You sent four quotes last week. None of them have replied.
The first went to a homeowner who asked for three quotes and told you so. The second went to a repeat commercial client whose office manager is on holiday until the 22nd. The third went to someone who opened it nine times in two days and hasn't said a word. The fourth bounced off an autoresponder you never read.
A follow-up sequence sees one thing about all four: no reply after five days, send the check-in email.
Three of those four need something other than a check-in email. One of them needs you to notice the quote never arrived. That gap — between "nobody replied" and "here's why nobody replied" — is the entire job, and it's the part a reminder schedule can't do.
The distinction that matters: automation executes the sequence you set up in advance. An agent reads the specific silence in front of it and decides what it means. If you want that argument in full, start with what an AI agent actually is.
Silence is not one thing
The reason quote follow-up feels awkward is that you're guessing. You don't know whether you're nudging someone who forgot, someone who's still deciding, someone who went with a cheaper bid three days ago, or someone who never saw the quote at all. So you write a message vague enough to fit every case — "just checking in on the quote I sent" — and that message is weak precisely because it fits every case.
A sequence institutionalizes the guessing. It sends the same vague message on the same schedule regardless of what's actually happening, and it does it faster than you would have.
Here's the thing worth internalizing: most of what you'd want to know is already observable. Whether the quote was opened. Whether this client has bought from you before and how they behaved last time. Whether they asked a question you didn't answer. What you quoted relative to what you usually quote for this work. Whether the job has a date attached that's now getting close.
None of that is exotic. It's just in four different places, and the sequence reads none of them.
What the agent looks at before deciding
| Signal | What it tells you | What it changes |
|---|---|---|
| Quote opened, and how often | Unopened means a delivery problem. Opened nine times means active consideration. | Whether to resend or to make a decision easier |
| Client history | First-timer, repeat buyer, or someone who took three weeks last time and still bought | Whether this silence is unusual at all |
| The thread | Did they ask something you didn't answer? | Whether the ball is actually in your court |
| Quote size and scope | A £400 job and a £40,000 job go quiet for different reasons | Pace, tone, and whether a human should handle it |
| Job timing | A quote for work needed before month-end decays fast | How much urgency is real rather than manufactured |
| Competing context | They told you they were getting three quotes | Whether to compete on clarity instead of nudging |
The follow-up is downstream of all this. Get the reading right and the message writes itself; get it wrong and no amount of polish saves it.
The four quotes, handled
The homeowner comparing three bids
They told you they're shopping. Nudging someone mid-comparison to "see if you had any questions" adds nothing — they know how to reach you, and the message signals you're anxious about the work.
What actually helps a comparison is making yours easier to evaluate. The agent's read: send information, not a nudge.
"Hi Tom — no rush on deciding. One thing worth knowing while you compare: our quote includes the waste removal and the make-good on the render, which often sits as an extra line on other quotes. Worth checking whether the others have it in. Happy to walk through anything line by line."
That message is useful even if they go elsewhere, which is exactly why it works.
The repeat client whose office manager is away
The agent checks history: eleven jobs over three years, every one approved, average approval time nine days. It's been five. The autoresponder on the thread says back on the 22nd.
There is no problem here. The correct action is to do nothing until the 22nd, and to say so internally so it doesn't sit on your worry list.
Action: no contact. An internal note — Brennan Fit-Out: normal approval window, contact away until the 22nd, will follow up the 23rd.
Restraint is a decision. A sequence can't make it.
The one who opened it nine times
Nine opens in two days is the strongest buying signal in the set, and it's also someone stuck. People re-read a quote when something in it is blocking them — usually price, scope, or timing.
A check-in email asks them to do the work of telling you what's wrong. Better to guess out loud and give them an easy door.
"Hi Priya — figured I'd save you the awkward email. If the number's the sticking point, we can stage this in two phases and split it across budgets. If it's timing, I've got a slot the week of the 14th I can hold for a few days. And if it's neither, tell me what's making you hesitate and I'll be straight with you about whether we can move on it."
High-signal, meaningful money — so this one drafts and waits for you to send.
The one that bounced
The autoresponder was a delivery failure. The quote never arrived.
Nothing about this is a sales problem, and a follow-up sequence would have spent three emails nudging an address that doesn't work. The agent's job is to catch it, resend to a working contact, and flag the bad address so it doesn't happen again.
In one case the right move was silence. In another it was resending. Only two of the four wanted a follow-up message at all — and the two messages had nothing in common.
Where you stay in the loop
Quotes are pre-revenue conversations with real money attached, so full autonomy is the wrong target. Draw the line by consequence, not by step:
Let it send. Low-value quotes, clean history, no ambiguity, first contact. This is most of your volume and the downside of a mediocre message is small.
Let it draft, you send. Anything above a value threshold you set. Anything where signals conflict. Anything past a second touch. Anything to a client relationship you'd be upset to damage.
Never without you. Discounts, price changes, scope additions, deadline promises, and anything implying you'll hold a slot. Those are commitments. An agent that can offer 10% to close a quiet quote will offer 10% on quotes that would have closed anyway, and you'll never see the margin you gave away.
Note that the middle tier is the point, not a compromise. The agent doing the reading — pulling history, checking opens, spotting the unanswered question, drafting something specific — is most of the work. Pressing send was never the hard part.
