A forecast is only as good as what it is built on.
Deciding what to buy and when is the decision that quietly costs the most: too little and you are out of stock on the thing that sells, too much and your cash is sitting on a shelf. There is a whole industry selling software for it. Before any of that, this page covers what Cin7 Core already does about it, where that stops, and why the first question is not which tool but whether the numbers underneath are worth forecasting from.
Four answers ship with the platform.
They are not one feature at four sizes. Each solves a different shape of problem, two are included in a standard subscription and two are separately subscribed, and a good number of businesses need only the first of them.
A minimum before reorder and a reorder quantity per product, set for each location rather than globally, with low stock reorder and backorder handling on top. It needs a supplier attached to the product record to do anything useful, which is the first place this quietly fails.
A report that projects sales from recent velocity and supplier lead times, then turns the answer into reorder suggestions you can raise as a purchase, a transfer or a production order. Safety stock is a toggle on it, and the two settings use genuinely different arithmetic.
A full supply run: demand from sales, supply from stock on hand, transfers, production, assembly and suppliers, with the suggested orders split, merged and confirmed from the run itself. A scheduler view puts the same supply and demand on a calendar and a Gantt chart.
Algorithmic sales forecasting out as far as 24 months, with overstock and stockout reports, smart reorder, a purchase budget planner and forecast accuracy reporting so the model can be tuned rather than trusted. Lead times can be read from your own history instead of typed in.
Where it stops, and it does stop.
Two of the four are add-ons. MRP is a subscription of its own and needs the minimum before reorder setting configured for each location before it produces anything. ForesightAI is a subscription of its own too, and it needs at least six months of sales history in Cin7 Core to run. Nine is the number we would plan around, because a forecast from a thin history is still a forecast and still looks confident.
Everything here reads Cin7 Core and only Cin7 Core. Demand you can see coming but have not entered as an order is not in the arithmetic: the retailer who has told you verbally, the promotion booked for next month, the container you know is late. Marketing spend and channel forecasts are outside it too, and no setting brings them in.
There are working limits worth knowing early. ForesightAI recalculates once a day, so a change on the warehouse floor can take up to 24 hours to appear on those screens. Only one MRP run can be active at a time, and once a run is completed it is read-only. Suggestions are also suggestions: nothing here places an order for you, and in our experience that is the right way round.
- ✓ Reorder points and the velocity report are included
- ✓ MRP and ForesightAI are separate subscriptions
- ✓ ForesightAI needs six months of history, and nine to be worth planning on
- ✓ Nothing outside Cin7 Core is part of the calculation
- ✓ ForesightAI refreshes once a day rather than live
- ✓ Suggestions are raised by a person, not placed automatically
Forecasting is a data problem before it is a software problem.
You would not put a second storey on foundations you had not checked. A forecast is the same thing: it takes your stock on hand, your costs, your lead times and your sales history at face value and does arithmetic on them. If those are wrong, the output is not roughly right. It is confidently wrong, presented in a chart, and someone spends real money on it.
This is the part no vendor demo covers, because the demo runs on their data rather than yours. It is also the part that decides whether any of this works. In our experience most of the value in a forecasting project is released before the forecasting tool is chosen, and a fair number of businesses that came to us for one found the reorder suggestions they already had started working once the inputs were right.
It is not a counsel of perfection. Nobody has flawless data and waiting for it is its own kind of failure. The test we would apply is narrower: is each of these good enough that a decision made on it would be a decision you would defend? Where the answer is no, that is the work, and it is usually cheaper than the subscription you were about to buy.
Where the list below reads as a long afternoon rather than a quick check, a system audit is the honest starting point. It finds which of these is actually the problem rather than treating all six at once.
- ✓ A forecast repeats your history, including the parts you would not repeat
- ✓ The arithmetic is public: velocity, lead time, safety stock, what is on order
- ✓ Wrong inputs produce a confident answer, not an obviously wrong one
- ✓ Most of the fixing is configuration and housekeeping, not software
- ✓ A tool bought to skip this step inherits the same numbers
If the number on the screen is not the number on the shelf, every suggestion downstream inherits the gap. Stocktakes, in-transit stock and the adjustments nobody wrote down are the usual culprits, and no forecast repairs them.
The reorder arithmetic multiplies lead time by velocity, so a lead time typed in once at go-live and never revisited moves the answer more than most people expect. Where the history is good enough, having the system calculate them is usually the better setting.
A purchase budget plan is only as good as what you believe a unit costs, landed. If freight, duty and handling were never allocated, the plan is arithmetic on the wrong figure and it looks exactly as convincing as one on the right figure.
Forecasting reads what is in Cin7 Core and nothing else. History from before go-live is not there unless it was deliberately brought in, and a promotional spike nobody flagged reads as ordinary demand and gets repeated next year.
Duplicated SKUs, a product replaced by its successor with no link between them, and families set up three different ways all split one product’s demand into several thin histories. Each of those histories then forecasts badly on its own.
Reorder settings resolve per location and per supplier, with the location-specific setting winning. If products have no supplier attached, or locations were added later and never configured, whole sections of the catalogue quietly produce no suggestion at all.
One name, and an honest answer about it.
Our other platform pages group this list by how each app reaches Cin7 Core. This page cannot, and the reason is the most useful thing on it: we have not verified the mechanism and we have not implemented it. So it is named, and that is where we stop. This section listed four tools until we read our own record back and found it said nothing about three of them.
StockTrim is the planning tool we are most often asked about next to Cin7 Core, and the one we have a relationship with. We are naming it because you would meet it anyway and would rather hear about it from someone with nothing to sell. What we are not doing is telling you how it connects, or how it behaves once connected, because we have not done that work and the honest version of this page says so.
