What will Deligatr claim, and what will it not?
We claim only what we can trace to something measured or published. One fabricated figure discounts every real finding beside it, so we would rather say less.
The short answer
Every number on our pages is either sourced, from our own measured work, or labelled illustrative. We never state a conversion rate we did not measure, never invent an industry average and never use fear or false urgency.
The same rule applies to this documentation. The only external figures come from published third-party research, cited where they appear. Figures about our own work come from our own working files.
What do we always do, and never do?
Six things we always do, and six we never do.
Our promise
We always
- Trace every number to something measured or published
- Say plainly when we could not check something
- Show what is working, not only what is broken
- Tell a measurement apart from an assessment
- Label every illustrative example as illustrative
- Disclose when a result comes from our own sister company
Our red lines
We never
- State or imply a conversion rate we did not measure
- Invent an industry average
- Name a competitor that did not come up in a real search
- Tell you “this is costing you X” without a source for X
- Use fear or false urgency
- Turn one case into a promise of results
How do we source the numbers we publish?
A number reaches a page only after it has a named source, a sample size where one exists, and a label that says what kind of number it is.
External figures come from published research
We name the publisher, the study and the sample size where it is known, on the page where the figure appears. Where a finding is a correlation, we say so.Our own figures come from our own working files
For example the keyword counts behind our site architecture, or the answers recorded in a live AI visibility test. We do not round them up.Derived figures are labelled as derived
When we divide or compare published numbers ourselves, we show the working or say that it is derived.Illustrations are labelled as illustrations
Mock-ups use generic names and made-up wording. They show the shape of a thing and are never presented as results.Anything else is removed or stated without a number
If we cannot trace a claim to a source, it either goes, or we say it in plain words with no figure attached.
See the research behind our method on AI visibility research.
What is the difference between measured and assessed?
A measurement comes from a live test. An assessment is our judgement of signals we can observe. We never blur the two.
Measured: AI visibility
Assessed: conversion readiness
The same method powers the free AI Visibility Checker and our published research method. How the two scores are read is covered on the Audit tab.
How do we disclose our own results?
When we use a result from Webtec, we say plainly that Webtec is our sister company, the agency Deligatr was built inside, and that it is one case.
Webtec, one case
Cost per meeting before and after switching to outbound
Webtec is Deligatr’s sister company, a UK agency, so these figures are in pounds. Before the switch, meetings came from Meta ads with a 50% no-show rate. After it, meeting volume doubled in the first quarter.
£400
per attended meeting, before (Meta ads)
~£160
per meeting, after (outbound)
Different bases (attended versus booked), so not a like-for-like comparison.
Webtec figures, one case, reported in pounds. Not an average and not a forecast for your business.
Our disclosure rules
- Webtec is never presented as an independent customer.
- Quotes from Webtec are labelled as coming from the Webtec team, our sister company.
- One case is never turned into an average, a range or a promise.
- Client case studies carry the line “Results from one client. Yours will vary.”
What does a person check before anything reaches you or a prospect?
Every automated output has a human checkpoint before it reaches a prospect, and uncertain data is flagged for review instead of guessed.
Every sequence approved
A person reviews outreach before it sends.
Replies read by a person
Nothing is auto-closed without someone reading it.
Batches spot-checked
Generated call hooks are checked before callers use them.
Doubt is flagged, not guessed
Uncertain data goes to review instead of being filled in.
Claims checked before they ship
An unsourced statistic gets sourced or removed.
Limits stated
Audits say what we could not check.
Who owns my data, and is visitor identification compliant?
Who owns your leads, how outreach stays compliant and what happens to your data if you leave are covered on their own page.
One point belongs here. Rules on identifying website visitors differ by market, particularly the UK, the EU and California, so identification is configured with each client around the consent requirements of the markets they sell into, before it goes live. The mechanics are on the Visitor ID tab.
Questions about our claims
Short answers on sourcing, disclosure and guarantees.
Does Deligatr guarantee results?
We do not guarantee a result number. What we do guarantee is the work: a human approves every message before it sends, copy and design are redone at no extra cost if they miss the agreed standard, you get a named senior owner, a weekly report and a review call, and if setup is not live in about a week we say so and tell you why. Results depend on your market, your offer and what a customer is worth to you. On the discovery call we tell you plainly what we would expect, and we say so if a channel is not the right fit.
Where do the statistics on Deligatr pages come from?
External figures come from published third-party research and are cited where they appear, with the sample size where it is known. Figures about our own work come from our own working files. Anything we cannot trace to a source is removed or stated without a number.
Is Webtec an independent client?
No. Webtec is our sister company, the agency Deligatr was built inside. Its results are one case, and we say so wherever they appear.
Why are some examples marked as illustrative?
Mock-ups of an inbox, a message or an AI answer use generic names and made-up wording so you can see the shape of a thing. They are never presented as real results, and they are always labelled.
What is the difference between measured and assessed?
A measured result comes from a live test, such as asking AI assistants real buyer questions that never contain your name. An assessed result is our judgement of signals we can observe, such as calls to action in a page. We never present one as the other.
What should I do if a figure on this site looks wrong?
Tell us through the contact page. We would rather correct a number than defend it.
Looking for prices? See Pricing or the full pricing page. Back to the Overview.