Author: Next

Every company says the same thing before migration: “Our data is fine.” Then you open the customer file and find 4,200 rows for what turns out to be 1,900 real customers, three versions of the same address, and a column labelled “Contact 2” that eight different people used for eight different purposes.

Migration is where good projects are won or lost, and it happens on one weekend. Here is what that weekend looks like when it goes well, based on a 180-person distribution business we moved off eleven years of spreadsheets.

Six weeks before: the cleanup nobody enjoys

Rule one is that you never migrate everything. You migrate what is still alive.

• Customers: only those with activity in the last 18 months, duplicates merged by phone number and tax ID
• Items: anything not sold or purchased in two years gets archived, not carried forward
• Open documents only — unpaid invoices, live purchase orders, current stock. Historical invoices stay in the old system as read-only PDFs
• One owner per data set, named in writing, with a deadline. “The team” is not an owner.

This phase caught 61 duplicate suppliers at the distribution business. One of them had been paid twice in the same week eleven months earlier because two departments each had their own version.

The weekend itself, hour by hour

1. Friday 6 p.m. — trading stops, final stock count begins, old system set to read-only
2. Saturday morning — data loaded and checked against control totals: total receivables, total payables, total stock value
3. Saturday afternoon — ten real transactions run end to end by actual users: a quote, an order, a delivery, an invoice, a return, a payment
4. Sunday — corrections made, control totals re-checked, first users trained on the exact tasks they will do Monday morning
5. Monday 8 a.m. — live. A support person sits beside the order desk for the entire day

The control totals are the part people skip. If your receivables are $1,412,880 in the old system and $1,412,880 in the new one, you can sleep. If they differ by $3,400, you find out on Saturday, not on the tenth of next month.

What to do for the first thirty days

Run both systems for two weeks, but only for reporting — never for entry. Double entry is what breaks migrations, because sooner or later someone enters the same invoice twice.

Expect your first month-end close to take longer than usual. Ours took nine days instead of five. The second month took six, and by month three it was four. Plan for that dip and tell your finance team in advance. A slow first close is normal; a slow fourth close means something was left undone. If you are planning a migration, write your control totals down before the weekend arrives. That single sheet of paper is the difference between a calm Monday and a very long month.

Professional services firms rarely lose money in dramatic ways. They lose it in fifteen-minute chunks that nobody bills, and in scope that everybody noticed but nobody wrote down.

This particular firm — 60 people, engineering consultancy, about $9 million in annual revenue — was profitable on paper but permanently short of cash. We spent a week inside their project records. The pattern was almost boring in how consistent it was.

Before: what the numbers actually looked like

• Timesheets submitted on Friday for the week, from memory, with an average of 6.5 hours recorded against a 9-hour day
• Fixed-fee projects with no tracking of hours consumed, so nobody knew a project was overspent until it closed
• Client change requests handled by email, absorbed silently, and never converted into a variation
• Retainer clients billed the same amount for three years while the work behind them roughly doubled

When we added it up, roughly 9% of billable capacity was evaporating each month. On their revenue, that is about $22,000 a month that was worked but never invoiced.

After: four changes, no new headcount

The ERPNext Consulting engagement focused on making the honest path the easy path. Nothing here required a policy memo longer than one page.

1. Timesheets entered daily at the end of the working day, on a phone, taking under two minutes
2. Every project shows a simple bar: hours used versus hours sold, visible to the project lead from day one
3. Change requests are logged as a date, a description, and an estimated hours figure before any work starts
4. Retainer agreements reviewed every quarter against actual hours consumed in the previous one

The resistance came from one place: senior consultants who felt daily time entry was surveillance. We answered it with data. Once the firm showed a project that had quietly consumed 340 hours against a 220-hour fee, the argument ended on its own.

Three months in

Recorded billable hours rose 11%. Two retainer clients agreed to increased fees. Six change requests that would previously have been absorbed turned into paid variations worth $41,000. Net effect: cash flow turned positive for the first time in two years.

The uncomfortable truth is that most service firms don’t have a pricing problem, they have a recording problem. Start with one project this week. Write down every change request, no matter how small, and count the hours at the end. The number will tell you whether you need a new strategy or just a better habit.

