The only thing as unpredictable as the weather in the seafood industry is cost accounting. From the uncertain nature of fishing to fluctuating market prices and variable yields, managing profitability is a constant struggle for seafood CFOs and financial controllers. At the heart of this challenge lies a fundamental business question:What does it cost to produce each seafood product you sell?

Figuring out the profit margin on a salmon fillet is, I would argue, more complex than calculating the cost of goods sold (COGS) for an iPhone. The difference lies in the nature of the complexity: iPhone complexity is controlled and predictable. It’s about managing a staggering number of known variables across global supply chains. Seafood complexity has far fewer variables, but they are wildly unpredictable. It’s about constantly adapting to fish price volatility, yield variation, “give-away” variance and uneven quality.

This unpredictability is a perfect problem to be solved by AI, whose core functionality is to be a “prediction machine.” AI finds underlying patterns in data, providing you with insights that you otherwise wouldn’t have. AI can predict a value (price, cost or yield), classification (grade) or anomaly (problem or error) based on pattern recognition in data.

To illustrate my point, I’ve generated a month of production data in a fictional seafood processing plant using five raw material lots. Production runs from 1,800 to 3,500 kg of raw material per day. During the month, fish prices rise. Initially, there are three lots in the inventory valued from $3.8 to $4.9 per kg. After a week, a fourth lot (D) is purchased at $5.5 per kg and a fifth lot (E) for $5.6 per kg around mid-month, raising the value of the inventory.

As the raw material inventory is depleted, lot D and E are purchased to replenish quantities, maintaining a healthy balance, as the chart below illustrations.

I then used three different methods to calculate costs. (See Seafood Cost Accounting Methods below for more details.)

• Standard Costing is commonly used in food processing where large volumes of homogeneous product are produced continuously. For this reason, it’s less useful for seafood, given so much variability in product quality, portion size, etc. It’s typically used just for the packaging or ingredients portion of costing.
• Job Order Costing or Activity-Based Costing (ABC) is a highly accurate way to allocate actual costs to a specific batch, production lot or job order. It’s often used for fresh or live products where there are no raw material inventories and short sales cycles.
• Weighted-Average Costing is based on the value of your raw material in inventory. Given price volatility, an algorithm is used to calculate the weighted average based on the cost and amount of each raw material lot. It’s typically used when products are made from frozen inventories.

If the production yield is fixed, as it is at 49 percent in the chart below, you can see the Weighted-Average Cost rising as more expensive raw material is added to the inventory. You can also visually see how wildly volatile Job Order Costing can be, from a low of $7.76 to a high of $11.70 per kg, depending on what raw material lot is used in production.

But what if yields also fluctuate, based on the quality or average fish size of each lot? In this scenario, where yields vary from 46 to 53 percent, both Job Order and Weighted Average Costing become more volatile. However, if you are paying higher prices for better quality—and therefore improved yield—Job Order Costing could become less volatile, since higher fish prices could be mitigated by higher yields. In our scenario, that’s not the case. Higher raw material prices are resulting in costs clearly rising, from $9.59 in the fixed yield scenario to $10.03 in the dynamic scenario.

The wavy graph above is enough to make any CFO seasick.

In my experience, I see a lot of seafood companies run by financial “dead reckoning” since they don’t have real-time accurate costing data. They know their financial position from the last month or quarter, but their current financial position is an estimated guess based on imperfect data. But just like in maritime navigation, errors compound over time. The longer you go without a fix, the further your estimate drifts from reality.

I was once working with a Japanese seafood company that seemed like a perfect customer. They ran a spotless processing plant, a stellar example of Japanese “lean manufacturing.” However, they managed their entire data via a WhatsApp group and Excel. While the plant was squeaky clean, their data was not. Their financial manager only figured out their profitability long after month’s end. It was financial dead reckoning at its worst: they eventually went out of business before I made a sale.

At ThisFish Inc., we’ve developed complex costing models for seafood processors, using both Job Order and Weighted-Average Costing methods (See our case study on Orca Specialty Foods). The critical operational variables are the purchase prices and the daily production yields, which typically contribute more than 70 percent of the cost of goods sold for a seafood product. For both a salmon processor and tuna cannery, we then took years of good quality production data to build yield prediction models using a machine learning algorithm. If you can accurately predict your yield, then you’ll also know your production costs in advance.

Real-time digital data is the navigational equivalent of GPS for seafood costing, knowing your exact profit or loss on each sales order or job. And with this good quality data, you can then develop predictive AI models which act like a radar, giving you notice of financial dangers ahead.

SIDE BAR | SEAFOOD COST ACCOUNTING METHODS

Standard Costing

• How It Works: Cost are accumulated over a period—typically a month—for an entire process such as the production of one SKU. The total cost is then averaged over the total number of units produced to calculate a cost per unit. This is how most accounting software deals with costing by giving a standard or fixed value at the end of the month.
• Best Use Case: It is best used when material prices don’t fluctuate much and production is standardized. In seafood, standardized costing is often used for labour, packaging and ingredients costing. Key
• Challenge: Standardized costing typically lags finished product pricing, leading to unfortunate surprises at month’s end. You only know your costs and, most importantly, profit margins at month’s end. You are always looking in the rearview mirror, using last month’s costing to estimate gross margins.

Job Order Costing

• How It Works: All costs are tracked specifically for each individual job order. For raw material, the cost of the fish is divided by the production yield to determine the actual raw material cost to produce each item. Processors may also track how much labour went into each batch or track the cost of any wasted packaging or ingredients during production.
• Best Use Case: With fresh and live seafood sales, where there are volatile raw material costs and market pricing, having accurate data on every batch and sales order can be critical for profitability. Production yields are also critical.
• Challenges: Most accounting and ERP software can’t handle this dynamic, detailed seafood costing, and so companies struggle with complex Excel spreadsheets to manage activity-based accounting. Real-time manual reporting can be a struggle too, with calculations prone to human error.

Weighted-Average Costing

• How It Works: If there are four raw material lots in the inventory, the total weight of the inventory is divided by the total cost of that raw material good, generating a weighted average. As quantities of each lot are put into production, the weighted average will fluctuate slightly, up or down depending on if the lot was above or below the weighted average. Also, when new raw material is added to the inventory, the new lot quantity and price change the weighted average. Also known as rolling average.
• Best Use Case: This method is best used when manufacturing standardized products—a can of tuna or fixed weight seafood item—from frozen raw material inventory. This method avoids the wildly fluctuating costing that you often see with Job Order or Activity-Based Costing.
• Challenges: Companies can struggle to calculate real-time inventory costs, as the inventory changes each day. It typically requires advanced production and inventory software.