Fish fillet yield is easy to state and surprisingly easy to misread. A processing report may say 42% fillet recovery, a buyer may cite 55% yield from headed-and-gutted fish, and a factory dashboard may show 38% saleable product after trimming. All three numbers could arise from the same physical shipment, because their denominators, cuts and product specifications differ. A percentage without a starting mass and an ending definition is not a complete processing result.
Dressing percentage, fillet yield, skin removal, deboning, trimming, glaze addition and packing losses belong to different steps of seafood processing. The purpose of this guide is to translate each step into a transparent mass balance. We will work with fictional production records, check the arithmetic, distinguish water-related weight changes from edible recovery, and explain why a high reported yield is not automatically a better outcome if quality or specification differs.
A useful rule appears before any calculation: name the incoming state and the outgoing state. “Skinless boneless trimmed fillet, percent of whole ungutted fish mass” communicates much more than “fillet yield 40%”. Once the states are explicit, the reader can connect them with measured weights, track uncertainties and avoid combining incompatible percentages. This is the same discipline used when translating numerical evidence in science: the denominator is part of the meaning.
Begin with the anatomical and commercial product definitions
A whole fish may arrive alive, chilled, frozen, bled, gutted, headed, or in an intermediate dressed form. Head-on gutted, headed-and-gutted, skin-on fillet, skinless fillet, trimmed fillet, boneless portion and recovered mince are not synonyms. They differ in which tissue is present. If a yield table uses “H&G”, confirm whether it means headed-and-gutted for the particular operation; similar acronyms can have local meanings. If a product label says “loin”, request the cut specification instead of assuming a universal anatomical portion.
Dressing percentage normally compares the mass of a dressed product with a stated original state. Yet “dressed” may be used differently by processors or species conventions. A report might define it as gutted only, another as headed-and-gutted, and a third as fully prepared retail units. Rather than memorising one meaning and imposing it everywhere, read the process flow and identify which organs, head, fins, skin and bone have been removed. The valid ratio is mass after the specified dressing step divided by mass of the explicitly stated input.
A fillet recovery calculation works the same way. When the output is 180 kg skinless trimmed fillets from 500 kg whole ungutted fish, yield is 180 / 500 × 100 = 36% on a whole-fish basis. When the same 180 kg came from 360 kg gutted head-on input, yield is 50% on a gutted-input basis. Both describe the same 180 kg outcome but tell different stories. A translator who retains the percent but deletes “of gutted head-on weight” can turn an accurate supplier statement into an inaccurate purchasing comparison.
A reader should also distinguish product yield from edible yield. Trim and mechanically recovered mince might be edible and saleable under suitable processing and food-safety controls, while a skinless premium fillet specification rejects that material. A “total edible recovery” metric can therefore exceed “skinless fillet recovery” without violating mass conservation. The products cannot simply be pooled if their quality, end use and saleability differ. Keep each mass stream and its specification in its own row.
Map the processing sequence before multiplying percentages
An illustrative sequence is whole fish, gutted fish, headed-and-gutted fish, skin-on fillets, skinless fillets, trimmed saleable fillets. Every transition has an input mass and an output mass. Percentages across stages are conditional: fillet recovery from headed-and-gutted fish is a fraction of that stage’s input, not of the original whole fish. To recover overall yield from stage yields, multiply the fractions in their physical order and confirm that every numerator becomes the next stage’s denominator.
Suppose a factory starts with 1,000 kg whole fish. Gutting retains 900 kg, heading retains 760 kg, filleting retains 410 kg, skinning retains 375 kg, and trimming retains 340 kg. The stage yields are 90%, 760/900 = 84.44%, 410/760 = 53.95%, 375/410 = 91.46%, and 340/375 = 90.67%. Their product, using unrounded fractions, is 340/1,000 = 34% overall skinless trimmed yield. Adding the stage percentages would be meaningless; averaging them would also be wrong. Each stage answers a conditional question.
A frequent error is multiplying a rounded yield from one operation by a rounded yield from another and treating the result as an exact mass. If a stage is recorded as 84% rather than 760/900, rounding already removes information. Reconstruct total mass from raw kilograms where possible, then round the final percentage to a sensible precision. A small difference caused by rounding is not evidence of missing fish. A much larger difference that persists with raw records deserves investigation.
