Skip to content

In development · early access list open

See the next trend before it peaks.Know the margin before you order.

TrendFalcon is trend software for online shops, built to predict the next rising trend. It reads what people pay attention to every day, checks whether a rise holds or is only a spike, and works out what one product would leave you after supplier cost, freight, duty, VAT, fees and the marketing it takes to win each order. Every number shows its source and the date it was read.

For Swedish shops: a hand-made decision file for one product category is coming soon. See the decision file, with a worked example →

This page previews the product we are building. Every figure in the product views is sample data. The early-access list is open; workspace access dates and pricing are not set yet.

Our goal: the trend platform every online shop runs on.

The vision
Every online shop that sells goods asks the same question, season after season: what do we sell next? We are building the platform that answers it: the trend, the evidence, the landed cost, the margin, a test and a signature, in one loop. We start in Sweden. The goal is every market.
Why now
Lists of what is trending are given away, and the EU now charges customs duty on low-value parcels from outside it. A list shows what is popular today. A shop needs to know what will still be rising when its stock arrives, and what it leaves after every cost, marketing included. That is the part we are building.
The moat
A record built to compound: every test is designed to add what was known, what was done and what happened. The record of outcomes shared with consent is built to improve the next ranking. That record cannot be copied from a list, and it is never sold.
For a shop
  • Spot the rise while there is time to stock it. Every rise is read from dated sources and checked for whether it holds, so a one-day spike is not mistaken for a trend.
  • The margin before the first order. Landed cost and margin are worked out by code, line by line, before a single unit is bought.
  • A no before the money is spent. One place from trend to signed decision. Every figure shows its source and the date it was read, and what is not known is said.

The research behind itWhere it goes

  • Path: Sauna hat
  • Rising
  • Falling

Drag to turn. Select a node to open its view. Every node is sample data, except the three real topics.

00The loop

The same graph as a list

01 · Trend

  • Encyclopedia readingHow often a subject is read, per language edition. Shown in the product, as the source’s own terms allow
  • Search interestWhat people look for, as weekly volume ranges. Shown in the product, as the source’s own terms allow
  • Retail listings and pricesWhat shops in the market list and charge, with the date. Shown in the product, as the source’s own terms allow
  • Supplier cataloguesPrice tiers, stock, lead times and shipping. Shown in the product, as the source’s own terms allow
  • Market statisticsWhat a normal season looks like in each category. Shown in the product, as the source’s own terms allow
  • Event calendarsKnown dates that move demand. Shown in the product, as the source’s own terms allow
  • Other platform signalsAttention on platforms that do not allow display. Internal analysis only, never shown
  • Test outcomesWhat happened when a product was tested. Yours, shared only with consent
  • Sauna hat, SwedenRising. +60% in four weeks
  • Felted wool slippers, GermanyRising. +36% in four weeks
  • Reflective dog vest, SwedenSeasonal. +14% in four weeks
  • Window bird feeder, GermanyRising. +49% in four weeks
  • Pickleball paddle cover, GermanyToo early to tell. +20% in four weeks
  • Linen bed throw, SwedenPeaked. −2% in four weeks
  • Dog cooling mat, GermanyCooling. −40% in four weeks
  • Balcony planter box, SwedenCooling. −37% in four weeks
  • Knitting needle case, SwedenRising. +26% in four weeks
  • Cork yoga block, GermanyPeaked. +9% in four weeks
  • Boot drying rack, SwedenSeasonal. +15% in four weeks
  • Padel grip tape, SwedenCooling. −20% in four weeks
  • Matcha, real dataReal data · Wikipedia pageviews. Too early to tell
  • Pilates, real dataReal data · Wikipedia pageviews. Too early to tell
  • Sneakers, real dataReal data · Wikipedia pageviews. Too early to tell

02 · Products

  • Sauna set: hat, seat pad and linen towel€79. Margin €28.60 per order
  • Wool felt sauna hat€29. Margin €8.85 per order
  • Embroidered sauna hat€39. Margin €13.61 per order
  • Sauna hat and felt seat pad€55. Margin €20.81 per order
  • Sauna set for two€135. Margin €48.29 per order

03 · Suppliers

  • Supplier A, inside the EUOffer dated 3 Nov. Goods €21.00, 12 days to deliver
  • Supplier B, outside the EUOffer dated 28 Oct. Goods €13.50, 42 days to deliver

