The interesting question is not accuracy
People arrive at a comparison like this expecting a verdict on which tool computes more accurately. In Guna Milan that question is close to settled: the 36 points are eight lookups against classical tables that have not changed in centuries, applied to two Moon positions. Prokerala implements them properly. So do we.
The genuinely interesting question is why two properly-built tools can still hand you different numbers for the same two people — and what that difference is actually telling you.
Where Prokerala is the better tool
Stated up front, because a comparison that only flatters its author is not worth reading.
- Presentation of the breakdown. Prokerala's koota-by-koota report is among the clearest in the category. Showing all eight scores with their maximums, legibly, is a design achievement that many larger platforms have not managed.
- A mature, documented API. If you are integrating chart or matching data into your own product, Prokerala's astrology API is the more established option for straightforward endpoints. That is a real advantage and we would point a developer there for that job.
- Multi-language coverage. Its language support across report surfaces is broader than ours.
- Standalone report quality. For a single, self-contained matching report you can print and hand to a family, it is a strong product and does not require an account.
- Panchang and calendar depth. Adjacent to matching but often needed alongside it, and well covered.
Why two correct tools disagree
There are three real sources of divergence, and only one of them is a bug in anybody's code.
1. Cancellation rules — the big one
Classical literature describes exceptions that restore a zeroed koota. The Bhakoot exception most implementations agree on is shared rulership: when both Moon signs are ruled by the same planet, the 7 points come back. Exactly three sign pairs qualify — Mesha with Vrishchika (both Mars), Vrishabha with Tula (both Venus), and Makara with Kumbha (both Saturn). We apply it, and most serious tools do.
Nadi is where consensus breaks down. Sources describe exceptions for the same nakshatra with different rashis, the same rashi with different nakshatras, shared nakshatra lords, and more — and they do not agree with each other. A tool that auto-applies one of those lists and a tool that does not will report different totals for the same couple, and neither is making an arithmetic error. They are following different authorities.
We chose not to auto-apply any Nadi cancellation. A same-nadi pairing scores zero and is flagged. The reasoning is that auto-applying means picking a tradition on your behalf and hiding that choice inside a number you have no way to interrogate. We would rather show the flag and say that recognised exceptions exist, so the conversation happens with an astrologer who can name the authority they are following.
2. Ayanamsa and borderline positions
We use Lahiri (Chitrapaksha), as most Indian practice does. Different ayanamsa choices shift the sidereal Moon slightly, which matters only when the Moon sits near a boundary. But because a nakshatra is just 13°20′ wide, borderline cases are not rare, and a nakshatra that flips takes Tara, Yoni, Gana and Nadi with it — up to 21 points.
3. Assumed birth times
If either tool substituted a default hour for a missing birth time, the resulting nakshatra is an assumption, not a reading. Two tools assuming different defaults will diverge, and neither result is more correct. This is the divergence most worth ruling out first: check whether you actually supplied both birth times.
A concrete case where the rule decides everything
Two couples, identical Bhakoot geometry, opposite results — decided purely by whether one planet happens to rule both signs.
| Pairing | Moon signs | Distance | Shared ruler? | Bhakoot |
|---|---|---|---|---|
| A | Mesha & Vrishchika | 6/8 — in the failing set | Yes — both Mars | 7 of 7, cancelled |
| B | Vrishabha & Kanya | 5/9 — in the failing set | No — Venus and Mercury | 0 of 7, dosha stands |
A 7-point swing decided by a rulership table, with nothing else about either chart entering the calculation. If you ever see two tools differ by exactly 7 or exactly 8 points, a cancellation rule is almost certainly the reason — those are the Bhakoot and Nadi weights.
Side by side
| Dimension |  AskSoma |  Prokerala |
|---|---|---|
| 36-point Ashtakoot calculation | Yes | Yes |
| Clarity of koota breakdown | Full breakdown shown | Excellent — a category benchmark |
| Documented astrology API | Yes | More established for chart endpoints |
| Multi-language report coverage | Narrower | Broader |
| States which kootas your inputs cannot reach | Yes | Not prominently |
| Conversational follow-up on a result | Yes — AI astrologer | No |
| Automatic Bhakoot shared-ruler cancellation | Yes | Yes |
| Automatic Nadi cancellation | No, by design — flag shown instead | Varies by report |
| Ayanamsa | Lahiri (Chitrapaksha) | Sidereal, Lahiri by default |
Compiled from public product surfaces in 2026. Platforms change; verify anything that will affect a decision.
Which to use
Use Prokerala if you want a clean standalone matching report to print or share, or if you are integrating matching into your own software and want a documented API.
Use AskSoma if you want to know what your particular inputs could and could not determine, and to ask follow-up questions about the result rather than receiving a static document.
Using both is entirely reasonable, and if the two disagree, the disagreement is informative: it is almost always pointing at a cancellation rule or an assumed birth time, and both are worth knowing about.
What neither tool can tell you
- Whether a marriage will last. The 18 / 24 / 28 thresholds are practice conventions with no outcome data behind them.
- Anything outside the Moon. The 36 points never consult the Sun, Venus, Jupiter, the Lagna, the houses or the dashas.
- Manglik status without both birth times — it is measured from the Lagna, which advances a full sign roughly every two hours.
- Which cancellation tradition is the correct one. That is a question for a human astrologer, and the honest answer is that authorities differ.
Frequently asked
Why do AskSoma and Prokerala give different Guna Milan scores?
Almost always because of cancellation rules rather than arithmetic. The eight koota lookups are fixed, but classical sources describe exceptions that can restore a zeroed Bhakoot or Nadi, and the lists are not unanimous. A tool that auto-applies a cancellation and one that does not will report different totals for the same couple, and both can be defensible. Ayanamsa choice and rounding of the Moon's position can also shift a borderline nakshatra.
Is Prokerala kundli matching accurate?
Its ephemeris and Ashtakoot implementation are solid, and its presentation of the koota breakdown is among the clearest available. Accuracy in this domain is largely settled — the 36 points are fixed classical tables. The meaningful questions are which cancellations a tool applies and whether it tells you what your inputs could not reach.
Does Prokerala have a kundli matching API?
Prokerala offers a documented astrology API covering matching and chart data, which is a genuine strength and a real reason to choose it for developer or integration work. AskSoma also exposes a developer API, but Prokerala's is the more established option for straightforward chart and matching endpoints.
Which is better for Ashtakoot matching, AskSoma or Prokerala?
For a clean, well-presented standalone Guna Milan report or for API integration, Prokerala is the stronger choice. AskSoma is aimed at a different job: stating what your specific inputs can and cannot determine, and letting you ask follow-up questions about the result conversationally.
Should a tool cancel Nadi dosha automatically?
It is a genuine judgement call. Classical sources describe several Nadi exceptions but do not agree on them, so auto-applying one means picking an authority on the user's behalf and hiding that choice inside a number. AskSoma does not auto-cancel Nadi for that reason. Tools that do are not wrong, but you should know which convention you are being shown.