OUR STANDARD
What we will not say.
Our systems take the shape of your business. They never take the shape of an answer they do not have. This page is the machinery that makes the second half true.
Why a rule set instead of a promise
You have been sold confidence before: a number with a decimal place and nothing behind it.
The problem is rarely dishonesty. It is that most analysis has no mechanism making it impossible to overclaim. The template asks for a percentage, so a percentage appears, whether or not the data can produce one.
So we built the mechanism instead of writing a values statement.
Every claim carries a grade, and the grade decides which words are allowed
| Grade | When | What may be said |
|---|---|---|
| E4 Robust | Three or more independent lines, consistent, current, enough observations | A figure, a likelihood and a range |
| E3 Medium | One or two lines, or indirect, or a weaker fit | A range and a likelihood. No bare figure |
| E2 Limited | A single line, stale, or a proxy only | A direction and a confidence. No probability. No number |
| E1 Absent | Not in what you sent | No claim. A named gap, and what would close it |
Three things drop a claim one grade: evidence about a different period or segment than the question; independent sources disagreeing; and the one we say out loud, a range wide enough that the recommendation flips inside it. When that happens the honest statement is that the analysis does not decide the question, and we make it.
"Likely" always arrives with its number attached, and never without a reference class. Not "70% likely to recover", but "in seven of the ten months that followed this pattern in your own data, it recovered without intervention".
- E4 Robust
Your DSO from MSME buyers is 78 days, 33 days over the statutory norm. Reducing it to 45 releases Rs 1.4 crore of working capital.
- When
- Three or more independent lines, consistent, current, enough observations
- What may be said
- A figure, a likelihood and a range
- E3 Medium
Your DSO from MSME buyers is between 71 and 84 days, likely (55 to 80%) above the statutory norm. Reducing it would release roughly Rs 1.1 to 1.7 crore.
- When
- One or two lines, or indirect, or a weaker fit
- What may be said
- A range and a likelihood. No bare figure
- E2 Limited
Your DSO from MSME buyers is rising and is above the statutory norm. We can give you the direction but not the size: one source, and it is two quarters old.
- When
- A single line, stale, or a proxy only
- What may be said
- A direction and a confidence. No probability. No number
- E1 Absent
We cannot tell you anything about your DSO. It is not in the data you have shared. Your Tally receivables ageing would close this, and it is worth closing, because this is where working capital usually hides.
- When
- Not in what you sent
- What may be said
- No claim. A named gap, and what would close it
Four kinds of not knowing
| Kind | What we say | What we do |
|---|---|---|
| Absent | Nothing in your data speaks to this | Name the file that would. We do not estimate |
| Stale | Last read Open item: <date>, Open item: <N> periods old. At that age it supports a direction, not a size | Drop the grade |
| Unpriced | This is measured but has never been costed. A business case on it is an argument, not a calculation | Give the mechanism. Refuse the number |
| Contested | Two of your sources disagree. This rests on the first, and here is what changes under the second | Show both. Never average silently |
Every gap we name carries five things: what is missing, why, what it would change, how to close it, and what closing it is worth. A gap that would not change the advice is declared immaterial and set aside. Listing every conceivable missing input without saying which ones matter is its own kind of uselessness.
Three registers, never mixed in one sentence
- Observed
- This is in your data
- Inferred
- This follows from your data, by a mechanism we will name
- Assumed
- This is not in your data. We supplied it, here it is, and here is whether the recommendation survives if it is wrong
The test every paragraph has to pass
Strip every figure and every proper noun out of the paragraph. If what remains is true of businesses in general, it does not ship.
This catches the exact failure mode of machine generated advice: a sentence that reads perfectly, sounds insightful, and would be equally true of any company on earth.
And on this website
The same rules. Every figure here has a source you can check.
The one widely quoted statistic we could have used, that roughly 95% of enterprise AI pilots produce no measurable return, is methodologically contested. So it is not on this site as a headline, even though it argues for us.
There are no client logos we have not earned, no case studies we cannot name, and no roadmap features shown as though they shipped. That is why the Business Pulse page says design partner rather than available.
TRY IT ON YOUR OWN DATA
Bring the messiest file you have.
Put it through the same road our clients' data travels. It will either read correctly, or tell you exactly what to change. Nothing is stored, and you do not need an account.