How to choose a Shariah compliance API: seven questions to ask

Choosing a Shariah compliance API comes down to what the provider returns beside the verdict, not how many securities it covers. A halal investing product is only as defensible as its data layer: the first time a user asks why your app says one thing and a competitor says the opposite, you need the name of the standard, the ratio it was applied to, and the filing the figures came from. An API that returns a bare pass or fail leaves you with nothing to answer that with.

The seven questions below are worth asking any provider, including us. What matters in each case is less the answer itself than what it tells you about how the data was built.

Why coverage count is the wrong first question

Raw coverage count is the wrong first question because it is a gate, not a ranking. Either a provider covers the securities your users actually hold or it does not, and the way to find that out is to send a sample of tickers from your own order book and see what comes back — not to read a headline number.

Past that gate, breadth stops discriminating. A provider screening every listed company in the world under one rulebook tells you less than a provider screening far fewer under five, if your job is to defend a verdict rather than display one. The first gives you a single opinion about every company; the second gives you five opinions about the companies your users own, and only the second survives a support ticket.

Headline coverage figures are also the least checkable thing on a vendor's site. Most are rounded, and most carry neither the date they were counted nor the venues they were counted across, which means nobody — including the vendor — can reproduce them. Ask what the number counts, when it was counted, and which exchanges are inside it. Ours is in the table below, rounded down from a live count, with that date and those exchanges printed beside it.

The seven questions to ask any Shariah screening provider

Each one is answerable in a sentence by a provider who knows how their own data is built.

1. Which standard produced this verdict?

A good answer names a published standard in the response itself, not on a marketing page. Shariah screening is not one rulebook: AAOIFI, Dow Jones Islamic Market, FTSE, MSCI and S&P set different thresholds against different denominators, so the same company can clear one and fail another. Lockheed Martin is the clean illustration — doubtful under AAOIFI, halal under Dow Jones and S&P, not halal under FTSE and MSCI — and every one of those is a correct application of its own rules. If a provider cannot tell you which of the five you are looking at, you cannot tell your users either.

2. Do you return the ratio, or only the result?

A good answer returns the measured figures the verdict was built from, not a pass/fail flag. A verdict is a comparison, and a product that can show a user the ratio it measured beside the limit that standard draws has explained itself; a product showing a green tick has only asserted something. AAOIFI fails a company whose interest-bearing debt reaches 30% of its twelve-quarter average market capitalisation, so a response carrying the ratio lets you show a user where they sit against that line instead of asking them to take the verdict on faith. The practical test is whether you could re-derive the verdict yourself from the fields in the response. If you cannot, you are buying an opinion rather than data.

3. Where do the financial figures come from?

A good answer names the filing. Ratios are only as good as the accounts underneath them, and there is a real difference between figures extracted from a company's own annual report and figures taken from a summary feed that has already decided what counts as debt. Ask which document, which fiscal period, and whether anyone can point at the line that was read. Ours come out of SEC 10-K and 10-Q filings, UK Companies House annual accounts, and Canadian 40-F and investor-relations disclosures.

4. How does a changed verdict reach me?

A good answer is a change feed, not advice to re-pull the universe every night. Verdicts move when companies file, and the user who needs to hear about it is the one holding the position. Ask what the notification path carries as well as whether it exists: a feed that tells you a verdict changed but not which rulebook moved it sends you back to computing a diff you should never have had to compute.

5. How do you screen a fund?

A good answer screens through to the basket and publishes how much of that basket it could see. A fund verdict assembled from a fund's own marketing material is not a screen. The figure that makes a fund verdict meaningful is the share of portfolio weight the provider could actually rate: a verdict drawn over nearly all of a basket and one drawn over much less of it are different claims, and a provider who does not publish the difference is quietly selling you the weaker of the two. Our own fund screen is a portfolio-weight test derived from AAOIFI, and the visible share is published beside every fund verdict: below 95% of a basket we never publish a fund as halal or doubtful, and a fund under that line is rated at all only when the non-compliant weight we can already see breaches the 5% limit. The reasoning is on our ETF page.

6. What happens when you do not know?

A good answer has a distinct state for it, and that state is never “compliant”. Ask what the API returns for a security it has never rated, and separately what it returns for one it has rated but cannot see enough of — a thinly disclosed basket, a company that has stopped filing. If both collapse into the same field as a real verdict, the coverage number you were quoted earlier is not counting what you thought it was counting.

7. Can I evaluate this without talking to you?

A good answer is public documentation, a key in minutes and a free tier with a published cap. A provider whose only entry point is a demo booking has told you what every later support question will feel like. Read their error responses before you read their pricing page: what an API does when you send it something wrong is the part of the contract you will meet most often.

Akinda at a glance

The five attributes a buyer usually compares, with the basis for each rather than a round number.

The AAOIFI screen behind that primary verdict fails a company whose prohibited income exceeds 5% of total revenue, whose interest-bearing debt reaches 30% of its twelve-quarter average market capitalisation, or whose cash and short-term investments reach 30% of that same denominator. The thresholds, the comparators and what each of the other four standards does differently are set out on the methodology page, with the AAOIFI rules themselves on their own page. Field names, error codes and rate limits per plan are in the developer documentation.

Seven questions are a framework, not a shortlist. If you already have vendor names on a page, our comparison of the halal screening APIs runs several of these same criteria across the alternatives.

What this article is not

This is a description of how published standards differ and what a data provider can honestly show you — not a religious ruling about which standard to follow. That judgement belongs with a qualified scholar, and ours is named and credentialed on this site. What a vendor can do is apply each standard faithfully, publish all of them, and show the numbers underneath — which is also what makes one worth choosing.