Category: ERP

Fourteen days. That is how long a 120-person building products manufacturer took to close the books every month. Not because the accounting team was slow — because they spent most of those days waiting on paper from other departments.

Nine months later, the close finishes in four days. The headcount did not change. Here is the difference, day by day.

What the old fourteen days looked like

• Days 1–2: chasing shipping logs and receiving paperwork from two plants.
• Days 3–5: reconciling inventory by hand, because the system count and the floor count told different stories.
• Days 6–8: reviewing invoices that had been entered late or entered twice.
• Days 9–13: accruals, reconciliations, and manager review in a stack of spreadsheets.
• Day 14: publish, and immediately start preparing for the next close.

What the new four days looks like

1. Two days before month end: bank reconciliations and standard accruals are prepared in advance. Nothing here depends on the final day’s activity.
2. Day 1, by 11 a.m.: all transactions are posted. Shipping and receiving have a hard daily cutoff, and it is enforced by the warehouse manager, not accounting.
3. Day 1, afternoon: the inventory snapshot reconciles automatically against cycle counts taken daily, so there is no month-end surprise.
4. Day 2: an exception report lands with nine items on it, not 400. Someone reviews nine things in an hour.
5. Day 3: department managers review and sign off in the system.
6. Day 4, by noon: close is published.

The two changes that made four days possible

The first was a daily cutoff. If a shipment is not entered by 4 p.m., it belongs to the next day. That one rule removed the entire “chasing paperwork” phase.

The second was moving from reconciliation to exception reporting. The old close reviewed everything, which meant people spent days confirming that nothing was wrong. The new close assumes the data is right and asks the system to show only what looks off.

Here is the part that surprised the CFO: the accuracy improved. Fewer late entries meant fewer corrections, and the audit trail got cleaner because every posting happened in the system instead of on a clipboard.

This kind of change is mostly process design, which is what separates useful ERP Consulting from a software install. The ERP Consulting team that helped here spent more time with the warehouse manager than with the accounting staff.

Tomorrow, ask one question: which single transaction type causes the most rework at close? Fix the cutoff rule for that one type next month and watch what happens to day three.

The proposal said $180,000. Sixteen months later, the controller of a 95-person specialty food distributor had spent $340,000. Nothing shady happened. Nobody was oversold. The gap was made of line items that were never in the quote because nobody thought to write them down.

Here is where the extra $160,000 went.

The costs that never make the proposal

• Data cleanup, $38,000: 14,000 item records with duplicated units of measure, plus vendor and customer cleanup. Nobody budgets for cleaning what you already own.
• Your own people’s time, $96,000: a project manager at 60%, an AP clerk at 40%, a sales admin at 30%, over ten months. This is real money even though no invoice arrives.
• Temporary help, $21,000: six weeks of extra hands after go-live, when everyone is slower than usual and orders still have to ship.
• Extra training, $18,000: the vendor included two days. Two days covered the menus, not the jobs.
• Reports built after go-live, $27,000: every manager wanted “the report we used to have” in the first 90 days.

The one line item nobody forecasts

The subscription went up at renewal — $10,000 more in year two — because headcount and transaction volume had grown. It was in the contract. Nobody had read that clause closely.

Add it up: $60,000 software, $70,000 implementation services, and $210,000 of everything else. Your vendor quote is roughly a third of the real number.

How to get a number you can actually defend

1. Ask each department head for hours per week they will spend on this, multiply by a loaded hourly rate, and put it in the budget as a line.
2. Add 25% of the software cost for data cleanup. If your data is older than eight years, add 35%.
3. Budget for one temporary person for the first six weeks after go-live.
4. Set aside 15% of the total for reports and small fixes in the first 90 days.
5. Read the renewal clause and write the year-three number into the business case.

This is where ERP Consulting earns its fee, and it is not the configuration. A good advisor tells you the $340,000 number before you sign, so the board approves the real figure instead of discovering it in month eleven.

Before your next steering meeting, take 30 minutes and multiply your team’s weekly hours by a loaded rate. You will probably find six figures of cost that has never appeared on a single page of your project plan.

Same software. Same 14-month timeline. Same regional implementation partner. Plant A in Ohio has 180 employees and closes its books in six days. Plant B in Tennessee has 165 employees and still takes 17. They went live eight months apart.