Process diagrams should show reject streams as well as the main product. Gutting removes viscera and sometimes other material; heading produces heads; filleting produces frames; skinning produces skins; trimming produces cutoffs. Some streams may be reused, sold or rendered. If the report accounts only for fillet mass, that does not mean all the remaining mass is “waste”. It may include recognised co-products, water, thaw drip and unmeasured material. Use neutral labels until their handling and destination are documented.
Worked example: convert a conditional fillet yield into whole-fish yield
A fictional batch has 700 kg whole fish. After gutting and heading, 560 kg remains. Filleting produces 280 kg skin-on fillets. Therefore H&G recovery is 560/700 = 80%; skin-on fillet recovery on H&G basis is 280/560 = 50%; and skin-on fillet recovery on whole-fish basis is 280/700 = 40%. Multiplying 0.80 × 0.50 confirms 0.40. These values are mathematically consistent and express three separate process descriptions.
Now skinning and trimming leave 252 kg saleable fillets. The last-stage retention relative to 280 kg is 90%. The whole-fish saleable recovery is 252/700 = 36%, and H&G saleable recovery is 252/560 = 45%. If a processor quotes “45% finished fillet recovery” to a buyer comparing whole fish inputs, that buyer could overestimate output unless the H&G denominator is explicitly included. The number has not changed but the implied decision has.
The remaining 448 kg of mass is not automatically 448 kg of environmental waste. Its components might include marketable heads, frames, skin, recovered mince, viscera, drip and water transfer. A good worksheet tracks those flows separately and identifies what is saleable, reused, disposed of, or unmeasured. This matters operationally because higher finished fillet recovery may improve economics, but a facility might also obtain value from co-products or alter its product mix.
Product specification can make the same fish produce different yields
A premium boneless, skinless and deeply trimmed fillet removes tissues that another buyer might accept. The more restrictive specification reduces saleable mass even when cutting skill remains unchanged. Compare two hypothetical buyers receiving the same 300 kg skin-on fillet lot. Buyer A accepts skin-on pieces with minor dark muscle, so 288 kg passes inspection. Buyer B requires skinless portions with a stronger trim and accepts 235 kg. The recoveries of 96% and 78.3% describe different outputs. Calling the first processor more efficient without acknowledging specification would be misleading.
Small fish can be harder to fillet efficiently than large fish because fixed anatomical losses and handling tolerances may occupy a greater share of body mass. Size, shape, maturity, season, condition, and fish species affect anatomical proportions. Mechanical equipment also has setup tolerances and is not equally suited to every size band. A yield comparison should therefore retain the fish size distribution and cut specification, not just the average input mass. If incoming cohorts differ, a difference in recovery may have nothing to do with operator performance.
Some studies present averages across hundreds of fish, while others use a handful of specimens or large commercial runs. Their applicability depends on measurement method, species, size, season and processing state. The FAO compilation of fish yields demonstrates why reported values vary even within the same species: definitions such as gutted head-on weight, whole weight, skinned versus skin-on fillets, and trimmed versus untrimmed flesh are not interchangeable. A responsible reader extracts the original denominator rather than lifting a single attractive benchmark.
Frozen weight, glaze and thaw drip require separate accounting
A frozen fish product may carry surface ice glaze intended to protect it during frozen storage. Gross packaged mass can include packaging, glaze and product, while a net weight claim should follow applicable rules regarding what can be included. If 120 kg of frozen packages contains 4 kg packaging and 7 kg glaze, the underlying unglazed product estimate would be 109 kg only if those quantities were correctly measured and defined. Do not assume glaze is edible fish or silently treat it as fillet recovery.
Thawing can change the measured weight of fish through drip loss and surface water removal. Consider 100 kg frozen unglazed fillet that becomes 97 kg drained fillet after controlled thawing, with 3 kg recorded drip. That is a 3% weight loss relative to frozen unglazed input; it does not imply that precisely 3% of the original protein or EPA/DHA mass has been lost. Drip composition matters. Similarly, a processor that improves water retention may record a higher mass yield without producing more fish tissue. Mass and composition should be evaluated separately.
A clean yield report states the temperature and handling state of each weighing: chilled and drained, frozen unglazed, glazed, thawed and drained, or prepared for retail. Weighing at inconsistent stages creates artificial gains and losses. The solution is not to invent a universal correction factor; it is to standardise weighing procedures and record the state. When comparing different studies, report the incompatibility explicitly rather than converting with undocumented assumptions.
