Welcome to The Ops Digest!
Each week, we break down one real operations problem and one AI build you can actually use. No hype. No grunt work left on your plate.
This week: the flip side of the last two issues. We’ve covered how accounts gradually leave. This one is about how the next good account starts showing itself months before your prospect list would ever surface it. Your best customers shared a fingerprint before they were big, and it’s sitting in your order history right now.
Here’s how it works.
Your Next Customer Looks Like Your Last One
Think about your best account. Not the biggest — the best. Strong margin, full basket, orders that don't need babysitting. Call it Halvorsen Fabrication.
Now ask an uncomfortable question: if Halvorsen walked in today as a stranger, would your prospecting process find it? Three years ago, Halvorsen was a 40-person shop that wasn't on anyone's target list. It came in as an emailed PDF quote request, got handled well, and grew into the account your quarterly review brags about.
Here's what almost nobody does next: go look at what Halvorsen's first 90 days actually looked like in your order data. Because the accounts that grew into your best customers almost always looked different from day one — and they looked different in ways your CRM never captured. The prospect list says “machinery, 25–100 employees, Midwest.” The order data says something much sharper: first order had eleven lines across four categories, the second order landed nine days later, and the buyer asked about two products you didn't quote.
One of those is a demographic. The other is a fingerprint.
ONE WINNER, ONE ALSO-RAN — FIRST 90 DAYS Halvorsen Fabrication, first 90 days: First order, 11 lines across 4 categories — cutting tools, abrasives, coolant, PPE. Second order 9 days later. By day 90: ordering twice a month, 6 categories, and a question about vendor-managed inventory. Meridian Supply Co., first 90 days: First order, 2 lines, both commodity items, clearly price-shopped. Second order 41 days later. By day 90: one more order, same 2 SKUs. Both showed up in the CRM the same way: “New account — machinery.” Same rep, same onboarding, same follow-up cadence. Three years later, one is your reference customer. The other still buys 2 SKUs when your price wins. The difference was visible in the first month. Nobody was looking. |
The reflex is to say the rep should have spotted it, which is unfair — again — because the tools reps are handed are built around company attributes, not buying behavior. Industry code, headcount, region. Those tell you who a company is. The fingerprint tells you how a company buys. And how your winners bought in their first 90 days is the most predictive prospect filter you own, because it's built from accounts that actually became profitable — not from a persona workshop.
You're about to scroll past this.
Which fits, because “profile your best customers” sounds like a strategy offsite, not a Tuesday.
So we made it a Tuesday. One page, ten minutes per account.
We turned the fingerprint into a worksheet: pull the five traits from your best accounts, score any prospect or first-time buyer against them, and know in one glance whether you're looking at a Halvorsen or a Meridian.
Why Your CRM's Industry Field Can't Do This
Start with why fit matters more than volume. Research from TOPO (now part of Gartner) found that organizations with a strong ideal customer profile achieve 68% higher account win rates. Not marginally better — 68%. And the teams getting that lift aren't defining “ideal” by firmographics alone; they're defining it by how their actual best customers behave.
Now the timing half. The Ehrenberg-Bass Institute's well-known finding — the 95-5 rule — says that at any given moment, roughly 95% of your potential buyers are not in market. Blanket outreach to a list of lookalike companies mostly lands on the 95%. What moves a lookalike from “fits the profile” to “call this week” is a trigger: a new facility, a job posting for a purchasing manager, a first-time quote request that arrives with a Halvorsen-shaped basket.
And here's the signal most teams never think to use, because it lives in operations rather than sales: your own customers' order patterns. When three of your customers in the same vertical spike on the same category in the same quarter, something is happening in that vertical — a project cycle, a regulation, a supply gap. The fourth company in that vertical isn't a cold call. It's a warm one that doesn't know you yet.
None of this requires buying intent data. The evidence lives in your ERP, the same place last issue's basket audit lived. You're just reading it in the other direction: instead of asking which accounts are quietly leaving, you're asking which strangers look like the accounts that quietly grew.
