison websites** send high volumes and expect fast answers at peak times. - **AI assistants and chat** are a growing channel and need an engine that AI clients can talk to. The safest choice is one model serving every channel. If each channel has its own copy of the rates, prices drift apart and every rate change has to be made several times. Ptally serves real time APIs, embedded forms and chat and voice agents from the same model, and is live on price comparison websites at peak aggregator traffic. The quote call is documented on our [API page](https://ptally.ai/developer/api). ## 4. Audit trail Your capacity provider, your compliance team and in time your regulator will all ask the same question: why did this customer get this price? The engine should answer it without anyone rebuilding the calculation by hand. Check that: - every quote records the exact model version that priced it; - every version records who changed it, who approved it and when; - you can replay a quote and see every input, lookup, formula and rule; - declines and referrals record which rule fired. On Ptally each quote response includes the version that priced it and the referral and exclusion rules that fired. Adding `?debug=true` returns every step the engine evaluated. Versions move through draft, approval and publishing, so nothing prices live quotes without sign-off. ## 5. A cost model that fits a new book Established rating engines are often sold with an annual licence and a separate implementation fee. For a carrier with a large book that can make sense. For a new MGA it means paying before the first policy is written, with money that could fund underwriting and distribution. Look for a cost model that starts when premium starts. Our article on [what a rating engine costs](https://ptally.ai/blog/how-much-does-an-insurance-rating-engine-cost) compares the options in detail. Ptally is free to build, with no implementation fee. The Performance plan charges up to 1% of gross written premium, so the cost grows with the book. The Per model plan is a flat fee per live model for MGAs that prefer a fixed budget. Enterprise adds a self host option, SSO, audit and approvals, and custom SLAs for multi product MGAs. ## Questions to put to every vendor 1. What do we pay before the first live quote? 2. How long from a signed-off rating guide to a live quote? 3. Can our capacity provider read the model and see the tests? 4. Does every quote record the version that priced it? 5. Can one model serve our API, website and aggregator channels? 6. How long does a rate change take, and what does it cost? ## Where Ptally fits Ptally is an AI enabled insurance pricing platform covering any line in the UK, Europe and the US. It has served over 50 million quotes. Models are built from your rating guide, spreadsheet or filing, tested on every change and published with approvals and full version history. AI clients can connect over MCP to read, validate and test models. Pricing experts can build your first model for you as part of the no implementation fee offer. See the plans on our [pricing section](https://ptally.ai/#pricing), explore a working model on the [developer page](https://ptally.ai/developer), or read the [API reference](https://ptally.ai/developer/api).