There is a moment every multi-store retailer recognises. Store A is out of a bestseller and losing sales. Store B has eleven of them sitting on a shelf. Someone says, “Just transfer them.” Three weeks later, nobody knows where eight of those eleven units went.

I have watched this play out across chains from four stores to sixty. The transfer itself is never the problem. The assumptions around it are.

Myth one: transfers solve shortages

Transfers move a problem, they don’t fix it. If store A is short every week and store B is long every week, you have a buying or allocation problem, not a logistics one. In one 14-store chain we reviewed, 62% of all stock transfers in a quarter were moving items between stores that had both been over-ordered in the same original purchase. Fixing the allocation logic cut transfers by 40% and freed up a full day a week of staff time.

Before you approve a transfer, ask a blunt question: if this item keeps needing to move, why is it being bought that way in the first place?

Myth two: in-transit stock doesn’t matter

It matters enormously. A van breaks down, a driver forgets the manifest, a store manager signs for nine boxes instead of ten. If your system only removes stock at the sending end and adds it at the receiving end, you have created an invisible gap where shrinkage lives.

• Dispatch is scanned item by item, not counted as boxes on a clipboard
• Receipt is scanned at the destination before the driver leaves the building
• Any transfer open more than 48 hours appears on a daily exception list with a named owner
• Monthly, someone reconciles transfer variance by store and by driver

That last bullet sounds bureaucratic. It isn’t. Two stores accounting for most of the variance usually points to one process or one person, and you can only see it if you look.

Myth three: approvals slow everything down

Uncontrolled transfers are what slow you down, because they create a reconciliation job for somebody at month end. A good ERPNext Consulting design puts a value threshold in place and lets everything below it move instantly.

Try this next week: pull your transfer history for the last 90 days, sort by value, and find the level below which 80% of transfers sit. Set the approval threshold there. Low-value moves go free, high-value moves get a check, and your finance team stops spending the first three days of every month hunting for missing cartons.

“Same customers. Same prices. Same machines. Last month we made money, this month we didn’t. Explain that.” The owner of a 140-person metal fabricator was not angry at me. He was angry at his own numbers.

We pulled twelve months of job records and found the answer in about four hours. It was not a pricing problem. It was a recording problem, and it had been quietly costing him margin for two years.

The four silent margin leaks

Every job in that shop looked profitable on paper. The paper was wrong in four predictable ways.

• The bill of materials still listed 2.0 mm sheet, but purchasing had switched to 1.8 mm eleven months earlier — the costing never followed
• Scrap on the press was thrown in the bin and never entered, so materials were always “on hand” in the system and losses never hit the job
• Machine time was charged at a 2019 rate that assumed a $22 hourly cost; the real number was closer to $31
• Rework jobs were re-opened under the original work order, which made the original job look fine and hid the extra 60 hours

Add those four together across 1,400 jobs a year and you get a shop that thinks it’s earning 18% gross margin when it’s earning 13.5%.

What changed, and how fast

We didn’t buy machines or hire anyone. We made four small rules stick.

1. Scrap gets a reason code and a quantity at the machine, entered by the operator, not the supervisor at day end
2. Any change to supplier material thickness triggers a bill of materials review within five working days
3. Machine hourly rates are recalculated every quarter from actual power, depreciation, and labour costs
4. Rework always opens a new work order linked to the original, so the cost lands somewhere visible

The ERPNext Consulting work here was mostly about making these rules the path of least resistance. If entering scrap takes 40 seconds, it happens. If it takes four minutes, it doesn’t.

The Monday report that ended the argument

Every Monday at 9 a.m. the owner now sees a one-page variance report: standard cost versus actual cost for every job closed in the previous week, ranked by dollar gap. Top five get a conversation. The rest get ignored.

Within one quarter, average gross margin climbed 4.5 points — roughly $380,000 a year on their volume. Nothing was sold at a higher price. They simply stopped shipping profitable quotes and losing money on the shop floor. If your monthly margin bounces around with no explanation, the explanation already exists in your job records. You just haven’t given anyone a reason to write it down.

Ask a business owner what an ERP project costs and you’ll get a licence number. That number is almost never the problem. The problem is everything that happens after the go-live date slips.

