Home

Services

SEO Services Social Media Website Development AI Automation About Reviews Portfolio Blog Contact Free First Look
Chatbots & Enquiry Triage

One Question Decides Everything: What Does It Say When It Does Not Know?

The greeting, the tone, the little avatar — none of it matters much. What matters is the moment somebody asks something the bot was not built for. A bot that answers confidently there will invent a price, a policy or a delivery date, and your customer will hold you to it. A bot that says so and fetches a person is genuinely useful. Same technology, opposite outcomes, one design decision.

No credit card required · 30-minute call · No obligation to sign up

12+ years in digital marketing Working with Chatbots businesses
✓ Built In Your Accounts, Not Ours ✓ Quoted & reported in GBP (£) ★★★★★ Trustpilot 5.0 ★★★★★ Google 4.9
A customer support conversation open on a laptop screen
The behaviour that decides itNot knowing
Deflection-rate statistics quotedNone
The Numbers First

Six questions it will be asked that it cannot answer

Every one of these arrives in the first week. What the bot does with them is the difference between a useful thing on your site and one you switch off in a month.

  • What would this cost for me?The most common · and the most binding
  • Can you do it by Friday?It has no idea about your diary
  • Do you cover my area?Answerable, if it was given the list
  • Something has gone wrongWhich needs a person, immediately
  • A question about a competitorWhere any answer is a bad one
  • Something legally loadedRefunds, guarantees, obligations

The first row is worth pausing on. A bot that offers a number because it sounds helpful has made a commitment in your name to somebody who screenshotted it. Pricing questions should either return a genuine range you have agreed to publish, or a person — and which of those it does is a decision you make, not one the model makes on the day.

Straight Talk

Give it a small subject and a fast exit

A bot that knows a little, precisely, is worth several that know everything vaguely. Narrow it to the questions you can actually answer in writing — opening hours, coverage, what happens at a first appointment, how long things take, what to bring — and let it say plainly that anything else needs a person. The narrowing feels like a reduction in ambition and it is what makes the thing trustworthy.

The exit matters as much as the answers. When it hands over, the person receiving it should get the whole conversation, not a fresh start, because being asked to repeat yourself to a human after failing with a robot is the moment people give up. And the handover has to say honestly what happens next: a person now, or a reply by a stated time. Silence dressed as a handover is worse than the bot admitting defeat.

What it draws on should be a small set of things you control and can update — your own pages, your own answers — rather than the whole internet. That is what stops it improvising, and it also means fixing a wrong answer is an edit rather than a rebuild. When somebody asks something it got wrong, that becomes a new entry, and the bot is better on Thursday than it was on Tuesday.

Then it needs watching, particularly early. Read what people actually asked in the first fortnight, because two things fall out: the questions it should have been able to answer and could not, and the questions you did not know your customers had. The second list is frequently more valuable than the bot, and it feeds straight into the website.

  • Narrow subject, precisely answered — Knowing a little exactly beats knowing everything vaguely.
  • Hand over the whole conversation — Repeating yourself to a human is where people give up.
  • Its knowledge is yours and editable — A wrong answer becomes an edit, not a rebuild.
  • Read the first fortnight of questions — Two lists: what it missed, and what you did not know they asked.
  • Pricing answers are a decision — A published range, or a person. Never whatever sounds helpful.
A support team reviewing customer conversations together

A bot that offers a number because it sounds helpful has made a commitment in your name to somebody who screenshotted it.

.
Ghalib Ashrafi Founder & Digital Strategist · 12+ years across search, social & web

What building one actually involves

Six things, and the first is deciding what it will refuse to answer.

01

A Written “Do Not Answer” List

Pricing beyond a published range, dates, anything legally loaded, anything about a competitor. Agreed by you, so the model is never deciding these under pressure.

02

A Small, Editable Knowledge Set

Your own pages and your own answers, not the whole internet. It is what stops improvisation, and it makes fixing a wrong answer an edit rather than a rebuild.

03

A Handover That Carries Context

The whole conversation goes to the person, with a stated next step — someone now, or a reply by a named time. Never a silent transfer into nothing.

04

The Things It Genuinely Knows

Hours, coverage, what happens at a first appointment, how long things take, what to bring. Unglamorous, constantly asked, and answerable with certainty.

05

The First Fortnight, Read

What it could not answer and should have, and what customers asked that you never knew they wondered. The second list often improves the website more than the bot.

06

Consent and Data, Handled

A chat widget is a third party receiving your visitors. It loads behind consent like everything else, and where the data goes is decided rather than defaulted.

NarrowWhat it knows, and knows exactly
WrittenThe list of what it will not answer
WholeConversation, carried to the person
EditableA wrong answer is a fix, not a rebuild

What clients say

Real clients, quoted in their own words — published with their permission.

More of them, in full, on our reviews page.

— Our Proprietary Methodology —

The Visibility Framework™, applied in Chatbots

The method doesn’t change by market. What it’s pointed at does.

