Quote automation
Quote automation gets bought to make quotes faster, and the typing was never the slow part. The slow part is that the price depends on twelve things and only two people can decide it, and one of them is on holiday. Write those rules down and the quote takes a minute. Skip that, and you have automated a document nobody trusts.
Where the two days actually go
- Waiting for a number only one person knows. Volume break, a special rate for that customer, the discount someone promised last spring. It is in a head, not a system. Everything queues behind that head.
- Rebuilding the same quote from scratch. The last one like it exists, in someone's sent folder, in a version nobody is sure about. So it gets rebuilt, and the errors get rebuilt with it.
- Costs that moved and nobody told the quote. Materials, freight, currency. The template still carries a figure from March. A quotation automation that copies the old number faithfully is a machine for losing money at scale.
- The approval loop. Below X anyone can send it, above X it needs a signature — except nobody wrote X down, so everything gets checked or nothing does.
What has to exist before a quote can be generated
This is the honest gate, and it is short. If you cannot fill in this list, no software fixes it — and I would rather say that in the free hour than after an invoice.
- A price list that is true today, with a date on it and an owner's name against it.
- The discount ladder, written as rules: by volume, by customer tier, by product family. Rules, not precedents.
- The floor. The margin below which a quote does not go out without a signature, per product family, because a blanket floor is always wrong somewhere.
- What is being quoted, as data. If every job is described in free text in a different way, that is the real project, and it is worth doing anyway.
Where the margin actually leaks
Speed is the reason people buy this. Margin is the reason it pays for itself, and the leaks are dull.
- The rounded-down line. A hundred small courtesies a year, each defensible on its own, none of them ever added up. A machine adds them up and shows you the total. That number changes conversations.
- The quote nobody chased. Sent, never answered, never followed up, quietly counted as a loss. Chasing is scheduling, which is the easiest thing here to automate and the last thing anyone does.
- The discount that outlived its reason. Given for a launch three years ago, still applied. Rules with dates on them expire. Precedents in an inbox never do.
- Quoting work you should refuse. The fastest quote engine in the world will happily produce a beautiful price for a job that loses money. Encoding what you do not quote for is part of the build.
Wrong fit if
- Every job is genuinely bespoke and priced by judgement — engineering-to-order, restoration, anything where the estimate is the expertise. Automate the assembly of the document, not the number.
- You send a couple of quotes a week. The gain is real but small; spend the money on chasing the ones you already sent.
- Your price rules are deliberately unwritten because they flex per customer. That is a strategy, and it is incompatible with this, and it is your call to make — not mine.
- Your product data lives in free text. Fix that first. It pays off in more places than quoting.
Margin and assembly, already running
- Promotion guard — margin discipline as a running system: it keeps products out of forced discounts nobody agreed to, on a schedule, without anyone watching a calendar.
- Recommendation shelves — structured product data doing work — 968 products tagged by their real attributes, and the matching built from those tags instead of a maintained list.
- Photo factory — assembly at volume: one product photo in, a full listing out, images, video and text. One real batch was 32 products in a single run.
Honest answers
What does quote automation need from us before it can work?
A price list that is true today, a discount ladder written as rules rather than precedents, a margin floor per product family, and the thing being quoted described as data instead of free text. That list is the project. The document generation on top of it is quick.
Can it send a quote without anyone looking?
Inside the rules, yes: standard product, standard terms, margin above the floor, existing customer. Below the floor, non-standard terms, or a new customer with credit exposure, it stops and routes to a person with the reason attached.
We price by judgement. Is this useless for us?
Not useless, but narrower. The number stays yours. What automates cleanly is everything around it — pulling current costs, assembling the document, versioning it, chasing the answer. That alone usually recovers more time than people expect.
Does it work with our CRM?
It has to, or the quote lands somewhere nobody looks. Where a proper integration is not affordable yet, a scheduled export both ways is honest and it works. What matters is that one system holds the truth about the quote, not two.
What is the fastest useful version?
Current costs plus written discount rules plus a generated document, with every quote logged. No approvals, no CRM, no chasing. That version is small, it is dull, and it usually shows you the discount total nobody had ever added up.
Show me your last twenty quotes
I will tell you which ones your own rules could have produced without a person, where the margin went, and what has to be written down before any of it can be automated. If the answer is that your pricing is judgement and should stay that way, I will say so, and it will have cost you one free hour to find out.
Book that hour →