I sat in on both post-mortems. The difference had nothing to do with the software, the budget, or how hard people worked.

The difference was who sat in the steering meetings

Plant A put a line supervisor, a shipping lead, and an AP clerk in the room next to the directors. Those three people flagged 60 small problems before go-live, mostly around how work actually flowed on second shift.

Plant B’s steering team was four directors and a project manager. Every decision passed down through layers, and the people who had to live with the answers heard about them last.

Training on your own mess beats training on sample data

Plant B did a four-hour classroom session on a clean demo dataset. Everyone passed. Then they hit their own inventory, with its missing bins and odd units of measure, and froze.

Plant A trained three weeks before go-live in 90-minute sessions using their own live data and their own worst orders. Slower to schedule, far faster to stick.

• Train in the room where the work happens, not a conference room across the parking lot.
• Use ten real transactions from last month, including one ugly one.
• Keep sessions at 90 minutes and cap the group at eight people.
• Have each person complete one full cycle on their own before they leave.

The 30 days after go-live decide everything

Plant A stationed two floor-level fixers per shift for the first month — people who could unblock a picker in 30 seconds instead of filing a ticket. Plant B routed every issue to a help desk with a two-day response time.

By week three, Plant B’s crew had built a shadow spreadsheet system to get work out the door. Nobody killed it. It is still running today, which means the system of record is not actually the system of record.

That shadow spreadsheet, not the software, is what you are really paying to avoid. Good ERP Consulting focuses on the first 30 days as hard as the go-live date. The ERP Consulting contract usually ends at go-live, and that is exactly when the real work starts.

If you are mid-implementation right now, name two people per shift who can fix problems on the spot for the first month. Put their names on a wall. That single move is worth more than another week of configuration.

Before any data move, I ask one question: how many customers do you actually have? A 25-person industrial fastener distributor said “about 15,000.” The real number was 6,200. The rest were duplicates, dead accounts, and one salesperson’s eleven creative spellings of the same buyer.

Nobody did anything wrong. Every system collects duplicates the way a garage collects paint cans. The problem is what happens when you carry all of it into a new one.

Why 8,400 extra records is normal, not a scandal

Accounts get created at the counter when someone is in a hurry. A customer changes names after an acquisition and keeps the old number. A credit hold gets resolved by creating a fresh account instead of fixing the old one.

Move that mess as-is and your new system will simply be a faster way to be confused.

Seven questions to settle before you move a single record

1. Who owns each data set — customers, vendors, items, open orders — by name, not by department?
2. What makes two records “the same”? Match on tax ID or a normalized address, not on company name.
3. Which records will nobody miss? Define it plainly: no order in 36 months and no open balance.
4. What do you do with historical transactions when the account gets merged?
5. How will you handle the 200 records that are genuinely ambiguous, and who breaks the tie?
6. What is the cutoff date, and does everyone know that nothing gets entered after it?
7. How many times will you run a full test load before the real one? Three is the minimum.

The rule that saves the most pain

Freeze the data before you clean it. Pick a Friday, stop new account creation entirely, and clean what exists. Cleaning while people keep adding makes the job twice as long and never quite finished.

Second rule: the people who use the data clean the data. A 62-person distribution team I worked with assigned customer cleanup to the two order-entry clerks who knew every account by voice. They finished 6,200 accounts in nine working days.

An outside team can write the match rules and run the loads. That is legitimate ERP Consulting work. But the judgment calls — is this the same company, is this account worth keeping — belong to the people who talk to those customers every week. Every ERP Consulting project I have seen fail at migration failed on that handoff, not on the technology.

This week, export your customer list, sort by name, and count how many times the same company shows up. That number is your real project scope.

“We can get a quote out in two days,” the operations manager of a 40-person electrical supply distributor told me. “So why does cash always feel tight?”

We pulled the timeline on 60 orders that shipped in March. Quote to order: two days, exactly as promised. Then the trail went cold. Ship date to invoice date averaged 19 days.

Where the nineteen days actually went

• 3 days waiting for a signed order acknowledgment that sat in somebody’s inbox.
• 4 days in credit review, because every new-account check ran through one controller by hand.
• 6 days waiting on proof of delivery, which the carrier sent in a weekly batch instead of daily.
• 6 days in the billing queue, where invoices waited for a person to match them to a pick ticket.