- StockTrimDemand forecasting and replenishment, named pending a confirmed mechanismnot implemented by us, and not documented in our stack
Why this page is shorter than the vendors’ are.
Every one of the platforms we do work with has a page here describing what it connects to, how, and what breaks. This one has a line, and the difference is not an oversight. Our internal record of what our stack does is the same record this page reads from, and for StockTrim it holds a name and nothing more.
We could copy the vendor’s own description and it would read well. It would also be us repeating a claim we have not tested, on a page whose whole argument is that untested numbers are the problem. So no connection method, no feature comparison, no accuracy figures and no pricing. If you have found a figure for how accurate a forecasting tool is, ask who measured it and on whose catalogue.
What we would do instead is scope a specific tool against a specific need. That means writing down what you actually want the forecast to decide, checking whether Cin7 Core already answers it, and only then looking properly at whichever tool fits, including talking to the vendor with you. It is slower than a recommendation and considerably more use than a blind one.
- ✓ We name them because you would find them anyway
- ✓ We have no partnership, commission or target on any of them
- ✓ No connection method stated, because we have not verified one
- ✓ No accuracy claim, for these or for anything native
- ✓ A specific need scoped against a specific tool, not a blind recommendation
The Inventory Velocity Report and the rest of the built-in reports are where most of this reads from, and getting more out of them is often the cheaper half of the job. Reporting and BI
We make the inputs trustworthy, then set the logic up.
Cheapest answer first, as ever. Reorder settings that work cost nothing beyond the configuration; MRP and ForesightAI are subscriptions and projects, so it is worth being sure before you buy one.
Are the inputs worth forecasting from
Stock on hand, costs, lead times, the product list and the sales history behind it. This is fig. 02 done properly, and it is where most of the difference is made. It also tells you honestly whether a forecasting subscription would be money well spent yet.
Set the reorder logic up per location
Minimum before reorder, reorder quantities, suppliers attached to products, safety stock decided rather than defaulted, and locations configured so the suggestions come out for the whole catalogue instead of half of it.
Teach the run, not just the settings
Somebody in your team should be able to execute an MRP run, read what it suggests, know why it suggested it and change their mind about it. A planning setup only we understand is a dependency you did not ask for.
What we are not.
We do not forecast for clients, and we do not tell anyone what to buy. Nobody outside your business knows what you have heard from a buyer, what you are launching, or how much risk you are willing to carry on a container. Our work is making sure that when you or the system produce a number, the inputs behind it are clean. The decision stays yours, and it should.
We are not demand planners either. There are people whose whole trade is planning methodology, and where a business needs that we would rather say so and work alongside them than improvise it. What we know well is the system the planning has to run in.
And we are not your accountant. A purchase budget plan touches cash flow, stock valuation and what you can afford to commit, and those are questions for whoever signs off your books. We can say where each figure came from and how the system calculated it, which is genuinely useful, and we would stop there.
What people ask before starting.
Do we need a forecasting tool at all?
Often not. If your catalogue is manageable, your suppliers are steady and demand moves gently, reorder points per location and the velocity report cover it, and both are already included in what you pay.
The honest triggers for looking further are seasonality you cannot hold in your head, lead times long enough that a wrong call takes months to correct, several locations replenishing each other, or a purchasing budget that has to be planned rather than reacted to.
Do you recommend StockTrim?
Not blind. We have not implemented it, and we would not repeat a vendor’s description of its own product as though it were our finding. Anyone recommending a forecasting tool without asking about your catalogue first is guessing on your behalf.
What we would do is work out what the forecast has to decide, check whether Cin7 Core already answers it, and then look at whichever tool fits that need with you, vendor included. If we later do the work properly on one of them, this page changes.
How much history does ForesightAI need?
Six months in Cin7 Core is the hard minimum for it to run. Six to nine is the window where the output starts being worth something, and we would quote the conservative end of that rather than the minimum.
History from before go-live does not count unless it was deliberately brought in, which is one of the few arguments for migrating any historical data at all. If you went live three months ago, the answer is to wait rather than to buy.
Our reorder suggestions look wrong. Is that the software?
Usually not. The arithmetic is not a mystery: it multiplies sales velocity by lead time plus safety stock and subtracts what is available and what is on order. When the answer looks wrong, one of those four inputs normally is.
The common ones are a lead time that was a guess at go-live, products with no supplier attached, and stock on hand that has drifted from what is on the shelf. Those are all fixable, and none of them needs another subscription.
What does the work cost?
Our own time is £99 an hour, quoted as estimates by phase rather than a fixed scope. Setting reorder logic up on a tidy system is a short piece of work. Where the inputs need repairing first, the tidying is the larger half and we would want to look before giving you a range.
The add-ons are Cin7’s to price and we would not quote them, and StockTrim is the vendor’s to price. We take no commission on any of it.
Can it plan production as well as purchasing?
Yes, and that is one of the better reasons to look at MRP rather than reorder points. A run can suggest a purchase, a transfer between locations or a production order, which matters when the same demand could be met three different ways.
Planning production properly needs the manufacturing side configured as well, with a bill of materials on the products involved, and the usual approach is two passes: the finished goods first, then the components. That is a scoping conversation rather than a setting.
What we have written about costs, stock and running the operation.
Tell us what you keep getting wrong.
The line that keeps running out, or the pallet that has not moved in a year. Book a call and we will tell you whether Cin7 Core already has the answer, whether the inputs need work first, or whether this is genuinely a job for something on top.
please check your timezone and am/pm when booking
What to expect
- ✓ A straight conversation, no scripted sales pitch
- ✓ The included answer first, where there is one
- ✓ Estimates by phase before you commit to anything
- ✓ Plain speaking about what we have and have not done before