This Week, Try This: The Fingerprint Pull
The raw material is one export, same shape as last week's: order lines, not invoice totals. Customer, order date, SKU, product category or description, quantity, unit price, cost if you can get it — plus, if your system records it, how the order arrived and who sent it.
Pull it for two groups. Group A: your ten best accounts by margin — not revenue — covering their first 12 months with you. Group B: ten accounts that showed up around the same time and stalled. The contrast is the whole trick. Attach the file with the prompt below.
You are analyzing order-line data for a manufacturer or
distributor. The file contains the first 12 months of
order lines for two groups of customers. GROUP A became
high-margin, multi-category accounts. GROUP B stalled.
STEP 1 — FOR EACH ACCOUNT, COMPUTE ITS FIRST-90-DAY
FINGERPRINT:
- Lines and distinct categories on the first order
- Mix of specialty items vs contract/commodity items
- Days between first and second order
- Orders per month by month 3
- Order channel and buyer role, if the data includes them
STEP 2 — CONTRAST THE GROUPS:
- List the fingerprint traits where Group A and Group B
separate most cleanly, ranked by how well each trait
splits the groups
- Discard traits that do not separate the groups
STEP 3 — BUILD A LOOKALIKE SCORECARD:
- Turn the top five separating traits into a checklist a
rep can score any new prospect or first-time buyer
against, weighted by how strongly each trait separated
- State the score threshold above which an account
resembles Group A
RULES:
- These are signals, not destiny. Label them as such.
- Skip accounts with under 12 months of history.
- If a field is missing (channel, role, cost), say so.
Never infer it.
- Same-quarter comparisons only, to keep seasonality out.
What comes back is a scorecard built from your own winners: the five traits that separated the accounts that grew from the accounts that didn't, weighted and ready to use. Then put it to work in two places. Score every first-time buyer from the last 90 days — the Halvorsens currently sitting in your system labeled “new account” — and hand the top scorers to reps as growth calls, not courtesy calls. And keep it next to the quote inbox: when a stranger's first RFQ arrives looking like Group A's first orders, that quote gets your best turnaround time, not the standard one.
First run is under an hour. Most of that is the ERP export, again.
An Honest Take
Most prospect lists are sorted by company size, and I think that's the single laziest decision in B2B sales. Size predicts how much effort an account will take. It says almost nothing about whether they'll buy the way your best customers buy. Meanwhile the accounts most likely to become your next Halvorsen are already in your system — they showed up last quarter, ordered once, and got filed under “new account” with the same follow-up cadence as everyone else.
If I could change one habit, I'd make the fingerprint score part of new-account intake, the same way credit checks are. But I'd settle for something smaller: the next time someone says a prospect “looks like a good fit,” ask them — looks like it how? If the answer is industry and headcount, that's not a fit. That's a category.
The Bottom Line
Defense was the last two issues: catching the accounts that are quietly leaving. This is offense, and it runs on the same fuel. Your order history already knows what a great customer looks like at day one — the basket, the cadence, the questions they ask. Most teams never read it that way, so the next Halvorsen gets the same treatment as the next Meridian, and the difference only becomes obvious three years later in the quarterly review.
One export, one prompt, one scorecard. Somewhere in last quarter's new accounts there's a future top-ten customer currently getting your standard follow-up cadence. Find it before your competitor's rep does.
Spotting the lookalike is step one. Keeping them is friction removal.
The fastest way to lose a Halvorsen in its first 90 days is to fumble the orders: slow confirmations, re-keyed line items, a wrong SKU on the second PO.
Y Meadows automates order entry from inbox pickup to ERP posting: matching customers and SKUs, applying your pricing and credit rules, and getting the order into your system without the manual scramble.
So when your next best customer sends their first PO, the experience tells them they picked the right supplier. Want to see how it works on orders like yours?