04 · Delivery

  • Freight and duty per unit€1.30 inbound freight and €0.00 duty, per unit
  • Delivery to the customer€5.40 per order, to the customer’s door
  • Days to deliver12 days to deliver, 21 days left to the season: in time

05 · Costs

  • Landed cost€27.70: goods, inbound freight, duty and delivery
  • Planned ad cost per order€3.00 of ads planned to win each order
  • Margin per order after ads€25.60 after €3.00 of ads; the shop’s minimum is €22.00
  • Money at risk€1,395: the first order’s stock and the test’s ad budget, before the first sale is known

06 · Decision

  • Decision briefThe decision on one page, with every number’s source

07 · Evaluation

  • Rules1 of 4 questions still open
  • Timing12 days to deliver: in time for the season
  • Test planLocked before launch. At most €726 at risk
  • Phase 1: screen160 visits, spend cap €176. Stop if fewer than 3 orders
  • Phase 2: confirm500 visits, spend cap €550. Pass at 84 orders
  • Stop rulesFour rules, written before launch
  • Pass rule84 orders or more in the confirmation
  • WatchAfter the decision: confirmed, fading, time to stop
  • This testNot run yet. Its outcome will join the record
  • Knitting needle case, SwedenPassed. 28 orders from 660 visits
  • Window bird feeder, GermanyStopped at the screen. 0 orders from 160 visits
  • Linen bed throw, SwedenPassed. 32 orders from 660 visits
  • Cork yoga block, GermanyStopped at confirmation. 16 orders from 660 visits
  • Dog cooling mat, GermanyStopped at the screen. 1 order from 160 visits
  • Balcony planter box, SwedenPassed. 26 orders from 660 visits
  • Boot drying rack, SwedenStopped at the screen. 2 orders from 160 visits
  • Padel grip tape, SwedenStopped at the screen. 1 order from 160 visits
  • Dated record of callsEvery call is written down and dated before its outcome can be known

01Trend

Sense a signal, and understand what kind of rise it is.

A signal rises where people’s attention moves. Attention signals come in every day today; supply and price signals are being added, each with its source, its date and what we are allowed to do with it. Is it a real trend, a season or a spike? A spike is not a trend. In this preview the written rule asks for a rise to hold for weeks; the check against the season is not run on live data yet. How many sources show the same movement is counted beside it.

Signal sources

Sample data
Concepts tracked
1,240
Categories
6
Markets
2
Signal families
8
  • Encyclopedia readingHow often a subject is read, per language editionUpdated: DailyCovers: 10 yearsPlanned display
  • Search interestWhat people look for, as weekly volume rangesUpdated: WeeklyCovers: 5 yearsPlanned display
  • Retail listings and pricesWhat shops in the market list and charge, with the dateUpdated: WeeklyCovers: 2 yearsPlanned display
  • Supplier cataloguesPrice tiers, stock, lead times and shippingUpdated: DailyCovers: 1 yearPlanned display
  • Market statisticsWhat a normal season looks like in each categoryUpdated: MonthlyCovers: 15 yearsPlanned display
  • Event calendarsKnown dates that move demandUpdated: YearlyCovers: Next 18 monthsPlanned display
  • Other platform signalsAttention on platforms that do not allow displayUpdated: DailyCovers: 1 yearInternal only
  • Test outcomesWhat happened when a product was testedUpdated: Per testCovers: From the first testWith consent

Encyclopedia reading

Planned display

Built to count how often a subject is read, joined across language editions into one concept.

Sample reading: Sauna hat, Sweden

Same movement as the trend.

Reading index, week 20 to 45

  • This week254
  • The season alone would give109
  • One concept, in every languagebastumössa · Saunahut · sauna hat
The source
  • UpdatedDaily
  • History10 years
  • UsePlanned: shown in the product, as the source’s own terms allow
  • Coverage6 of 6 categories in Sweden, 6 in Germany

Planned display: shown in the product, as the source’s own terms allowInternal only: internal analysis only, never shownWith consent: yours, shared only with consent

Sample data. The kinds of source are the plan; every figure here is invented for the preview.

Coverage

Sample data
Market
32 of 36 cells covered
Coverage in Sweden, category by signal
ReadingSearchRetailSuppliersStatisticsEvents
coveredcoveredcoveredcoveredcoveredcovered
coveredcoveredcoveredcoveredcoveredcovered
coveredcoveredcoveredblind spotcoveredcovered
coveredcoveredblind spotcoveredcoveredcovered
coveredcoveredcoveredcoveredblind spotcovered
coveredblind spotcoveredcoveredcoveredcovered

Select a category to read its row.