Measurement precision and the danger of a denominator switch
A weighing instrument can have excellent repeatability yet introduce systematic error if incorrectly calibrated. A platform scale suitable for 500 kg batches may not accurately resolve a 200 g trimming subsample. Record instrument resolution, calibration status, tare procedures and whether gross or net masses were used. The difference between 37.4% and 37.5% yield might be less than the combined weighing and sampling uncertainty. Reporting to four decimal places adds appearance of certainty without new evidence.
Denominator switching is more consequential than ordinary rounding. Suppose a historical report uses 400 kg trimmed fillets from 1,000 kg head-on gutted fish, or 40%. A new report uses 400 kg from 1,200 kg whole fish, or 33.33%. A sentence saying “yield declined from 40% to 33.33%” is not a valid process comparison because input states differ. Obtain the whole-to-gutted conversion for both batches or compare outputs on one shared basis before attributing any decline to operations.
There is a second denominator problem inside co-products. A facility might say “bone-free mince recovery 20%” without stating whether it is 20% of frames sent to the recovery equipment or 20% of original whole fish. If 100 kg of frames yield 20 kg mince from a 700 kg whole-fish batch, frame-based recovery is 20%, but whole-fish recovery is 20/700 × 100 ≈ 2.86%. Both are valid. The value is actionable only when the relationship is clear.
Build a complete facility mass balance, not merely a fillet percentage
The basic accounting identity is incoming product mass plus other measured material additions equals all measured outgoing streams plus inventory change and any identified unaccounted difference. For a simple batch without added materials or inventory change, 1,000 kg input should be reconciled against fillets, heads, viscera, skin, frames, trim, recovered flesh, drip and other recorded streams. The equation does not by itself demonstrate that every stream is suitable for food or that every outgoing kilogram has commercial value. It is a physical completeness check, not a food-safety certificate.
Imagine 1,000 kg whole fish yielding 355 kg premium fillets, 85 kg trimmed edible offcuts, 120 kg heads, 150 kg frames, 95 kg viscera, 60 kg skin, 25 kg other measured rejects and 40 kg measured drainage. The listed outputs total 930 kg, leaving 70 kg unaccounted. That 70 kg difference is 7% of incoming mass. It might represent process-water exchange, unrecorded co-products, retained material in equipment, scales measured at different times, or ledger omissions. Do not call it “evaporation” or “loss” without evidence. A balanced ledger is a question to answer, not a reason to manufacture a plausible story.
If a separate inventory count shows 50 kg of product remained in chilled bins at the end, the discrepancy falls to 20 kg, or 2% of original mass. If wash water was absorbed by material and later discharged, the bookkeeping needs to include those water flows where material. The analyst should agree the process boundary first. A factory-wide daily balance, a single trimming station, and a chilled-packaging batch can each be correct yet incomparable because their boundaries differ. Labelling the boundary is as important as labelling the denominator.
Co-products can change the economic interpretation without changing the mass balance
Suppose two plants process identical 1,000 kg batches. Plant A produces 370 kg premium fillets and sells none of the cutoffs. Plant B produces 355 kg premium fillets plus 85 kg saleable trimming suitable for a validated food use. Plant A leads on premium fillet percentage, but Plant B may produce more total saleable edible mass. The relevant comparison depends on quality, price, processing costs, food-safety approval and demand. Reporting one aggregate yield hides the distinction between a premium fillet and a lower-priced co-product.
Economic yield is not necessarily mass yield. With hypothetical prices of $12 per kg premium fillets and $3 per kg saleable trim, Plant B’s material revenue is 355 × 12 + 85 × 3 = $4,515 before other costs and revenues. Plant A’s 370 kg premium fillets at the same price yield $4,440, assuming no other outputs sold. The difference is $75 per tonne of whole fish under these invented prices, not a universal result. Changing co-product price, handling cost or specification could reverse the comparison. To discuss profitability, include all relevant costs, not just kilograms recovered.
Nor should every co-product automatically be classed as usable animal feed or edible material. Appropriate handling, traceability, contamination control and regulations govern destination. The seafood traceability question is examined separately in eduKate’s guide to lot lineage and processing mass balance. A processing ledger can assign material streams accurately while the food-safety system decides whether each stream may enter a particular market. Those are complementary, not interchangeable, assurance systems.
Trimming decisions are quality-control decisions
A trim loss is not always a mechanical failure. Workers may remove skin fragments, bones, dark muscle, blood spots, parasites where applicable, visible damage or parts that do not meet an agreed specification. A rise in trimming may reflect more rigorous customer requirements or incoming quality concerns rather than lower worker skill. A fair evaluation separates necessary quality-related removal from avoidable excessive cutting. Otherwise a manager can inadvertently reward high mass yield achieved by failing to meet a specification.
