Cin7 ForesightAI is worth paying for in a small manufacturing business when you carry roughly 100 or more active SKUs, have at least six months of sales history in Cin7 Core, and lose real margin to stock-outs or cash to overstock. Below about 50 steady-selling SKUs, a spreadsheet forecast usually does the job for less.
Cin7's pricing page shows no price for the add-on, so the useful number is the break-even price: the most you can pay before the saving disappears. Work that out first, then get the price from Cin7 or from the add-on screen in your subscription settings. If the quote is below your break-even by a good margin, run the trial. If it is above, you have saved yourself a month of set-up.
What you would actually be buying
As of November 2026, Cin7's pricing page lists Core at $349, $599 and $1,199 a month (Standard, Pro, Advanced) for new customers. ForesightAI sits on top as an add-on for any of those plans, and the page's FAQ says only that optional add-ons such as advanced forecasting cost extra. You add it from the user menu under My subscription > Update subscription. Its help centre says ForesightAI is now built into Core: forecasts, analytics and settings are inside the product rather than a separate app, and it recalculates once a day.
From the help pages, the pieces that matter to a maker are these:
- Sales forecast graph at Purchases > Smart Reorder > Sales forecast. It shows the forecast alongside sales history, safety stock and an inventory forecast.
- Forecast adjustments: Change by % (say +15% during a sale), Set % trend (a change that compounds each month), or Change quantity, where each month's figure is either added to the forecast or set as a total that holds whatever the model later says. This is where a human overrides the model for a promotion or a lost customer.
- Product strategy, chosen at Settings > ForesightAI settings: Gross margin, Revenue, Sales frequency or Sales quantity. It steers how products are prioritised.
- Settings by supplier, location and product, including lead times, which the forecast needs to be right.
- Winners & Losers, a view under Sales of products moving up or down.
- Assembly settings on locations: mark a location as Assembly and, according to the help pages, ForesightAI suggests production jobs, raw materials and finished-goods transfers there, using an assembly lead time and assembly period for each bill of materials. A co-manufacturing option covers outside assemblers.
- Smart Reorder methods per product: Automatic, Use sales forecast, Use minimum inventory, or Use customer orders, which Cin7 describes as best for make-to-order or irregular products.
Two preconditions come with it: the add-on must be on your subscription, and ForesightAI needs at least six months of sales history in Cin7 Core. A two-week free trial is mentioned on the set-up page. I could not find any published accuracy figure, so I make none; the point of this tutorial is to let you measure your own.
The Complete Excel Course with AI Integration
From formulas and pivot tables to Power Query, DAX, VBA and AI with Copilot, ChatGPT and Claude — hands-on from the first lesson.
By Sagnik Bhattacharya and Virginia Thorn, Primetime Emmy Award winner
Buy nowThe payback test in three parts
You are comparing three things: money lost to stock-outs, money tied up in excess stock, and the time spent doing forecasts today. The add-on has to beat the total, and also beat the cheaper alternative of improving your spreadsheet.
Part 1: what stock-outs cost you
Pull a year of sales orders and note the products that ran out, how many days, and the average daily sales of each before the stock-out. Lost sales are not always lost: some customers wait. Be honest with a recovery rate, for example half of the orders come back late. The margin you lost is lost sales times gross margin.
Part 2: what excess stock costs you
Take average inventory value and a carrying cost rate, which covers money tied up, storage, insurance and the risk of write-offs. Fifteen to twenty-five percent a year is a typical range people use, but use your finance person's own figure. A forecast that cuts average stock by 5% saves 5% of that carrying cost.
Part 3: what it costs you today
How many hours a week does someone spend working out what to make or buy? Multiply by their loaded hourly cost. Software might save some of that, but be careful: setting up and checking forecasts takes time too, particularly in the first quarter.