Sample data. A dashed cell is a blind spot: the product is built to show where it cannot see.

Radar

Sample data
Market
12 of 1,240 sample concepts · by stage, then change
Follow
Every step after this one follows it

Followed through every stepCan be followedThe other rows open their trend page

Sample data. Twelve invented trends; each stage is worked out from its series by the written rule.

Trend page

Sample data

Sauna hat, Sweden

RisingUp +60%

One concept in every language: bastumössa · Saunahut · sauna hat · followed through every step

0125250week 202530354045Confirmed week 38

Weekly indexWhat the season alone would give

week 45

Week 45: index 247. The season alone would give about 109, in a band from 98 to 120. +127% against the season: above the band its season gives.

Why “Rising”
  • Last four weeks against the four before+60%
  • What the season alone would give+13%
  • Difference from the season+47 points
  • Weeks in a row above its season10
  • Confirmedweek 38

Three weeks or more above its season, and growing more than 10 points faster than the season alone would.

Sources
  • Encyclopedia readingSame movement
  • Search interestSame movement
  • Retail listingsSame movement
  • Supplier cataloguesSame movement
  • AudienceAdults 25 to 54
  • CategoryWellness

Sample data. The rule is written down, so anyone can work the stage out again.

Being built to
  • Collect demand, supply and price signals every day, in every market served.
  • Keep the history of every signal, with its source and the date it was read.
  • Join one concept across sources and languages into a single trend, staged by a written rule anyone can reproduce.

02Products

Explore product ideas.

From attention to products: these product ideas are illustrative, and buyer demand and supplier prices are not yet verified. A trend is not sellable; a product is. Here the trend that is followed becomes five product ideas, two dated offers each. Choose one and the steps after this follow it.

Products and offers

Sample data
Currency
Fixed illustrative rate: 1 EUR = 11 SEK
Sample shopSweden · wellness · minimum margin €22 per order, after ads
TrendSauna hat · Rising
Goods
€21.00
Minimum
30 units
Delivery
12 days
Dated
3 Nov
Checks
4 of 5

Still open: return terms

Goods
€13.50
Minimum
150 units
Delivery
42 days
Dated
28 Oct
Checks
3 of 5

Still open: sample checked, return terms

Sample data. Supplier A and Supplier B are invented, and so are their prices.

Being built to
  • Turn a rising product into concrete products with compared, dated offers.
  • Fit every decision to the shop that asks: its market, its prices, its minimum margin.

03Suppliers

Find the manufacturer.

Two dated offers for the product you chose, side by side: the unit price, the minimum order, the days to deliver and the checks each supplier passes.

Two dated offers

Sample data

Sample data. Supplier A and Supplier B are invented, and so are their offers and dates.

Being built to
  • Compare dated supplier offers, each with what it was checked for.
  • Coming: supplier offers read from the shop’s own quotes, kept with their dates.

04Delivery

Bring it in, and get it out in time.

Inbound freight, duty and delivery to the customer, and whether the goods arrive before the season’s first strong week.

Freight, duty and timing

Sample data
  • Inbound freight per unit€1.30
  • Duty per unitNone inside the EU
  • Delivery to the customer, per order€5.40
  • Days to deliver, Supplier A12
  • Days to the first strong week of its season21
  • In time for the seasonYes

Coming: each carrier’s last order day before the season, read from dated delivery cut-offs.

Sample data and estimates. Carrier cut-offs are coming; nothing here is a carrier’s promise.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Being built to
  • Price inbound freight and duty from published rates, with their dates.
  • Coming: each carrier’s last order day before the season, from dated delivery cut-offs.

05Costs

Every cost, the ads included.

The supplier’s unit cost, freight, duty, VAT and fees are built, line by line, into the landed cost and the margin. Then the ads it takes to win each order come off, and the shop’s minimum is checked against what is left. The money at risk is the first order’s stock and the test’s ad budget.