Consider 200 kg untrimmed fillets. A batch is inspected and 12 kg is removed for standard specification trimming, while another 5 kg is removed due to documented defects. Final accepted fillet mass is 183 kg, or 91.5% of untrimmed fillet mass. If a report says 8.5% trim loss, it should distinguish the 6% standard trim and 2.5% defect-related loss when records allow. If both categories use the same denominator, their percentages add to 8.5%; if one category is calculated after the other, they cannot simply be added without converting to a shared basis.
A second decision concerns tolerance. A supplier might quote a target recovery of 92% with acceptable lot-to-lot variation. A measured value of 91.8% is not automatically a failed process if uncertainty and specifications permit it; nor is 92.2% automatically proof of improvement. A specification should define sampling, acceptable defects, product temperature, inspection criteria, and the denominator. Translate “target”, “minimum”, “observed” and “guarantee” as distinct claim types. They imply different obligations.
Sampling can systematically overstate recovery
If a supervisor measures yield only on the largest fish, the sample may not represent a mixed-size commercial batch. Likewise, selecting neatly filleted trays for audit while excluding difficult damaged fish creates selection bias. To estimate batch recovery, the sampling design should cover relevant size classes, shifts, operators and incoming conditions, or else calculate the precise subgroup to which the sample applies. Averages should be weighted by appropriate mass or count, not automatically averaged across arbitrarily sized groups.
For example, 100 small fish yield 22 kg saleable fillet from 70 kg whole mass, a 31.43% recovery. Fifty larger fish yield 60 kg fillet from 150 kg whole mass, a 40% recovery. An unweighted average of the two group percentages would be about 35.71%. But the combined mass-based recovery is (22 + 60)/(70 + 150) = 82/220 = 37.27%. The latter is the correct combined yield for the batch because numerator and denominator are summed first. Group averages can answer a different question, but the choice must be explicit.
Similarly, pooling ten fish measured individually does not prove a machine’s average yield under production conditions if all ten were handled by a specialist in a controlled test. Repeating an observation under the same unusual conditions provides better precision about that demonstration, not necessarily representative factory performance. The report should distinguish controlled validation, routine production audit and ad hoc complaint investigation. A translated percentage without study design often looks more general than the evidence actually is.
The effect of species, maturity and incoming quality
Different species have different head proportions, bone structures, flesh shapes, skin adhesion and fat distribution. Even within one species, maturity and season can affect condition and tissue proportions. That is why generic yield percentages must not be used as universal conversion factors. A cod conversion drawn from a reference manual does not automatically apply to a small pelagic species, a crustacean, a farmed salmon or a fish processed into a different product. Use a species- and product-specific source or a validated facility measurement.
Fish condition before processing matters as well. Handling damage, bruising, rough storage, delayed chilling and poor raw-material quality can increase trimming or rejects. A report that attributes lower yield solely to the filleting machine may ignore the incoming raw material. Conversely, the machine’s blade settings, maintenance and fish orientation can matter significantly. An informative investigation keeps two sets of evidence: incoming fish specification and process performance. The causal question is answered only when both are observed.
Size grading connects to this issue but is not the same metric. eduKate’s fish size-grading guide asks how fish are distributed across weight classes and how that affects selection. Fillet yield asks what fraction of a defined incoming state becomes a defined processed output. A larger market-ready share does not guarantee the same fillet recovery percentage, and average size alone does not settle the matter. The two reports can complement one another if fish lots and definitions are aligned.
Distinguish gross weight, net weight and saleable weight
Gross weight often refers to a shipment or package including materials outside the fish product. Net weight commonly refers to product quantity after excluding packaging, subject to applicable labelling law. Saleable yield then measures the subset of output passing specified quality and market conditions. If 500 kg gross packaged boxes contain 30 kg carton and pallet materials, 20 kg permitted surface glaze and 450 kg fish product, it would be wrong to call all 500 kg edible fillet. The precise net-declaration requirements must be reviewed for the relevant jurisdiction and product category.
A simple shipping discrepancy can arise because the production yield report uses 450 kg post-trim fillet weighed before glazing, while the warehouse picks 470 kg of frozen glazed units and the freight carrier bills 500 kg gross. Each number describes a different stage. To reconcile them, record product mass, glaze, packaging, time and water state. Replacing all with a single “shipment weight” destroys the traceability of mass. The same logic applies when converting yields from fresh input to frozen finished product.