A worked example: a 14-person garden furniture maker
Take a small maker of powder-coated steel garden furniture, 14 staff, selling to retailers and online. The figures are illustrative.
| Input | Value |
|---|---|
| Active SKUs (frames, colours, parts) | 180 |
| Annual revenue | $1,800,000 |
| Gross margin | 40% |
| Sales lost or delayed to stock-outs, as share of revenue | 3% ($54,000) |
| Share of those orders never recovered | 50% ($27,000) |
| Average inventory value | $300,000 |
| Carrying cost rate | 20% a year ($60,000) |
| Hours a week on planning and reorders | 6 at $35 an hour ($10,920 a year) |
Margin lost to unrecovered stock-outs: $27,000 × 40% = $10,800 a year. Carrying cost: $60,000 a year.
Now the scenarios for what a better forecast might achieve. These are assumptions for testing, not claims about the product:
| Scenario | Stock-out margin recovered | Carrying cost saved | Planning time saved | Yearly benefit |
|---|---|---|---|---|
| Cautious | 25% = $2,700 | 3% of stock = $1,800 | 1 hour a week = $1,820 | $6,320 |
| Middle | 40% = $4,320 | 6% = $3,600 | 2 hours = $3,640 | $11,560 |
| Optimistic | 60% = $6,480 | 10% = $6,000 | 3 hours = $5,460 | $17,940 |
Divide by twelve and you get a break-even price of roughly $525 a month in the cautious case, $965 in the middle case and $1,495 in the optimistic one, before counting the extra hours you will spend on set-up and adjustments. If sales quote $400 a month for the add-on, this maker can test with confidence. If the quote is $1,200, it is paying for the optimistic case and has to prove it.
A cross-check: the quote plus the base plan is the whole cost of the system. If that maker is on the Pro plan at $599 a month for the sake of material requirements planning, an add-on quote must be read alongside that, because the decision is whether the extra spend for forecasting is justified, not whether Cin7 is.
Where the benefit gets smaller than the sum says
The sum above is generous in three ways, and each one tends to be true of small makers.
Lead times are long and lumpy. If a steel supplier delivers in six weeks and only in pallet quantities, a better forecast cannot shorten that. Reorder rules are limited by minimum order quantities. Check the supplier lead time and minimum order settings, because the forecast needs those right, and if they are wrong the forecast will not help.
Demand is driven by a few large customers. If two retailers place half your orders in big quarterly lumps, the forecast learns little from a smooth history. Your account manager's knowledge of the next order beats any model. Use forecast adjustments for known orders, and treat the model as a baseline.
Some products sell rarely. A part that sells three units a year is difficult for any forecasting engine. The spare-parts minimums tutorial covers how to handle slow movers with a fixed minimum, which is what you would do in the spreadsheet too.
The spreadsheet baseline you must beat
A forecast is only worth paying for if it is clearly better than a free one. Build the free one and score both with the same measure.
Export monthly sales by SKU for the last 24 months. For each SKU, forecast next month as the average of the last three months multiplied by a seasonal factor: this month last year divided by last year's yearly monthly average. In Excel, with sales in B2:B25 and the seasonal factor in column D, the forecast for the next month is:
=AVERAGE(B23:B25) * D26That is a rough method, though honest. Then do a backtest. Hide the last three months, forecast them from the data before, and compare with what actually sold. The usual measure is the weighted absolute percentage error:
WAPE = SUM(ABS(actual - forecast)) / SUM(actual)Compute it for your top 30 SKUs by revenue, since those carry most of the money. Suppose the spreadsheet scores a WAPE of 32%. During the trial, ask ForesightAI's forecast for the same period, or once it has run for a few weeks, compare its forecasts with what happens. If it scores 24%, it is a quarter better on the products that matter. If it scores 31%, the fee is hard to defend. I cannot tell you what it will score on your data, which is why you measure.
SKU count, history and stock-out cost: thresholds I would use
| Factor | Stay with a spreadsheet | Run the trial |
|---|---|---|
| Active SKUs | Under about 50 | 100 or more, especially many colour and size variants |
| History in Cin7 Core | Under six months (add-on cannot start) | Twelve months or more, so seasons show |
| Cost of a stock-out | Low: customers wait, or you can expedite cheaply | High: contract penalties, retailer chargebacks, lost repeat orders |
| Demand pattern | Steady, or driven by a few known orders | Many products with moderate, varying demand |
| Planner time | Under two hours a week | Over five hours a week and growing |
| Supplier lead time variety | One or two suppliers | Many suppliers with different lead times |
These are rules of thumb of mine, not Cin7's. They are also not independent: a maker with 400 SKUs and one big customer may still be better off with the spreadsheet plus account-manager knowledge. Where three or more rows land on the right-hand side, the trial is worth the effort.