At any given moment, $1,240,000 of shipped product had not been invoiced. That is roughly a month of payroll parked in a queue.

Errors made it worse. About 11% of invoices needed a credit memo and a reissue, which pushed the real collection date out even further.

Four changes, done in ninety days

1. Order acknowledgments moved to a click-to-confirm email link, timestamped automatically.
2. Credit checks under $10,000 got an automatic rule: net-30 if there is no past-due balance over 60 days. Only exceptions go to the controller.
3. Proof of delivery pulls daily from the carrier portal instead of waiting for the weekly file.
4. Invoices generate automatically from the ship confirmation at 4 p.m. daily. Billing only touches exceptions.

What showed up on the bank statement

Invoice lag fell from 19 days to 4. Days sales outstanding went from 52 to 41, releasing about $310,000 in working capital that had been sitting in paperwork.

Credit memos dropped from 11% of invoices to 3%, which quietly removed about 40 hours a month of rework for two people.

Here is the part worth sitting with. This company had brought in ERP Consulting two years earlier to speed up quoting — and quoting was never the problem. The executives kept measuring the step they were proud of and ignoring the four steps behind it.

That pattern shows up in most ERP Consulting work I have seen. The bottleneck is rarely where the complaint is. Before your next Monday meeting, take five orders that shipped last week and time every step from ship to cash. You will find your 19 days in about 20 minutes.

Tuesday, 6:40 a.m. A warehouse supervisor at a 62-person HVAC parts distributor pulls a bin for fan motors. The system says 140 on hand. He counts 89. That is roughly $9,300 of parts that cannot ship today, and three customers are already on the phone.

It was not one bin. Over the next week the team counted 4,800 items and landed at 78% accuracy. Read that again: for every five picks, about one was wrong.

Everyone blamed the warehouse crew. Wrong target.

These were not careless people. They had run the same racks for nine years on paper tickets and a lot of tribal knowledge. They were working around a process that made errors almost unavoidable.

The damage started at the receiving dock. Deliveries were entered into the system hours before anyone shelved them, so the on-hand number went up while the parts sat in a staging area nobody counted. Two large vendors sent packing slips that ran to a second page, and two different people would enter the same shipment twice.

Three changes that took accuracy from 78% to 96%

1. Receiving got one door and one rule: nothing gets entered until the pallet is physically on the shelf. Any delivery over $2,000 gets a second set of eyes.
2. Bin labels got a check digit, so A-14-3 and A-14-8 stop looking identical at a glance. Mis-picks on the top 200 items fell by more than half.
3. Cycle counting started on the 40 fastest movers — five items a day at 7 a.m., before the first truck arrives. Not a once-a-year weekend marathon.

The result only finance noticed

Ninety days in, accuracy sat at 96%. Emergency air freight dropped from 31 shipments a month to 6, taking about $4,100 a month off the freight line.

The bigger surprise was overtime. Receiving had been staying late to “catch up on paperwork.” Once entries happened at the shelf, the paperwork disappeared, along with 22 hours of weekly overtime.

Almost none of this came from changing software settings. That is the uncomfortable lesson from a lot of ERP Consulting projects: the system records whatever your process tells it to record, mistakes included.

And if you are paying an ERP Consulting team, watch what they do in week one. If they open a configuration screen before they walk your receiving dock, you are about to buy a very expensive report.

Tomorrow morning, pick ten items at random, count them, and compare to the system. Then write down where the number drifted. Do that for five days and the pattern will point at the one process worth fixing first.

Same company, same software, same industry. In 2019 they built a job costing setup with 400 cost categories, and it was dead within eight months. In 2022 they tried again with 40 categories, and it stuck. The difference wasn’t the tool. It was three decisions about people.

This is a 120-person specialty contractor doing $60 million a year with about 30 active jobs at any given time. The first attempt failed so completely that project managers were keeping private spreadsheets and emailing them to accounting every Friday.

Attempt one tracked everything and told nobody anything

The 2019 build had a category for every eventuality: travel, small tools, permits, disposal, equipment rental, sub-tier cleanup, and several dozen more. Entering a single purchase order took eleven fields.

