Signal Band

Companion guide

Explaining your results

For founders and buyers presenting a Signal Band report to an executive, board or investor. It covers how to read the report quickly, a five-minute briefing you can follow, and plain answers to the questions people usually ask.

Read the report in two minutes

  1. The headline. The first bold sentence is the single most important finding, with its number.
  2. What this means. Why that number matters to the business: the risk, the money or the time at stake if the open assumptions prove wrong.
  3. What we suggest. The few next steps that would move the result most, each tied to your own numbers.
  4. The four boxes. The headline numbers. The small line under each says whether it is good or not, and by how much.
  5. Top risks. The three things most worth worrying about, what happens if each proves wrong, and what the evidence shows today.

Everything after page 1 is the detail behind it. The Briefing notes page near the end gives you three talking points and answers to likely questions, drawn from your own report.

The five-minute briefing

Executives want three things: what you found, why it matters, and what you are asking for. A simple order that works:

  1. Headline (30 seconds). Read the headline sentence as it is written. For example: "We score 55% for investor readiness, 15 points short of the 70% bar."
  2. So what (90 seconds). Pick the biggest risk from the Top risks table and say what happens if it proves wrong. Then name your strongest evidence, so the picture is balanced.
  3. The suggestion (90 seconds). Walk through the first two suggested next steps: what you will test, how much evidence is enough, and by when.
  4. The ask (60 seconds). Name the decision you want, with its cost and date. Offer to re-run the check after the tests and show the score moving.

Tip: lead with the number, not the method. If someone asks how it is calculated, the appendix answers it in a page.

Venture Readiness: questions people ask

What does "investor readiness" actually measure?

How much of what your plan depends on is backed by evidence. Every plan rests on assumptions, such as "customers will pay" or "we can hire the team". Proven assumptions count in full, partly proven ones count in part, and untested ones count as zero. The ones that matter most carry the most weight. 70% is the bar investors commonly look for.

Is 55% out of 100, or 55% of the way to 70%?

Out of 100. A score of 55% is 15 points short of the 70% bar.

Why is my score lower than I expected?

Most first runs score low because most of the plan has not been tested yet. That is not a verdict on the idea. Each test you complete and add to the check raises the score.

The chart shows 10 stages, but the report counts 11 assumptions. Why?

The assumptions are grouped into the ten questions an investor works through, the "stages". Some stages hold more than one assumption: for example, Capability & Capital holds both "the team can build and sell this" and "we have enough cash". A stage's percentage is the average of the assumptions in it.

What is "Exposure if wrong"?

How much of the plan rests on important assumptions that are not proven yet. Lower is better, and 35 or less is the comfort level. A high number means that if one of those assumptions proves wrong, a lot of the plan would need to change.

What is a "high-impact unknown"?

An important assumption with no strong evidence yet. Investors usually accept three or fewer: more than that means too many big bets at once.

Does a good score mean the venture will succeed?

No. It means fewer of the plan's bets are untested. It is a measure of evidence, not a prediction.

How do I raise the score?

Work through the suggested next steps in order. Each one says what to do, how much evidence is enough, and which answer to update. Then run the check again, as often as you like during your 30 days.

AI Investment: questions people ask

What is "backed by evidence"?

The part of each claimed benefit that the evidence supports. A claim with no evidence counts as $0. A claim with strong evidence, such as live results or a test against a comparison group, counts for most of its value.

Why does a live tool with good results still show less than 100% backed?

Each claim is discounted by how strong its evidence is, not only by the result seen. Live production data counts the most and a vendor's claim the least, and one piece of evidence on its own is never treated as conclusive.

What do Fund, Pilot and Stop or redesign mean?

Fund to scale: strong evidence, a good return and at most one warning sign. Pilot: promising, but fund only the next proof, with a date and a cost. Stop or redesign: the evidence suggests the spend is not yet justified, because the backed benefit doesn't cover the cost, or there are several warning signs.

Does "Stop or redesign" mean the idea is bad?

Not necessarily. It means that, on the evidence entered so far, the spend isn't yet justified. A cheap pilot, a measured starting point or a lower running cost can change the decision.

What is "return on the evidence"?

The evidence-backed benefit minus the cost to run, divided by the cost to run. Below zero means it costs more than the evidence says it returns.

Why is the starting point (baseline) so important?

Without a measured "before", nobody can prove the "after". It is the most common reason AI business cases cannot show their value.

Where do the spend ranges come from?

The cost of the AI itself is simulated 2,000 times using your usage and pricing, giving a likely range rather than one number. The middle case is the typical month; the top of the range is a bad but plausible month.

Why does a measured result count, but not in full?

What actually happened is the best evidence of what will happen, so the share of the claimed benefit you measured sets how much is counted. It is then discounted by how dependable that kind of measurement is: live production data counts 90% of what it shows, a controlled A/B test 80%, a pilot 50%. A single measurement can be skewed by timing, a small sample or a good month. Example: live data showing 90% of the effect counts as 81% backed.

What if part of the benefit has no claim behind it?

It counts as unproven, and the report flags it with the dollar amount. If you claim it, you need to prove it: add a claim that says where it comes from, with evidence, or reduce the claimed benefit.

Why was a live initiative paused for review?

After 12 months fully operational, at least 80% of the claimed benefit should be showing up. If it is not, the check suggests pausing further spend while you work out how to reach the full benefit, or resetting the claim. In the first 12 months, a shortfall is treated as a pilot with a checkpoint, provided the benefit is not declining and the backed benefit covers at least half the running cost.

How do I improve the result?

Add evidence. Measure the starting point, run a short pilot against a comparison group, or add live results, then run the check again.

Common pushback, and how to answer it

"These are just your own numbers."

Yes. The report scores the evidence we entered with a published, repeatable method. Its value is showing which parts are proven and which are not, so we know where to test next.

"Why should we trust the score?"

Every figure traces to an input and a published rule, and the same inputs always give the same result. The appendix shows the cut-offs and the method.

"What do you want from us?"

Name the decision you want: funding for the next proof, a hire, a pilot or a pause. Put the cost and the date next to it.

"What happens if we do nothing?"

Use the "What this means" section: it sets out what is at stake, in money or time, if the open assumptions prove wrong.

Words that help

These results come from your own inputs, so present them as guidance, not as fact.

Decision support, not financial, legal or investment advice. Questions about a report: [email protected].