How Revalio works
A public scan shows possible gaps. Connected records show what actually happened.
- The free public scanPublic estimate
- It can only see what anyone can see on your website, so everything it produces is an estimate — labelled as one, with its workings attached.
- Your connected recordsObserved
- Once you connect the inbox your enquiries actually arrive in, Revalio is reading real messages from real customers, so estimates give way to counts of things that happened.
Same product, completely different class of evidence. Here is each step, and what Revalio does and does not know at each one.
- Public website scan
- Contact routes
- Response information
- Modelled exposure — estimate
- Connect real enquiry data
- Example draft
“Thanks for getting in touch. Are you still looking for help with this?”
Owner approval requiredSend from your mailbox
1. Scan public pages
You enter a web address. No account, no card, no call. Revalio fetches the publicly available pages of that site and reads them the way a prospective customer would. It respects robots.txt, and most scans finish within a few minutes.
What Revalio can know
- How an enquiry is meant to reach you, and what happens to it when it does, as far as your public pages show it.
- Enough to produce a report: a Revenue Response Score, a revenue exposure range with a confidence band, and named findings — every figure marked as an estimate.
What it cannot know
- Anything behind your website. It does not log in anywhere, and it does not touch your inbox, your CRM or your phone system.
- Benchmark assumptions are category priors that have not been calibrated against measured outcomes. They are labelled "uncalibrated" throughout.
- Enquiry volume cannot be observed directly from a website. Where no demand signal is available, the figure shown is a category benchmark and confidence is reduced accordingly.
- The model treats its four inputs as independent. Real correlation between them would shift the range.
- A public scan cannot observe the sales process, so the conversion estimate is adjusted by no more than ±20% from the category benchmark.
So the honest description of a free report is this: a structured, reproducible estimate of what a business shaped like yours is likely exposing, given what your website reveals about how it handles enquiries. It is not a measurement of your revenue, and it never claims to be.
These figures are estimates based on publicly observable information and benchmark assumptions. They are not a statement of actual lost revenue.
Benchmark assumptions are category priors that have not yet been calibrated against measured outcomes.
Connecting your enquiry data adds measured figures from your own records alongside these estimates; every figure stays labelled with where it came from.
The five stages of a scan
- Reading your website
- Finding how customers can enquire
- Checking response infrastructure
- Comparing against local competitors
- Modelling revenue exposure
What the free report contains
When the scan finishes your browser goes straight to the report. The link is private: reaching it needs both the address and an access token, so nobody finds it by guessing, and we do not publish it.
- A Revenue Response Score out of 100, broken down by component, so you can see which part of the funnel cost you the points.
- A revenue exposure range — a range rather than a single number, because a single number would be false precision — with the reasoning for its width.
- A confidence score and band: the model's own statement of how much it was able to observe about you rather than assume. Low confidence is shown as low confidence, not quietly rounded away.
- The findings. Each one is a named failure mode from a fixed catalogue, with a severity and a recommended fix, rather than free-text commentary.
- The methodology. The formula, the four inputs, every assumption, every source class, the model version and the seed the simulation ran with — so the same scan always reproduces the same figures.
How the exposure figure is built
The estimate is one multiplication, and it is printed on every report.
Estimated enquiries × estimated conversion rate × estimated average transaction value × estimated addressable leakage
Addressable leakage is the share of enquiry value a well-run funnel could realistically hold on to that a poorly run one does not. It is not a claim about money you have lost.
Rather than pick one number for each input, the model fits each to a range and combines them:
- Each input is fitted to a distribution with a stated 10th–90th percentile band.
- The four inputs are combined by Monte Carlo simulation with a fixed seed, so the same scan always reproduces the same figures.
- The published range is the 10th–90th percentile of the resulting distribution — not the arithmetic worst and best case, which would be far wider and much less informative.
- Independent estimators that agree narrow the range; estimators that disagree widen it.
- Where the business files accounts at Companies House, the estimate is capped by the maximum turnover its filing category permits. A figure that would exceed what a business of that filed size could legally bill is not quoted at all.
2. Connect enquiry records
You create an account and connect where enquiries arrive. Four routes work today: your mailbox, your website form by webhook, a CSV of past enquiries, and manual entry. Now the estimates start getting replaced.
What Revalio can know
- Real messages from real customers, in the inbox or form you connect.
- As those enquiries flow in, estimated inputs give way to counted ones: your actual enquiry volume and your actual response times. Once you record what happened to each opportunity, it can also count the share you won, though it never sees outcomes you did not record and never claims it caused one.
- History from day one, if you import a CSV of past enquiries.