Running the two-week trial so it tells you something
Two weeks is short. Plan the trial so you learn what can be learned:
- Before day one, fix your data. Check supplier lead times, minimum order quantities and the "Replenished by" setting for each SKU-location, because the help pages say it is mandatory on every product-location. Wrong settings produce a wrong forecast and you will blame the tool.
- Choose the strategy deliberately, once. The four options are Gross margin, Revenue, Sales frequency and Sales quantity. Cin7 suggests Sales frequency for new users; a maker with a cash squeeze might pick Gross margin so the engine protects the products that earn most. A change can take up to 24 hours to show and Cin7 advises changing strategy only a couple of times a year at most, so do not switch mid-trial. Note which you chose and why.
- Record the recommended reorders. On day one and at the end of each week, note what it suggests ordering and what you would have ordered by your own method.
- Test the adjustments. Add a Change by % adjustment for a known promotion and a Change quantity total for a product you are winding down, and see that the forecast behaves. Settings and adjustments reach the forecast at the next daily sync, so allow a day before judging.
- Check the component suggestions. If you assemble products from bought-in parts, mark the assembly location, then compare the raw-material suggestions for two or three finished items against your own bill of materials and current stock.
- Ask for the price in writing during the trial, with the minimum term, so the decision does not rest on a number you only get at the end.
A realistic trial result is mixed: the engine will get many products roughly right and one or two obviously wrong. A wrong result is useful when you can see why, for instance a one-off bulk order read as a trend. Whether you can correct that with an adjustment tells you how much work the engine will be each week.
A realistic mistake: trusting the forecast on a one-off order
Take a maker who, last spring, shipped a single bulk order of 400 chairs to a hotel group. The sales history shows a spike. A forecasting engine that treats the spike as demand will recommend building more stock for a repeat that never comes. The fix is housekeeping, not software: Cin7's own adjustments page names "a period of unusual demand you want to exclude" as a reason to adjust, so cancel the spike with an adjustment and write a note saying why. Doing that for the top 30 SKUs once a quarter takes about an hour, and it is something a spreadsheet user does anyway when cleaning data.
Quick sums for three other makers
A candle maker with 45 SKUs and steady online sales. Annual revenue $400,000, margin 55%, stock-outs rare because demand is smooth. Benefit might be $2,000 a year on a good day. Spreadsheet.
A pet-food co-packer with 120 SKUs and monthly contract orders. Demand is lumpy but known weeks ahead from customer forecasts. The best forecast is the customer's own purchase plan, so a model adds little. Use the customer data and a spreadsheet; revisit if non-contract sales grow. If this maker does trial ForesightAI, the Use customer orders method is the setting to test on contract lines, since it bases suggestions on confirmed orders rather than a forecast.
A cable-assembly maker with 350 SKUs, six suppliers and a 12-week lead time on connectors. Stock-outs stop production lines and incur penalties. At a $50,000 penalty bill a year, even a 20% reduction is $10,000, and the planner already spends 10 hours a week. This is the profile for the trial.
Using a chat tool to read the comparison
Once you have the backtest in a sheet, a chat assistant can help you read it, though not produce it. Export a table with columns SKU, actual units, spreadsheet forecast, ForesightAI forecast (or its recommended quantity) and unit margin, then ask:
Here is a table of 30 SKUs with actual units sold, my spreadsheet forecast and
the software forecast for the same three months. For each SKU give the absolute
error for both methods. Then list the 5 SKUs where the software was most wrong
and the 5 where it was most right relative to my spreadsheet, and the total
weighted absolute percentage error for each method. Do not estimate anything
that is not in the table.Illustrative output: "Spreadsheet WAPE 31.4%, software WAPE 25.8%. Software most wrong: SKU 114 (forecast 220, actual 40, one bulk order in history). Software most right: SKU 031 (forecast 188, actual 185)." The line to act on is the SKU 114 note, because the error traces to a one-off order, which you can fix with an adjustment. Always verify the totals yourself on two or three rows before you trust a summary of a spreadsheet written by a language model.