I reviewed a 250-person building products company that budgeted $120,000 over seven months. It took fourteen months and $310,000. Nobody stole anything. The money leaked out of ordinary places.

The hidden invoice nobody shows you at the start

• Nine months of double entry — two finance staff at 12 hours a week each, about $46,000 in payroll
• Overtime in the warehouse during three failed cutover attempts, $18,500
• Two experienced order desk staff who quit and were replaced through recruiters, $27,000
• Repeat stock counts and manual reconciliations, 400 hours across teams, roughly $21,000
• Orders shipped late to four major accounts, two of which reduced volume the following year

The licences were $38,000 of the total. Everything else was friction. That ratio is normal, and it is the reason scoping matters more than features.

Four scoping rules that keep projects boring

Boring is what you want. The engagements that go well share the same four habits, and an experienced ERPNext Consulting partner will insist on all of them before signing anything.

1. Phase one covers the three transactions that touch cash — quote to order, order to invoice, purchase to payment — nothing else
2. Every phase has a written finish line, a named business owner, and a date that does not move because of a new idea
3. Data rules are agreed before migration starts: which customer duplicates get merged, which items get retired, how far back history goes
4. The acceptance test is written by the users who will live in the system, not by the project team

Notice what’s missing from that list: reports, dashboards, custom screens, nice-to-haves. All of it can come in phase three, when people already trust the basics.

The question that predicts failure before it happens

Ask the project sponsor this: “Who is losing two hours a day for the next ten weeks, by name?” If the answer is vague, the project will slip. If the answer is three specific people with their calendars already blocked, you will go live on time.

One more thing worth doing: put a real number on your internal labour. Every hour of double entry during a rollout costs you twice — once in wages, once in the errors it creates. Track it weekly. When the sponsor sees a $4,000 weekly bleed, scope conversations get very short, very fast.

The warehouse manager called me on a Thursday afternoon. “We just counted the fasteners aisle. The system says 41,000 pieces. We physically have 26,000.” That gap had been sitting there for eleven months, and nobody could tell which customer orders it had already broken.

This is a 90-person industrial fastener distributor doing roughly $34 million a year. They already had software. They already had barcode scanners. What they didn’t have was a single reason for the sales team to trust the number on the screen — so sales kept their own spreadsheet, and that spreadsheet quietly disagreed with the warehouse.

Where the missing 15,000 pieces actually went

We spent two days in the warehouse instead of the conference room. Four things were eating the stock, and none of them were software failures.

• Receiving was recorded at the end of the week, so goods sat on the dock for up to five days as invisible stock
• Sales ordered in pieces, pickers picked in boxes of 500, and the conversion was typed by hand from memory
• Customer returns were put back on the shelf immediately but only entered into the system when someone remembered
• Free samples for trade shows left the building with a handshake and a note on a whiteboard

Once you write those four down, the fix stops being about technology. It becomes about who is allowed to move stock and when they must record it.

What the ERPNext Consulting team changed in six weeks

We didn’t rebuild everything. We touched three workflows, then watched them for a month.

1. Receiving confirms the count at the dock on a tablet before the truck leaves — no paperwork, no back-dating
2. Unit conversions are locked into the item record, so picking in boxes automatically writes off the right number of pieces
3. Returns require a restock scan before the credit note is issued, which killed the “shelf first, system later” habit
4. A 15-minute exception report lands at 8 a.m. showing any item with negative stock or no movement in 90 days

Week three was ugly. The exception report showed 2,100 items with mismatches, and the warehouse team wanted it switched off. We kept it on and made clearing it a daily 15-minute routine instead of a quarterly crisis.

The result, nine months later

Stock accuracy on A-class items hit 97.4%, up from 81%. Monthly backorders dropped from about 340 lines to under 60. The dead stock report surfaced $180,000 of material nobody had ordered against in a year, and selling that off released roughly $95,000 in cash.

The lesson I repeat to every distributor: your inventory accuracy is a behaviour problem that software can only expose, not fix. Before you spend anything on ERPNext Consulting, walk the aisle at 4 p.m. on a Thursday and count something. Then ask three people what the number should be.