Landed cost and margin

Sample data
€79
Selling price, VAT includedYour price
€79.00
VAT, 25% of the net priceRate
−€15.80
Refunds and lost parcels, 8.5% of net salesAssumption
−€5.37
Card fee, 1.6% of the price + €0.17Rate
−€1.43
Dispute allowanceAssumption
−€0.10
GoodsQuote · 3 Nov
−€21.00
Inbound freightEstimate
−€1.30
DutyNone inside the EU
€0.00
Delivery to the customerEstimate
−€5.40
Margin per order, before adsWorked out
€28.60
Planned ad cost to win each orderAssumption
−€3.00
Margin per order after ads
€25.60At or over the shop’s minimum of €22 · an estimate: 3 of its inputs are estimates
Landed cost
€27.70Goods, inbound freight, duty and delivery
Cost cap for the goods
€24.60The most the goods may cost and still leave the minimum after ads. The goods cost €3.60 less than the cap.
Money at risk
€1,395The first order’s stock, 30 units at €22.30 each, and an ad budget of €726: 660 paid visits at €1.10 each, before the first sale is known

Sample data and sample rates. The arithmetic runs in your browser, line by line. Where a quote is missing the product gives an estimate and says so.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Line by line

Sample data

Supplier: Supplier A · delivery 12 days Sample data · minimum order 30 units Sample data

Landed-cost build-up for one order, line by line
LineLabelPer order
Customer paysThe selling price, VAT included.Sample data€79.00
VAT to the state25% of the net price, the market’s rate. Rate dated 2026-11-12.Sample data−€15.80
Net revenuePrice less VAT.Sample data€63.20
Product from the supplierUnit cost on the supplier’s dated offer.Sample data−€21.00
Shipping to usFreight from the supplier, per unit.Sample data−€1.30
Import dutyNone inside the EU.Sample data€0.00
Delivery to the customerSample price. For comparison, DHL Paket publishes 14.49 EUR for a 2 kg parcel to SE (read 2026-11-11). Published price, read from the carrier’s page; VAT treatment not yet confirmed.Sample data−€5.40
Payment fee1.6% of the price plus a small fixed fee.Sample data−€1.43
Returns and lost parcels8.5% of the net price, allowed for.Sample data−€5.37
Dispute allowanceA fixed allowance per order.Sample data−€0.10
Margin per orderWhat is left of one order. Its inputs are invented, so it is not a result.Sample data▲ €28.60

Lead-time check: in time. Delivery takes 12 days Sample data; 21 days are left before the season’s first strong week Sample data. Estimate The weeks a trend needs are an estimate dated 7 October 2026, and “in time” is never a guarantee.

Sample data and dated estimates. Each line is labelled with which it is. The arithmetic is the landed-cost engine’s, run in your browser.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Being built to
  • Work out landed cost and margin by code, line by line. A language model never does the arithmetic.
  • Count the ads it takes to win each order, and the money at risk before the first sale.
  • Say for every line whether it is a quote, an estimate from our own dated tables, or a sample.

06Decision

One page you can sign.

The decision on one page, before any money moves: every figure worked out by code, every claim with a source and a date, and the minimum margin checked after ads. Select a figure to see where it comes from.

Decision brief

Sample data

Sample

Decision brief · Sweden · week 45

Sauna set: hat, seat pad and linen towel

Ready for a test, once one question is answered.4 of 5 gates passed. Open: fibre content on the label, in Swedish?

Evidence

Interest in the sauna hat in Sweden is over four weeks, where the season alone would give . The rise was confirmed in , and sources show the same movement.

Product and supply

Sauna set: hat, seat pad and linen towel from Supplier A (inside the EU): for the goods at , delivered in . A is on file.

Economics

At the margin before ads is per order, of net sales; after the of ads planned to win each order, is left. The money at risk before the first sale is known is . The goods may cost up to before the shop’s minimum after ads of is missed.

Rules and timing

3 of 4 questions are answered and one is open. Delivery in is in time for the season. None of this is legal advice.

The test

Screen with and a cap of ; stop under 3 orders. Confirm with ; pass at or more. At most is at risk.

  1. 1Reading and search interest, Sweden, weekly index to week 45
  2. 2What the season alone would give, from market statistics
  3. 3Supplier A, offer dated 3 Nov
  4. 4Supplier B, offer dated 28 Oct
  5. 5The shop’s own price and minimum margin
  6. 6Landed-cost arithmetic, rates dated 1 Nov
  7. 7Test plan, version 1, locked before launch
Worked out by code: every figure on this page
Signed by a person before it is sent

Sample data. The figures follow what you chose in step 02.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Being built to
  • Write briefs a shop owner can read in two minutes and verify.
  • Take every number from code and check the words against the numbers.
  • Have a person sign before anything is sent.