Checking it worked after you buy
Pick three measures before go-live and read them monthly for a quarter: number of stock-out days on your top 30 SKUs, average inventory value, and hours spent planning. Write down the starting values on the day you subscribe. If after a quarter stock-out days have not fallen and inventory has not fallen, and planning hours have risen, the add-on is costing you twice. Cin7 and your contract decide how easy it is to leave, so you want that answer before you sign, which is why the minimum term question sits near the top of the list below.
What to ask sales before you commit
- What is the price of the add-on for my plan, and how does it change as my SKU count or order volume grows?
- What is the minimum term, and what happens to my settings and adjustments if I cancel?
- How does it treat one-off orders, returns and stock-outs in the history?
- How are bills of materials and co-manufacturing locations handled in the reorder suggestions?
- What do customers with a similar range and history see after six months, in terms of the measures I can verify myself?
Treat any case study you are shown as an anecdote. The measure that matters is the backtest on your own data. If you also want a comparison of approaches beyond Cin7, the MRP add-on or chatbot tutorial sets out the alternatives for a small plant, and the stock count variance tutorial is a good thing to read first, because a forecast is no better than the stock levels it starts from.
Quote, break-even and accuracy gap: the final call
Do it in this order. Check you have six months of history in Core, as the add-on cannot work without it. Run the payback sum with your own figures and note your break-even price. Build the spreadsheet baseline and score it. Request the quote and the trial. At the end, compare three numbers: the quote, the break-even, and the accuracy gap against your spreadsheet. If the quote sits well below the cautious-case break-even and the accuracy gap is real, buy. If it is above the middle case, or the engine does not beat your spreadsheet, keep the spreadsheet and revisit when your range doubles.
Questions about Cin7 ForesightAI before you ask for a quote
What does ForesightAI cost?
Cin7's pricing page lists Core at $349, $599 and $1,199 a month for the Standard, Pro and Advanced plans for new customers, but gives no price for ForesightAI, which is a paid add-on on every plan. You add it under My subscription, so get the price and any minimum term in writing before the trial ends.
How much sales history does it need?
Cin7's help centre says ForesightAI requires at least six months of sales history in Cin7 Core. If you have moved to Core recently, history held in another system will not count. A maker with a young Core account should wait, or import history only if Cin7 confirms it is usable.
Is there a free trial?
The set-up article says a two-week free trial is available. Two weeks is short for a forecast, so you will mostly be judging the settings and the recommendations, not accuracy. Run your own backtest on past data alongside it so you have a number to compare.
Does it forecast raw materials as well as finished goods?
Cin7's settings page says that for a location marked as Assembly, ForesightAI suggests production jobs, raw materials and finished-goods transfers, using assembly lead time and assembly period settings for each bill of materials. Check those suggestions against your own bills of materials for two or three products during the trial before you rely on them.
Can a spreadsheet really compete?
For a modest range of steady-selling products, often yes. A trailing average adjusted for season and lead time gets close for items with regular demand. Software earns its fee on larger ranges, patchy demand and many suppliers, where doing the sums by hand each week stops being realistic.
Further reads
- AI Inventory Forecasting for Small Businesses: How It Works — How AI forecasting works in plain terms before you buy.
- How Small Shops Can Use AI to Forecast Stock and Reorder — A simple forecast and reorder routine without special software.
- Best AI Tools for B2B Marketing at a Small Manufacturer or Distributor — Ten AI tools matched to industrial marketing jobs: case studies from job notes, datasheet copy, trade emails and follow-ups, with prices and examples.
- What Does AI Quoting Software Cost a Machine Shop? — Pricing models, set-up time and a break-even sum on quote volume and win rate, plus how to test any quoting tool on your own past jobs first.
- AI Tools and AI Development: The Complete 2026 Guide — the AI hub, including every tutorial in the AI-for-business series.
Sources: Cin7 pricing page and Cin7 Core help-centre articles on setting up ForesightAI, strategies, ForesightAI in Cin7 Core, refining forecasts, product settings and forecast adjustments (checked 5 November 2026); all figures in the worked examples are illustrative.