An additional pitfall is using pieces rather than mass. A factory can report 950 accepted portions from 1,000 cut portions, or 95% acceptance by count, while accepted portions contain only 90% of the original portion mass. Larger rejects or unequal portion sizes make the count-based and mass-based percentages differ. Both can matter to operations, but they cannot stand in for one another. Ask whether the denominator is pieces, kilograms, sale value or some other quantity before drawing a performance conclusion.
A rigorous way to compare two processing lines
Imagine Line A and Line B each process five independent, comparable lots. A comparison should keep species, fish size, initial condition, product specification, handling state, equipment configuration, scale calibration and quality rules as similar as practical. Calculate recovery for each batch from its own measured masses and then decide whether to compare pooled mass yield, lot-level mean yield, or a statistical model that accounts for batch differences. Merely reporting the highest percentage from each line creates selection bias.
If the five lot-level recoveries on Line A are 35%, 36%, 35%, 37% and 36%, while B has 34%, 39%, 35%, 36% and 36%, both simple means equal 35.8% and 36% respectively? Recheck: A totals 179%, mean 35.8%; B totals 180%, mean 36.0%. That 0.2 percentage-point difference is small relative to the visible spread in B. Without appropriate replicate design and statistical analysis, the values do not prove B is systematically more efficient. A more stable line can be operationally preferable even if a few exceptionally high runs lift another line’s average.
If lot masses differ, a pooled ratio of total saleable kilograms to total input kilograms may answer the factory-wide material-efficiency question better than the unweighted mean of batch percentages. But pooled recovery can conceal variation with fish size or shift. A complete report may present both pooled mass recovery and the distribution across batches, explaining what each means. As in other quantitative translations, the right number depends on the reader’s decision.
Independent problem one: a misleading yield advertisement
A supplier advertises “55% fillet yield” and a processing buyer expects at least 550 kg finished fillets from 1,000 kg of whole fish. The supplier’s technical sheet specifies that the 55% is recovery of skin-on fillets from headed-and-gutted fish. In the buyer’s lot, 780 kg headed-and-gutted fish is obtained from 1,000 kg whole input. The expected skin-on fillet output under the supplier’s ratio is 780 × 0.55 = 429 kg, or 42.9% of whole input, not 550 kg. If skinning and trimming then retain 88% of the skin-on fillet mass, finished output is 429 × 0.88 = 377.52 kg, or 37.75% of whole input.
The advertisement’s value is not necessarily false, but its shorthand is dangerously incomplete. The buyer should request both conditional and overall yields, fish and product specifications and an explanation of whether the quoted rate is an expectation, trial result, long-run average or guaranteed minimum. This example is why a ratio is not fully translated until the denominator and stage are spoken aloud. The arithmetic check is 0.78 × 0.55 × 0.88 = 0.37752, matching 377.52/1,000.
Independent problem two: group weighting changes the story
Three production lots have input masses of 100 kg, 500 kg and 900 kg with corresponding saleable yields of 40%, 34% and 36%. A simple average of the three percentages is (40 + 34 + 36)/3 = 36.67%, which describes the average lot-level percentage if every lot receives equal weight. The output masses, however, are 40 kg, 170 kg and 324 kg. Total output is 534 kg from 1,500 kg input, giving a mass-weighted overall yield of 35.6%. The difference is real; neither calculation is arithmetically incorrect.
The factory manager assessing total recovered kilograms should generally use 35.6%. A researcher studying how yield behaves from lot to lot might examine the unweighted lot distribution as well, perhaps with a model for fish size or operator effects. If a report gives one figure without naming the aggregation rule, readers may make the wrong decision. This is not an advanced statistical trick; it is the consequence of different questions requiring different denominators.
Independent problem three: trim acceptance and moisture
A line records 300 kg untrimmed chilled fillets, 24 kg standard trim, 9 kg rejected for defects, and 267 kg accepted finished fillets. The accepted fraction is 267/300 = 89%. Standard trim is 8% and defect reject is 3% of the same 300 kg input. The three fractions reconcile exactly: 89 + 8 + 3 = 100%. If moisture changes between weighing stations, however, this purely anatomical balance may no longer close, and the water transfer must be recorded separately rather than disguised as an unexplained trim category.