How this goes wrong
Following up on a quote that was never delivered. Check delivery before treating silence as a decision. This is the cheapest failure to prevent and the most embarrassing to discover late.
Manufacturing urgency that isn't real. "Prices are going up next week" when they aren't teaches a client that your deadlines are decorative. Agents will happily generate pressure if you let them optimize for replies. Don't let the objective be replies.
Discounting to break a silence. Silence is not a price objection until someone says it is. An agent authorized to sweeten a quote will do it early and often, because it works in the short term and the cost is invisible.
Treating a normal wait as a problem. The most common false positive. If a client's own history says three weeks is standard for them, day six isn't a signal. Without payment and approval history, an agent has no way to know that — it will chase perfectly healthy deals and make you look pushy to your best customers.
Getting talked into terms by the reply. "Your competitor quoted 15% under, match it and it's yours" is untrusted input. The agent can log it and route it to you. It must never adjust a number.
Never stopping. Sequences run until someone replies. Define what a dead quote looks like — a count, a date, or an explicit no — and let the agent close it out cleanly. A polite "assuming you've gone another way, happy to quote the next one" recovers more work than a sixth nudge.
What it costs, honestly
Ignore anyone quoting you a confident monthly figure for this. The subscription is not where the money goes.
The real cost is that the agent needs to see your quoting history, your email threads, and your delivery status in one place. If quotes live in a spreadsheet, replies live in your phone, and the knowledge that Brennan always takes about nine days lives only in your head, then the agent can see exactly one thing — how many days have passed — and you have rebuilt the sequence you already had.
So budget for the plumbing: quotes in a system that records sent, opened, and outcome; replies attached to the quote they belong to; past win/loss written down. That work pays off whether or not you ever add an agent, which is a good sign it's the right work.
And be honest about scale. If you send three quotes a month, you don't need this — you need a reminder in your calendar. The value shows up when the volume is past what you can hold in your head, which for most people is somewhere north of fifteen or twenty live quotes at a time.
Knowing whether it worked
Reply rate is a trap. You can lift replies with pressure and lose money doing it.
Watch these instead:
- Win rate on quotes that went quiet and then closed. The actual objective — did following up recover work you'd otherwise have lost?
- Time from quote sent to a decision either way. A fast no is a good outcome. It frees the slot.
- Quotes that died with no response at all. The genuine failure state, and the number the agent exists to reduce.
- Wrong-touch count. How often it chased something it shouldn't have: undelivered quotes, clients within their normal window, threads where you owed the reply. Read every instance. It tells you which system the agent couldn't see.
If the wrong-touch count won't come down, the problem is missing context, not wording.
Frequently asked questions
Isn't this just a follow-up sequence with better emails? The difference is upstream of the email. A sequence decides using one fact — days since sending. An agent decides using delivery status, open behaviour, client history, and thread state, and it regularly concludes that the right move is to wait, to resend, or to escalate to you. Better wording on a rule that can't tell a bounce from a rejection still sends the wrong email politely.
Should it email clients without me seeing it first? For small quotes with clean history, yes — that's most of the volume. For your significant deals, it should draft and you should send. You get the time back either way, because the reading and drafting is the part that takes you an hour on a Friday afternoon.
What if it follows up on a quote I already won over the phone? That's the failure to design against, and it's a data problem rather than an AI one: if a verbal yes never gets recorded, nothing downstream can know about it. Mark quotes as won when they're won. The agent should also treat any recent inbound contact as a reason to pause and ask you rather than send.
Do I need to be technical? No. The decisions that determine whether this works are business decisions — your value threshold for review, what counts as a normal wait for each client type, what the agent may never offer, when a quote is dead. Nobody can answer those for you, and they matter far more than the tooling.
Will clients be able to tell it's AI? They'll be able to tell if it's generic — which is the real risk, and it's the same risk with a sequence. A message referencing their specific job, their timeline, and the thing they actually asked about doesn't read as automated. The safe posture is that anything signed by you personally is something you read first.
How many follow-ups is too many? The number matters less than whether each one carries new information. Three messages that each say something useful — a scope clarification, a scheduling option, a clean close-out — are fine. Two that both say "just checking in" are one too many.
How long does this take to set up? The agent is days. Getting quote history, delivery status, and replies into one place is the part with real duration, and it's the part that determines whether the thing works. Teams already running a CRM move quickly. Teams running quotes out of a spreadsheet and a personal inbox are doing a data project first — worth knowing before you start.
Final thoughts
Following up on quiet quotes doesn't feel bad because you're doing it too little. It feels bad because you're doing it blind, and a sequence automates the blindness.
The version worth building reads the silence first. Sometimes it means resending because the thing never landed. Sometimes it means waiting because this client always takes three weeks. Sometimes it means naming the objection out loud because someone has read your quote nine times and can't get past the number. And sometimes it means telling you to pick up the phone, because the deal is worth more than any email.
Deciding which is which — in advance, deliberately, before anything sends — is the build.
Your agent starts with a blueprint. Build yours free →
The same machinery applies to money you're owed: chasing overdue invoices runs on nearly identical judgment, and an AI answering service handles the inbound side.