The partner at a 40-person engineering services firm told me her utilisation was around 78%. She was confident. She’d been tracking it for years. Then we asked the engineers to log their time the same day they did the work.

Utilisation came out at 61%. The difference, over a year, was roughly 2,300 hours — work that had been done but never billed, or billed under a fixed fee that should never have been fixed.

The leak nobody wanted to look at

Time was being recorded on Friday afternoons, from memory, for the whole week. That’s not dishonesty. It’s just how humans work. People remember meetings and big tasks, and they forget the two-hour troubleshooting call on Tuesday that nobody scheduled.

Worse, those forgotten hours were invisible to project managers. So a project that was running 30 hours over budget looked fine until the invoice went out and the margin had already evaporated.

What actually changed

Three things, and the first one is the one people resist.

1. Time entered daily, in three minutes or less. One screen, project code, hours, done. No descriptions required for routine work.
2. Project budgets visible to the people doing the work, not just to the partners. If you’re at 85% of budget with 40% of the job left, you should know that on Tuesday.
3. Any work outside the original scope gets flagged in the moment and priced before it becomes free advice.

The daily entry rule lasted, but only after the firm dropped the requirement to write a description for every line. Cutting the form from nine fields to three raised compliance from 40% to 96% in six weeks.

The uncomfortable conversation

Once change requests were being recorded, the firm discovered it had been absorbing an average of 14 extra hours per client project, worth about £90,000 a year across the business.

Some clients pushed back. Most didn’t. The ones who did were mostly relieved to have clarity about what was included and what wasn’t — the fog had been costing them too.

Within a year, revenue per engineer rose 19% without a single rate increase. The work was always there. It just wasn’t being seen, and unseen work never gets invoiced.

An ERPNext Consulting engagement at this stage is less about systems and more about visibility: who sees what, and when. That’s what turned this firm around.

Try one thing next week. Ask your team to log their time the same day, even if it’s rough. Compare that week’s total against how you’d normally record it. The gap between those two numbers is the money you’re already leaving on the table.

The order was 40% bigger than anything this customer had ever placed. Everyone was delighted. Sales celebrated. Nobody checked the credit limit, because the customer had been reliable for six years.

Ninety days later, £240,000 was outstanding, the customer had stopped answering the phone, and a 35-person wholesaler was having a very uncomfortable conversation with its bank.

It wasn’t one bad decision — it was a missing step

Credit limits existed on paper. They just weren’t enforced anywhere. When a customer exceeded their limit, the system allowed it and sent a polite report to finance at month end, by which point the goods had long since shipped.

There was also no rule about larger-than-usual orders. A customer ordering three times their normal volume is a signal worth investigating — a new contract, a cash flow problem, or a business about to fold. All three look identical at the order screen.

The finance manager described her job at the time as “writing letters nobody answers”. That’s not credit control. That’s archaeology.

What they put in place

Six changes, in this order. None of them were expensive, and all of them annoyed at least one person for the first month.

1. A hard credit limit that stops an order at the point of entry, not after shipping
2. An automatic hold when a customer reaches 80% of their limit, so there’s a conversation before the wall
3. A weekly 20-minute aged debt review with one salesperson in the room, because collections need sales pressure, not just finance letters
4. Any new customer’s first three orders over £10,000 require a 30% deposit
5. Any order more than double a customer’s largest previous order gets reviewed before confirmation
6. A rule that sales commission is paid on cash received, not invoiced

That last one caused the loudest argument and produced the fastest change. Once commission depended on money in the bank, salespeople started making collection calls before finance had to.

Eighteen months later

Days sales outstanding fell from 68 to 41. Bad debt dropped to under £15,000 for the year. And the company took on four new large customers without any drama, because the process told them how to structure each one.

The sales director, who fought the deposit rule hardest, now uses it as a filter for customers worth having.

An ERPNext Consulting review later confirmed the pattern: the tools mattered less than the rule about commission. Incentives shape behaviour faster than policies ever will.

Open your aged debt report today and look at your five largest overdue accounts. If any of them can still place a new order right now without anyone needing to approve it, that’s the gap to close first.

Every Friday at 2pm, a 70-person metal fabricator held a meeting to decide what the factory would build the following week. It ran two hours. It usually ended with someone saying, “we’ll sort the rest out Monday.”