What it cannot know
- Enquiries that never reach a connected route. What arrives by phone or in person is only counted if you enter it.
- Settled money on its own. Influenced revenue never becomes verified revenue by itself — a settled figure has to be reconciled against it.
Connection requirements
- Your mailbox — Gmail, or any IMAP inbox on ordinary hosting, including most personal Microsoft accounts. If your enquiries land in an inbox and nowhere else, that inbox is the whole integration. The connection uses an app password, stored encrypted, used only to read that mailbox, and discarded when you disconnect. (A Microsoft 365 work or school account often has app-password sign-in disabled by its own admin — see the connection screen for what to check before relying on this for that kind of account.)
- Your website form, by webhook.
- A CSV of past enquiries, if you want history from day one.
- Manual entry, for what arrives by phone or in person.
Signing up and the free trial
If you want Revalio to do something about what the report found, you create an account. The trial runs for 14 days, with no card and no setup fee.
It does not convert itself when it ends. Revalio tells you the trial has finished and asks. Nothing is charged unless you say yes.
Signing in is passwordless: enter your email and use the one-click link we send back.
What Revalio counts as a result
This is where software usually overclaims quietly, so it is worth being exact.
Revalio reports nested figures, not separate wins. Opportunities acted on. Of those, the ones influenced — meaning a Revalio action happened before the outcome. Of those, the ones recovered — meaning the opportunity was subsequently marked won. Of those, the ones verified — meaning the figure has been reconciled against a settled amount.
Each sits inside the one before it. They are drawn against the largest, never stacked and never added together, because adding them would count the same money more than once. Each carries the class of the weakest evidence in it.
3. Review recovery drafts
You review each draft before it goes. When an enquiry goes unanswered past the window for your industry, or a quote sits unchased past its follow-up point, Revalio raises it and drafts a reply. It also creates plain tasks — call-back lists, chase lists — for the work that is not an email at all.
- By default: draft awaiting your approval
- A new account starts with a small set of automations already on, and every one of them drafts for your approval rather than sending — two simply raise a to-do. Any rule you add yourself is created switched off. A draft always waits for you. The approval screen is built for a phone and can be saved to your home screen.
- Opt-in only: auto-send enabled — sends without review
- Auto-send is a separate, explicit opt-in on eligible plans (Autopilot's Full Auto). You turn each rule on individually and choose, per rule, whether it sends itself or waits for you. Even on the automatic setting, a draft that states a price, offers a discount, makes a guarantee, promises availability or claims someone was already in touch is held back for a person to check, whatever the rule says.
What Revalio can know
- The customer's existing email thread, so a draft can reply inside it from your own mailbox.
- Whether a Revalio action happened before an outcome — that is what "influenced" means.
What it cannot know
- Your close rate. So it does not promise recovery, and any number that claimed to would be invented.
- Whether it caused a job to be won. Revalio saying it acted before a job was won is not the same as saying it caused the job to be won, and the product will not print the second sentence.
Email only. Revalio sends email. It does not send SMS, it does not send WhatsApp, and it does not run multi-channel campaigns. No text-message provider is connected to the product, so nothing in the interface offers to send one. We say so because this category routinely advertises channels it has not wired up. If it is not on this page, Revalio does not do it.
How sending and mailbox permissions work
The draft sends from your own mailbox, inside the customer's existing email thread, with the correct reply headers — so it arrives as a reply from you, not as marketing from a third-party domain. If the mailbox is connected read-only, the send fails rather than quietly going out from our domain instead.
You also choose what a mailbox connection is permitted to do, and it is a real permission boundary rather than a hidden button — enforced in Revalio's own code, not by the mailbox provider. An app-password connection is an unscoped credential: nothing about the password itself distinguishes reading from sending. Observe only means Revalio reads and refuses to call the send function at all. Approval means it can draft, and can send only what you have approved. The third setting allows approved rules to send on their own.
Estimated, observed, verified
Every figure Revalio shows carries a label saying where it came from, and the label sits on the number rather than in a footnote.
- Public estimate
- Modelled from publicly observable signals and category benchmarks. Not a measurement.
- Customer supplied
- You told us.
- Observed
- Revalio saw it happen in your connected data.
- Verified
- Reconciled against a settled figure.
A free report is almost entirely the first kind. The work of using Revalio is moving your figures down that list.
What Revalio does not do
- It does not buy or supply leads. Everything it works on is an enquiry that already came to you.
- It does not cold-email strangers on your behalf. It replies to people who contacted your business.
- It does not promise recovery. We do not know your close rate, and any number that claimed to would be invented.
- It does not send anything you have not either approved or explicitly switched on as a rule.
Revalio does not guarantee revenue recovery.