Suppose the accepted 267 kg subsequently loses 6 kg drained water during storage. The shipping mass is then 261 kg. Its ratio to original untrimmed fillet mass is 87%, but that number should not be attributed wholly to cutting loss. Distinguishing anatomical trim from later drip allows operations to investigate the right process. It also prevents a quality-control worker from being blamed for a storage-related weight change.
Independent problem four: quality can dominate a nominal recovery gain
Two hypothetical settings on a filleting machine process equivalent 500 kg batches. Setting A produces 205 kg nominal fillets but 15 kg fail the agreed bone-free specification, leaving 190 kg saleable. Setting B produces 198 kg nominal fillets but only 3 kg fail, leaving 195 kg saleable. Nominal recovery is 41% for A and 39.6% for B. Saleable recovery is 38% for A and 39% for B. On the actual specification, B performs better in recovered saleable mass despite making fewer kilograms before inspection.
If both settings require different rework time, energy or labour, those costs should also be considered before selecting equipment. An acceptable comparison includes product quality, rework, safety, co-products and economic value. Purely maximising pre-inspection mass can invite undertrimming or inadequate bone removal. This illustrates the wider principle: the output must be defined before the performance metric can be chosen.
A practical reconciliation procedure for incoming and outgoing ledgers
First, freeze the physical boundary of the batch. Decide whether records begin at whole fish receipt, headed-and-gutted intake, chilled thawed fillets or another state. Record the weighing date and material condition, including water and temperature. Second, identify all outgoing streams at the same boundary, including inventory held for later work. Third, calculate absolute mass quantities before percentages, so a later reader can recompute the rates. Fourth, group co-products by actual destination without assuming all removed material is waste.
Fifth, compare mass balances with calibration and operational uncertainties rather than arbitrarily forcing them to total exactly. A missing material stream is not fixed by renaming the discrepancy. Sixth, separate product specifications: a skin-on fillet yield is not a skinless boneless trimmed yield. Seventh, report each useful metric with its numerator and denominator and identify whether it is a measured batch result, a multi-batch average, a projection or a guarantee. Eighth, store the original records alongside the finished narrative so corrections remain traceable if later inspections uncover errors.
The review should end with an independence check: can a reader who did not attend the processing session reconstruct 340 kg/1,000 kg = 34% and identify precisely what those two masses include? If not, the report lacks necessary context. This standard is suitable for a factory manager, purchaser, educator, translator or auditor, regardless of location. It does not require making the article longer than necessary; it requires making every numerical claim reconstructible.
FAQs about fish yield and processing recovery
Does 50% fillet yield mean half of every whole fish is edible? Not necessarily. The denominator might be headed-and-gutted fish, while other edible co-products may exist. It can also be an average across a particular size range and cutting specification. State the actual input and output first.
Is a higher finished fillet weight always better? No. Surface water, retained glaze, bones, skin and unacceptable tissue may increase mass without improving saleable product under the buyer’s specification. Safety, quality and legal labelling matter. Compare equivalent products at consistent weighing states.
Can a published yield for one species be used for another species? Not as a general identity. Anatomy, size and processing technique can differ greatly. A published figure can guide what to investigate, but a validated comparable study or facility-specific estimate is needed for a confident operational forecast.
Why might yield rise after processing equipment is improved? Better alignment, reduced unnecessary trimming, cleaner cuts and fewer rejects are plausible mechanisms. But if incoming fish size or specifications changed at the same time, the before-and-after comparison alone may not identify which mechanism caused the improvement.
Why does a processing report show 47% recovery while the sales ledger shows 39%? The processing figure may refer to an earlier product state, different input basis or production before later trim, sorting, glaze removal and quality rejections. Reconcile the ledger stages rather than assuming one percentage was entered incorrectly.
A final check before accepting the English wording
A strong sentence says “The processor recovered 267 kg of specification-compliant chilled skinless fillets from 300 kg of untrimmed fillets, or 89.0% at the trimming stage.” A weak sentence says “The yield was 89%.” The first identifies the stage, acceptable output, masses and physical state. The second invites every reader to supply their own meaning. Clear writing turns processing evidence into a usable decision; the precision is not decoration.
Primary technical sources and related reading
FAO, Yield and nutritional value of commercially important fish species — species-specific tables that disclose whether a percentage refers to whole fish, gutted fish, skin-on or skinned fillets.
FAO, The Haddock — a worked species example that distinguishes yield of different product types.
Google Search Central, Creating helpful, reliable, people-first content — source for responsible editorial emphasis on accurate, substantial and helpful explanations.