• Project managers stopped coding costs because it was slow, so costs landed on the wrong job.
• Reports came out three weeks after month-end, which made them history instead of management.
• Nobody saw their own job’s numbers, because the whole report ran to 60 pages.

By month eight the reports existed and nobody opened them. The controller used a sentence that every ERP Consulting team hears at least once: “We have the data, we just can’t use it.”

The three decisions that made attempt two different

They rebuilt the same capability with one rule in mind — a project manager should be able to see a job’s health in under two minutes, on a phone, without asking anyone. Their IT lead still jokes that the biggest contribution was the phrase “delete that field.”

1. Forty cost categories, not four hundred. Anything unusual goes to “other” with a note, and the list grows by one only when the same note shows up five times.
2. Costs are coded by the person who spends the money, at the moment they spend it. No Friday catch-up. If a crew buys a permit, it’s coded from the truck.
3. Each project manager sees their jobs and only their jobs. Not a company-wide report. Four numbers: cost to date, percent complete, remaining budget, and hours over or under.

That was the whole redesign. Fewer fields, earlier coding, smaller view. It cost about a third of what the first attempt did, because most of that first attempt’s effort went into building reports nobody wanted.

What the project managers actually do with it now

The change shows up in how they talk. Instead of asking the controller how a job is doing, they know before the morning meeting. Two jobs were renegotiated mid-stream last year because the numbers surfaced at 40% complete rather than 90%.

According to the operations director, about $1.4 million in overruns were caught early enough to do something about. The 2019 version caught the same overruns three weeks too late to matter.

Before you redesign your job costing, ask three project managers to show you where they look for their numbers today. If the honest answer is “I call accounting,” you don’t have a reporting problem. You have a timing problem, and timing is fixable.

7:40 am. The dispatcher is on the phone with a technician sitting in a parking lot, waiting to find out whether the parts for his first job arrived. She doesn’t know. The warehouse doesn’t open for twenty minutes. Meanwhile three other techs are calling in with the same question.

That was a Tuesday at a 25-technician equipment service company running about 6,000 calls a year. By the end of the week we worked out that each tech was losing roughly 40 minutes a day to information that was late, wrong, or missing.

The schedule wasn’t the problem. The information was.

When we asked the service manager what he needed, he said “better routing software.” When we asked the technicians what they needed, they said something quite different.

• Which parts are on the truck right now, not which parts were loaded on Monday.
• Whether the customer’s equipment history shows this as a repeat failure.
• Whether the site needs a lift, a permit, or a two-person crew.
• Who to call if the gate is locked before 8 am.

Not one of those is a routing question. They’re all “what do I need to know before I walk in the door” questions, and every missing answer turns into a phone call, a trip back to the shop, or a second visit.

Forty minutes a day, in plain math

Twenty-five technicians, forty minutes each, 240 working days. That’s 4,000 hours of lost billable capacity a year, or roughly the output of two full-time technicians.

1. Put the job checklist on the tech’s phone the night before — parts, site requirements, equipment history, contact name, all in one place.
2. Require truck inventory to be scanned when it changes, not at the end of the day. The moment a part comes off the shelf, it’s recorded.
3. Flag repeat failures automatically. If the same serial number has been serviced three times in six months, the tech sees that before leaving, not after arriving.

Two of those three changes are process, not software. That’s typical. When ERP Consulting teams get called into a service business, the first useful question is usually “what doesn’t the tech know at 7 am?”

What customers noticed

First-time fix rate went from 68% to 81% over two quarters. Second trips dropped by nearly half, and that’s where the real money was hiding — an average second trip cost $190 in truck time and lost billing.

The service manager now reviews two numbers every Monday: first-time fix rate and repeat visits. It’s a habit that came out of a short ERP Consulting engagement and outlasted it by years.

The dispatcher got her mornings back too. Instead of fielding fifteen “do I have the part” calls, she spends the first hour on tomorrow’s schedule. That one shift is worth more than most software decisions.

Walk into your service operation tomorrow at 7:30 am and count how many calls come in before the first truck rolls. That number is your baseline, and it’s usually the fastest place to start.

A purchasing manager at a mechanical contractor pulled twelve months of invoices and did something none of her predecessors had bothered to do. She sorted them by supplier. The count came back at 214 suppliers for $6.8 million of annual spend, and only 31 of them had a contract.