07Evaluation

Test before money moves, then record the outcome.

How much to spend, when to stop and what counts as a pass are written down and locked before launch; the pass rule comes from the margin in step 05. Then what was sold, what it cost and what was left, and what a stop saved. Each outcome is recorded, so the next decision starts from what the last one showed. A sample run is never a result. In the sample record, most tests stop, and the record says so.

Ready to test?

Sample data
4 of 5 gates passedEach gate keeps its own reason. There is no combined score.
  • EvidencePassedRise confirmed in week 38; 4 of 4 sources agree
  • SupplyPassedTwo dated offers; Supplier A passes 4 of 5 checks
  • MarginPassed€25.60 per order after ads is at or over the shop’s minimum of €22
  • RulesOpen1 of 4 questions still open
  • TimingPassed12 days to deliver: in time for the season

Sample data. The questions are examples of what the product is built to ask. None of it is legal advice.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Test plan

Sample data

Will Swedish shoppers buy “Sauna set: hat, seat pad and linen towel” at €79, at an advertising cost the margin can carry?

The rules gate in step 07 is open. A plan like this would not be released until it passes.

Phase 1

Screen

Visits
160
Spend cap
€176

Stop if there are fewer than 3 orders.

Phase 2

Confirm

Visits
500
Spend cap
€550

Pass at 84 orders or more. Advertising per order is then €6.55 at most, so the margin of €28.60 before ads still leaves the shop’s minimum of €22.00 after them.

€176 if it stops at the screen€726 at mostA visit costs €1.10Break-even: 16.7% of visits must order to keep the minimum

Stop rules, written before launch

  1. The spend cap of a phase is reached.
  2. Fewer than 3 orders at the end of the screen.
  3. The supplier can no longer deliver within 12 days.
  4. Orders in the shop stop matching what the ad platform records.

Locked before launch · version 1 · a change makes a new versionYou run the test and hold the budget. TrendFalcon never spends.

Sample run

The sample run opens when all five gates in step 07 have passed.

Sample data. The price of a visit is a sample assumption; the plan is worked out again whenever the margin changes. The sample run reads invented orders, the same at every price.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Watch

Sample data

After a decision the watch goes on: the same written rule reads its trend again, every week. Here it reads the tests of the sample record that passed.

  • Knitting needle caseSweden · decided after its test of week 40ConfirmedRisingStill rising 20 points faster than its season
  • Linen bed throwSweden · decided after its test of week 30FadingPeakedAbove its season, but no longer growing faster than it
  • Balcony planter boxSweden · decided after the shop’s own test of week 28, run with the evidence gate openTime to stopCoolingFalling 14 points faster than its season explains

Sample data. Each decision is a test of the sample record that passed, read again from the series of its trend.

Your sample run in money

Sample data

Run the plan above and its units, revenue, cost and profit appear here.

Sample data. Nothing is sold in the preview.

Outcome record

Sample data
  • This test: Sauna set: hat, seat pad and linen towelSwedenStage at entry: RisingNot run yetWaiting
  • Knitting needle caseSwedenStage at entry: RisingScreen: 5 ordersConfirm: 23 orders, needed 20Spent €726Passed
  • Window bird feederGermanyStage at entry: RisingScreen: 0 ordersConfirm: not reachedSpent €176Stopped
  • Linen bed throwSwedenStage at entry: RisingScreen: 6 ordersConfirm: 26 orders, needed 22Spent €726Passed
  • Cork yoga blockGermanyStage at entry: RisingScreen: 4 ordersConfirm: 12 orders, needed 18Spent €726Stopped
  • Dog cooling matGermanyStage at entry: PeakedScreen: 1 orderConfirm: not reachedSpent €176Stopped
  • Balcony planter boxSwedenStage at entry: PeakedScreen: 5 ordersConfirm: 21 orders, needed 19Spent €726Passed
  • Boot drying rackSwedenStage at entry: SeasonalScreen: 2 ordersConfirm: not reachedSpent €176Stopped
  • Padel grip tapeSwedenStage at entry: CoolingScreen: 1 orderConfirm: not reachedSpent €176Stopped

Invented tests, by the stage each began at. 4 of the 8 are the shop’s own tests, run with the evidence gate open, and the record keeps every test.