Which meant Monday started with another meeting, and the first shift often stood around waiting for instructions. The production manager estimated they lost four to six hours of machine time every Monday. That’s roughly 250 hours a year on the biggest machine alone.

Why the Friday meeting kept failing

Three things kept breaking it. First, nobody could see real capacity — the schedule lived in three spreadsheets and a whiteboard, and the whiteboard won arguments. Second, rush orders arrived constantly and jumped the queue without anyone removing what they displaced.

Third, machine breakdowns were handled by memory. When the press went down for a day, everyone improvised, and the plan was quietly abandoned by Wednesday.

None of that is a software problem, by the way. It’s a visibility problem with a software-shaped solution.

What replaced the two-hour meeting

The ERPNext Consulting team did something simple first: they built a single view showing every open job, the hours it needed per machine, and the delivery date. It took three weeks to get reliable, mostly because the shop floor data was being entered a day late.

Then came the rule that changed behaviour: a rush order must name the job it displaces. If a salesperson wants to jump the queue, they pick what gets pushed. No picking, no jumping.

Within a month, rush orders dropped by half. Not because customers changed — because salespeople started asking whether the rush was worth pushing out a delivery they’d already promised.

• Job status updated by the machine operator at start and finish, not at end of shift
• Downtime logged with a reason code, three taps on a tablet
• A rolling six-day schedule instead of a fixed weekly plan, so Tuesday surprises don’t wreck the whole week
• Friday meeting shortened to 30 minutes, and only for exceptions

The numbers after four months

On-time delivery went from 71% to 92%. Overtime dropped around 30%. Work-in-progress value fell by roughly £95,000 because jobs stopped sitting half-finished waiting for a decision.

The most surprising result was cultural. Two machine operators started suggesting schedule changes, because for the first time they could see why the sequence was what it was.

If your planning meeting runs longer than an hour, you’re doing scheduling work inside a meeting that should have been done before anyone walked in. Track how many hours you lose every Monday morning for two weeks. That number will tell you whether this is worth your attention.

Most managers think the hard part of running two warehouses is the technology. It isn’t. The hard part is deciding who owns stock when it’s sitting on a truck between them.

I’ve watched a 45-person wholesaler discover £52,000 of inventory that had been “in transit” for more than 90 days. Nobody stole it. Nobody lost it. It just fell into a gap where both sites assumed the other one was counting it.

Mistake one: assuming one system means one truth

Putting both sites on the same system doesn’t end the arguments. It just gives both sides a shared screen to argue about. If Site A thinks it sent 40 units and Site B thinks it received 36, one system doesn’t tell you who’s right — it only tells you they disagree faster.

What fixes it is a rule about timing. Every transfer gets confirmed at dispatch and again at receipt, and any gap over 24 hours raises a flag the same day, not at month end.

Mistake two: treating transfers as paperwork

Transfers feel like admin. In practice, they’re where money quietly disappears. A truck that sits unloaded over a weekend, a delivery note signed but never entered, a pallet counted at the wrong site — each one is small. Together they add up fast.

• Every transfer gets a same-day dispatch confirmation, no exceptions
• Receiving has 24 hours to confirm or the transfer shows as outstanding
• Anything outstanding over three days goes on a Friday list with a name next to it

Mistake three: running both sites the same way

The busier site and the quieter site need different reorder points. That sounds obvious, but most companies copy the same settings across both because it’s easier than thinking about it.

One site might serve walk-in trade with unpredictable daily demand. The other might feed a handful of large accounts on predictable schedules. Same rules, different reality.

When the wholesaler in question set separate reorder points and separate counting frequencies per site, their total inventory value dropped 14% in five months while service levels stayed flat. They were simply holding the wrong stock in the wrong place.

A separate ERPNext Consulting review of the same business found the biggest single improvement came from something unglamorous: a clear rule that stock in transit belongs to the sending site until the receiving site confirms it. One sentence, agreed once, ended months of finger-pointing.

Before your next stock count, pull a list of everything currently in transit and how long it’s been moving. You’ll probably find a few items older than a week. That list tells you exactly where to start.