She wasn’t surprised that people bought from lots of places. She was surprised by the pattern. The extra suppliers weren’t cheaper — on average they ran 14% more expensive than the ones already under agreement.

Nobody rebels. They just get stuck.

Maverick buying gets described as people ignoring the rules. In practice it’s people solving a problem at 6 am with no better option.

• The approved fastener supplier has a $250 minimum, and a tech needs $40 of bolts. So he drives to the hardware store.
• The parts counter closed at 5 pm, and the job ran until 9.
• The preferred glove vendor was out of stock, and the crew needed gloves that morning, not Thursday.

Every one of those purchases made sense in the moment. Multiply it across 40 trucks and 240 working days and you get $310,000 of spend that nobody negotiated, compared, or tracked.

Three questions before anyone adds a supplier

They didn’t ban new suppliers. They made the decision visible instead.

1. Is this need already covered by an existing supplier? If yes, call that supplier first, every time, even for small orders.
2. What’s the annual value of this category? Above $25,000 company-wide, it goes to a bid. Below that, it goes on a list reviewed quarterly.
3. Who benefits, and what does it cost them to comply? Fix the minimum order, the pickup hours, or the stock level so the approved path is also the easy path.

That third question is the one most teams skip, which is why their policies get quietly ignored. It’s the pattern ERP Consulting teams see over and over: tighten the supply base and you get compliance for a month. Fix the reason people were going around it and you get it for good.

What the first 90 days actually returned

They consolidated to 61 suppliers, renegotiated three categories, and kept a short approved list for emergencies. Hard savings in year one came to $340,000, mostly from rebates and better pricing on the big categories.

The technician side got easier too. The preferred supplier opened a pickup counter until 8 pm and dropped the minimum to $75. Nobody has to be told to use it, because it’s simply faster than driving across town.

The finance director now runs that one-page supplier list every quarter, a habit she picked up during an ERP Consulting review and kept long after the project closed.

Pull your own top 30 suppliers by invoice count this week. If the names near the bottom are ones your leadership wouldn’t recognize, you’ve found your next 90 days of work.

The call came at 7:20 am on a Tuesday. A co-packer had shipped 4,000 cases of a snack bar with the old recipe printed on the box and the new recipe inside. The new version contained a tree nut oil the old one didn’t. Retailers pulled it, and the recall, freight, credits, and overtime came to $260,000.

Here’s the part that stings. The ingredient change was approved two months earlier, by the right people, in writing. The decision was fine. The follow-through wasn’t.

Everything was approved and nothing was updated

When we walked the change backward, it had passed through five pairs of hands and stopped in three places.

• The recipe was updated in the kitchen binder but not in the packaged product’s component list.
• Packaging artwork was revised on a shared drive. The printer still had version six.
• The line supervisor knew about the change. The second-shift supervisor, who started the following Monday, did not.

Five people each knew one piece. Nobody knew the whole thing, and that gap is where the $260,000 lives.

When ERP Consulting was asked to look at the recall cost afterward, the finding was uncomfortable. The company had a documented approval process and no documented handoff process. Approvals were covered. Handoffs weren’t.

The change process they built afterward

They kept it deliberately boring. One form, one owner, one effective date, and a physical check at the first production run.

1. One owner per change. The product manager signs their name and is responsible for every downstream step, including the printer.
2. A single effective date with a countdown. That date is reviewed at the daily production meeting for the two weeks before it lands.
3. First-run verification by someone outside production. Quality stands at the end of the line for the first 200 units and signs off with a photo.
4. Old versions get destroyed, not shelved. If the old packaging is still in the building, someone will use it at 11 pm during a rush.

The number that got the board’s attention

The company runs about 40 product changes a year. At the old error rate, roughly one in ten caused a problem worth chasing. The new process caught 11 issues in its first year before the trucks left the yard.

The general manager put it better than any policy document could: a change isn’t done when it’s approved, it’s done when the first good case ships. That sentence, written on a card and taped above the line, did more than any dashboard.

If you make anything with a recipe or a drawing, grab the last five changes your team made. Ask one question about each: who signed off that the packaging or documentation matched the product? If the answer is “the same person who made the change,” you have a hole worth fixing this month.