1 · Why we did this
Can Birkdale's Data Guide Pricing At The Moment A Quote Is Typed?
Phase 0 answered three questions: can we measure quote outcomes, is margin leaking, and should Birkdale build. All data below is Birkdale's own, read from BC without writing anything.
2 · The real funnel
Two Thirds Of "Quotes" Are Real. About A Third Of Those Win.
We classified every quote by what it actually was. Old quote numbers betray housekeeping (a "quote" archived in January carrying a number issued years earlier is a clean-up, not a customer), and desk behaviour separates price checks and samples paperwork from genuine selling. Bars show what remains at each step:
True win rate: approx. 30-35%. Of that, 23.8% is directly provable from BC's own stamp. The rest are orders the customer accepted but staff re-typed as fresh orders. We found those by matching customer, product and timing, then measured how often those customers were buying the same items anyway and subtracted it, so only genuine wins count.
1 in 4 wins loses its paper trail through that re-typing. That is why the tool must record outcomes automatically at order entry: a pricing model trained on broken labels learns the wrong lessons.
3 · The governance finding
Prices Are Saved Everywhere. Governance Covers Only A Slice.
BC holds 932,484 saved-price records across two estates, yet most trade routes through neither:
Price-Group Lists
records across 214 group price lists. Where a customer is in a group, it works: prices cluster far tighter. But 85% of the high-volume trade we tested has no group assigned at all.
Customer-Specific Prices
one-off negotiated prices accumulating since 2021, across 5,127 customers. This is the estate the build rationalises. (Plus 2,513 campaign and all-customer records, which completes the 932,484.)
Same customer, flat. Different customers, 20-40% apart. A customer's own prices barely move from order to order. Different customers pay 20-40% apart for the identical product, comparing like for like (same product, pack size and customer tier). Those price levels appear inherited, not actively set.
4 · Saved prices, the March problem
Three Quarters Of Saved Prices Are Dead Weight, And They Blunt Every Price Increase
Why increases do not stick: the April-July increase re-issued approx. 304,000 records wholesale, and the record timestamps show no per-record review stage. The estate rolls forward, and the leakage rolls forward with it.
5 · Below cost and free of charge
Two Separate Leaks: One Is Error, One Is Unowned Policy
Charged Below Cost
885 lines observed over the 12 months, priced above zero but below the cost BC held that day. Pricing error; the corridor floor prevents it.
Given Away Free Via Quotes
Real product cost on £0 quote lines. A mix of legitimate samples, replacements and unpoliced giveaway; it needs categorising. This is spend made visible, not savings assumed.
Trade customers only, measured against the cost held on the day each quote was raised, so there is no hindsight in the measurement.
6 · The size of the prize
Sensible Build Case: £70-90k A Year, Before The Extras
Measured bottom-up from Birkdale's own quote channel, on one £0-150k scale:
These three are separate and are not added together: the corridor figure is recoverable margin, below-cost is error prevented, and FOC is spend exposed for a policy decision.
Context: a pricing programme of this kind is normally worth 1-2.5% of revenue, which on Birkdale's £35.1M is £351-877k a year. Our own measured maximums sit above the top of that range, so the 1-2.5% framing is the conservative, externally quotable one. The quote channel also leaks proportionally more than the rest of the book, which is why a quoting tool is the right first fix.
7 · What it becomes
The Corridor: Try It Right Here
A real archived quote line (customer anonymised) inside real 12-month price bands. Click the button, then tick Manager view to see the governed floor:
The design principles behind it (Mark's methodology, our build):
- Net pricesReal prices, not list-minus-discount theatre
- Three customer tiersPartners, Foundation, Explorers, managed differently
- Prices with commitmentsSaved prices expire unless volume justifies them
- Win-rate guidance"At this price you win 7 in 10", never "how low can you go"
- Customer signals"Always buys when quoted" means edge toward stretch
- Outcomes capturedAutomatically at order entry, closing the 1-in-4 gap
8 · How it all joins up
One Pricing Brain, Every Quoting Doorway
John's ask: so many moving parts, make sure we complete the full circle. This is the circle. One corridor service holds the pricing intelligence; every place Birkdale quotes plugs into it, so no two systems can ever give a customer a different price:
- Fencing Quote BuilderTrade quotes, already live; corridors appear inside it
- Consumer Fence BuilderThe customer-facing app, same price brain behind it
- CS deskPhone, SMS and web-chat price checks, recorded instead of invisible
- Sales order automationUnpriced emailed orders get suggested prices in the existing review portal
- HubSpot / BC quotesThe rep's quote screen, corridor shown at the moment of typing
- Price increasesMarch and every one after, applied through the same governed estate
This does not mean rebuilding any of those tools; each keeps its front end and plugs into the one service. The bin-store builder and future tools join the same way. The list-versus-net price architecture question is the design decision for the kick-off session, with Mark in the room.
9 · Recommendation
GO: Guided Pricing With Guardrails, Not Autopilot
- Decisions requested: approve the build; confirm who owns price architecture; confirm who owns free-of-charge policy.
- First entry point: the rep quote screen and the Fencing Quote Builder, where a live, committed user group already works daily; CS desk and sales-order suggestions follow.
- Success measures, from today's baselines: corridor adherence (new); below-cost lines (0.8% of lines today); outcome capture (76% today); stale saved prices (approx. 19,400 today, trending to near zero).
- Guardrails and hygiene: the model is trained on Birkdale's data and serves Birkdale only, never resold or pooled; competitor prices are display-only reference with source and date, never an input to the model; floors sit behind manager approval; data can move to a Birkdale-owned database if preferred.
- Open items, stated plainly: a finance spot-check of BC unit costs on approx. 10 items (a formality given the quote-time cost basis, but it should be done); the carriage side sits with the separate margin work and needs the delivery zone rate card from finance; the "average Irish order is 34% carriage" line heard in July remains unverified and should not be repeated until checked.
Method note: trade customers only; costs as held on each quote's own date; wins counted from BC's own quote stamp, plus accepted-but-retyped orders matched on customer, product and timing with coincidental reordering subtracted; every headline stated as a range. Full workings in the findings document.