Your Pricing History Is a Dataset: Using Past Quotes to Win More Work

Every quote your business has ever sent records three things: what you thought the work was worth, what the customer thought of that, and, if you won it, what it actually cost you to deliver.

Most businesses have thousands of these sitting in an email archive and have never put them side by side. It is the single most reliably valuable data project a UK SME can run, and it does not require anything new to be collected.

This article is part of our guide to the data you already own.

The three comparisons

Everything worth knowing comes from three joins.

Quoted versus won. Which quotes were accepted, at what price, for which type of work, for which kind of customer. This tells you where your pricing sits relative to the market.

Quoted versus actual. What you said it would cost against what it did. This tells you where your estimating is wrong, and it is almost always wrong in a consistent direction on specific job types.

Won versus margin. Which work you win and which work you make money on, which are frequently not the same set. This is the uncomfortable one.

What it reliably shows

Across sectors, the same four findings come up so often that we can almost predict them.

One service line is systematically underpriced. Usually the one quoted from a template that was last reviewed several years ago. The gap between quoted and actual is consistent and nobody has noticed because each individual overrun looked like a one-off.

Discounting is concentrated. A small number of relationships absorb most of the discount, and a good proportion of those deals would have closed anyway. Discount given at the point of closing is rarely tested.

Win rate is flat across a wide price band. This is the finding that changes behaviour fastest. If you win roughly the same proportion of work at the top of your range as at the bottom, you are leaving margin on the table and price is not the deciding factor your sales team believes it to be.

Some customers cost more to serve than the quote reflects. Unbilled time, revisions, hand-holding. It clusters, and it clusters predictably.

Running it yourself

You can get most of the way without help, if your volumes are modest.

Define the question first. "Which job types do we underprice" is a question. "Let's look at our pricing data" is not, and it produces a spreadsheet nobody acts on.

Pick three years. Far enough back for patterns, recent enough that the business has not fundamentally changed.

Extract four fields per quote. Date, job type, quoted value, outcome. If you have it, add actual cost. Four fields is enough for the first pass; more fields will slow you down and add nothing.

Group by job type, not by customer. Customer-level analysis feels more interesting and tells you less, because the pattern is usually in the type of work.

Then look at the outliers. The jobs that went worst are where the learning is, and they are individually memorable, so people can tell you what happened.

For a few hundred quotes this is a fortnight of someone's time. For thousands spread across email it is an extraction project, typically £8,000 to £20,000.

A worked example

A 45-person engineering consultancy, £4.2m turnover. Twelve years of proposals in email, job costing in a spreadsheet, never compared.

Extraction across 1,800 projects: ten weeks, £14,000. The feasibility method is in the data you already own.

What it showed: one service line ran 19 per cent over on average, quoted from a template unchanged since 2019. Win rate on that line was 61 per cent, far above the 38 per cent average, which is exactly what you would expect from work that is too cheap. Two long-standing clients accounted for 44 per cent of all discount given.

What they did: repriced that line by 15 per cent, which is less than the overrun, deliberately. Win rate fell to 49 per cent, still above average. Net effect on existing volume: roughly £71,000 of additional annual margin.

No new customers, no new product, no new headcount.

The trap

The obvious conclusion from a pricing analysis is "put prices up", and applied bluntly it costs you work.

What the data actually supports is more specific: put prices up on the job types where win rate is well above your average, because that is the signal that you are underpriced relative to what the market will bear. Leave the lines where win rate is already marginal.

The same analysis usually also identifies work you should stop quoting for, which is the discipline behind productising expert knowledge, which is the harder decision and often the more valuable one.

Making it continuous

The single change worth making afterwards is recording the outcome of every quote in a consistent field, including the losses and why. Most businesses record wins properly and losses vaguely, which means half the dataset is missing.

Two fields added to something people already fill in, and in three years you have this analysis available on demand rather than as a project.

Key Takeaways

• Three joins produce everything: quoted versus won, quoted versus actual, and won versus margin.
• The four findings are consistent across sectors: one underpriced service line, concentrated discounting, flat win rate across a wide price band, and clusters of expensive-to-serve customers.
• A flat win rate across a price band is the clearest evidence that price is not the deciding factor you think it is.
• Raise prices on lines where win rate is well above average, not across the board. Blunt increases cost work.
• Record the outcome of every quote including losses. Two extra fields turns this from a project into a standing capability.

Frequently Asked Questions

We do not have actual cost data, only quotes. Is it still worth it?

Yes. Quoted versus won on its own tells you where your win rate is suspiciously high, which is the finding that most often leads to a price change. Actual cost makes it better and is not a precondition.

How far back should we go?

Three to five years for most businesses. Twelve years gives you a longer series and includes periods when the business, the market and your costs were materially different. Longer is not automatically better.

Our quotes are all bespoke. Does the grouping still work?

Usually, once you find the right categories. Bespoke work generally falls into five to ten recognisable types even when every quote is written from scratch. Ask the person who does the quoting what kinds of job there are and you will get the categories in five minutes.

Will this upset our sales team?

It can, particularly the flat-win-rate finding, because it contradicts a widely held belief that deals are lost on price. Share it as a question rather than a verdict, and let them interrogate the data. They usually find the explanation for the outliers faster than anyone else can.

---

Sitting on years of quotes you have never compared? Talk to Halo Technology Lab. Our strategy and scoping service starts with a single-year slice test so you know the answer is there before committing to the full extraction.