Step 01

Audit The Hours

Where time actually goes, task by task, scored on volume, repetition and the cost of getting it wrong. Ends in a ranked blueprint with estimated hours saved — yours to keep either way.

Step 02

Design The Guardrails

Before any building: what the agent may touch, where a human must approve, what happens when it is unsure, and which data is never allowed near a third-party model.

Step 03

Build & Evaluate

One workflow at a time, in your accounts, scored against real examples from your business before it touches live work. Shipped early so it meets reality while it is still cheap to change.

Step 04

Run & Improve

Monitored for cost, failures and quality drift. Models change, your business changes, and an automation nobody tends becomes a liability rather than an asset.

Honest, No-Nonsense Commitment

If the audit concludes that a task is not worth automating, we will tell you and refund the difference rather than build it anyway. And if a workflow we built does not hit the outcome we agreed in the blueprint, we keep working on it at no extra cost until it does or we take it out.

Investment

AI Automation pricing for Chatbots, in GBP

Quoted in pounds. Model usage is billed to your own account rather than through us, and it is usually smaller than people expect for a bot with a narrow subject — another reason narrowing is the right instinct.

Answers

The questions you get every day.

£1,500 – £3,500
  • Written do-not-answer list
  • Small editable knowledge set from your pages
  • Handover with the full conversation attached
Get a Quote

Multi-Channel

Site, WhatsApp and social in one place.

£8,000+
  • One knowledge set across every channel
  • Per-channel handover rules and hours
  • Evaluated before each channel goes live
  • Optional: combine all 4 services for full-funnel growth
Get a Quote

Every plan is scoped around your market — start with a free first look and we’ll recommend what fits, priced in GBP.

What you’re actually committing to

Most agencies keep this in a contract you only see after the sales call. We would rather you knew now, because it is the question everyone asks second — right after the price.

  • The audit is credited, not sunkPay for the audit, and the full amount comes off the build if you proceed. If you don’t, the blueprint is still yours to hand to anyone else.
  • A fixed build price after the auditQuoted once we know what we are building. If it takes longer than we estimated, that is our risk — the price only moves if you change the scope.
  • You own everythingAccounts, API keys, workflows, prompts, evaluation sets, logs and documentation — all in your name from day one, and still yours if we never work together again.
  • Running costs are yours and visibleAPI usage is billed by the provider directly to you. We never resell tokens or mark up usage, and you see the real number.
  • The retainer is month to month30 days’ notice, no exit fee. Stop it and your automations keep running — you are simply maintaining them yourself.
  • Human approval is the defaultAnything customer-facing or irreversible needs a person until the evaluation data justifies otherwise, and that decision is yours to make, not ours.

These are the terms as they appear in the agreement itself — nothing here is softened for the website. The full wording lives in our terms and conditions, and you get the agreement to read before anything is signed or invoiced.

Chatbots AI Automation questions, answered

What happens when it does not know something? +
It says so and fetches a person, with the whole conversation attached. That single behaviour is the difference between a useful thing on your site and one you switch off in a month — because the alternative is a bot that answers confidently at the edge of its knowledge and invents a price, a policy or a date that a customer will hold you to.
Only a range you have agreed to publish, or not at all. A bot that offers a number because it sounds helpful has made a commitment in your name to somebody who screenshotted it. Which of those two it does is your decision, written down, rather than something the model works out on the day.
No, and trying is what makes bots untrustworthy. One that knows a small subject precisely is worth several that know everything vaguely. Narrowing it feels like reduced ambition and it is exactly what makes the thing worth having.
You edit the knowledge it draws on, because that is a small set of your own pages and answers rather than the whole internet. A wrong answer becomes an entry, and the bot is better on Thursday than it was on Tuesday. That is only true if it was built narrow in the first place.
Read the first fortnight properly. Two lists fall out: things it should have been able to answer and could not, and things customers asked that you never knew they wondered about. The second list frequently improves your website more than any change to the bot.
A chatbot says things; an agent does things. The worst a bot can do is give a wrong answer, which is bad enough. The worst an agent can do is take a wrong action you then have to unwind — which is why that page is about permissions and this one is about honesty.
Keep Exploring

Related AI Automation pages

Same service, different angle — by market, by service and by industry.

Get A Free Chatbot Review

Send us the questions you get every week. We will tell you which a bot could answer with certainty, which should always reach a person, and whether it is worth building at all.

Want one that admits what it does not know?

Free 30-minute call · No obligation · Response within 24 hours

Get in touch with Ghalib Ashrafi HQ

Ghalib Ashrafi takes on work for brands in six markets — the UK, USA, UAE, Saudi Arabia, Australia and Pakistan. Every enquiry gets a reply within 24 hours.

Phone / WhatsApp: +92 343 2653224
Email: info@ghalibashrafi.com
Hours: Mon–Sat, 10am–7pm (PKT)
Still deciding? Get a free first look — no obligation. Get Free First Look