Entered as “Rising”
2 of 4 passed
Entered as “Peaked”
1 of 2 passed
Entered as “Seasonal”
0 of 1 passed
Entered as “Cooling”
0 of 1 passed

8 tests · 3 passed · €3,608 spentScreen: 160 visits, €176; fewer than 3 orders stops the test. Confirm: 500 more visits, €550; the orders needed come from each product’s margin.

Sample data. Eight invented tests, read with the same caps as the plan above.

Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

Dated record of calls

Sample data

A call is a dated yes-or-no question about a later week, written down and dated before the answer exists.

  • Call 0031Sauna hat, Sweden: still above its season in week 44?Dated week 38 · check code 9f2c…a41eHeld
  • Call 0029Knitting needle case, Sweden: still above its season in week 45?Dated week 39 · check code c05e…77f2Held
  • Call 0026Padel grip tape, Sweden: still above its season in week 45?Dated week 37 · check code 41b7…0c9dDid not hold
  • Call 0018Linen bed throw, Sweden: still rising faster than its season in week 36?Dated week 30 · check code e8a1…5b30Did not hold

Sample data. The real record starts empty, and is built to be published whatever it shows.

Sample result: not run yet. Run the plan in step 07 to see whether the example goes on or stops.

Being built to
  • Design every test before it starts: its size, its stop rules and the orders it must reach.
  • Tie the orders a test must reach to the margin, so a pass means the product can pay for its own customers.
  • Watch every decision afterwards: confirmed, fading, time to stop.
  • Join every outcome to what was known on the day the decision was made, and publish a record of calls anyone can check.

ResearchReal figures, with sources

Why this should exist.

The product views above use sample data. Nothing in this section does. These are findings from our own research and from public statistics, each with its source and its limits.

  1. 23.6%of EU enterprises with ten or more people employed sold online, in Eurostat’s 2025 survey

    Finding 01

    Every shop asks the same question.

    In Eurostat’s 2025 survey, 23.6% of EU enterprises with ten or more people employed had sold online in the year before, up from 17.2% in the 2014 survey. Those that sell goods all face the same decision, season after season: what to sell next. Sweden is where we start: Statistics Sweden counts 1,808 internet and mail-order retailers with one to nine employees, and 21,820 with none, a separate count of smaller firms.

    Sources: Eurostat, dataset isoc_ec_eseln2, indicator E_AESELL, EU, survey years 2025 and 2014, read on 10 November 2026; Statistics Sweden, Business Register, table FDBR07N, industry code 47.91, year 2025, read on 25 October 2026. Limits: Eurostat’s survey covers enterprises with ten or more people employed, in every activity except agriculture, mining and finance, so not only retail; each survey asks about the calendar year before it, and the 2024 survey gave 23.8%; the Swedish count is of registered companies, not active shops, and a shop that also has a physical store is counted elsewhere.

  2. Attentionis all a public signal measures

    Finding 02

    Attention is not sales.

    A view, a search and a listing measure attention or supply. None of them is an order. The step from attention to sales can only be measured where the orders are: in a shop. That is why the loop ends in a test and an outcome.

    Source: TrendFalcon research report of 9 October 2026, on every source then in use.

  3. 31.0%of rise signals were followed by lasting interest in German (31.5% in Swedish)

    Finding 03

    Most rises do not hold.

    In our own test on Swedish and German Wikipedia topics, a single rise signal was followed by interest at least a quarter above its earlier level, from the third week to the tenth, in 31.0% of cases in German and 31.5% in Swedish. A spike in a chart is usually not followed by interest that holds a quarter above its earlier level.

    Source: TrendFalcon attention protocol v2, run on 8 November 2026 on Wikimedia pageviews (CC0): 31.0% of 1,852 signals in German and 31.5% of 944 in Swedish. The plan was registered before the data was fetched. Limits: reading interest only, never sales; 52 topics chosen by hand; signals for one topic are close together in time, so their outcome windows overlap and they are not independent; one run. The rule in the preview above is a sample, not the rule tested here.

  4. 0 of 3published percentages came with a way to check them

    Finding 04

    Tools say more than they show.

    On 31 October 2026 we read the home pages of nine trend and retail-analytics tools. Three publish a percentage for how often their own model is right. On the pages we read, none gave the method, the test period, the baseline and the sample a reader would need to check it.

    Source: TrendFalcon research note R-1, 31 October 2026. Limits: home pages only, one visit each; two more sites refused the request and are left out; a method may sit on a page we did not open.

  5. €0.72of every euro is left before the goods are paid for

    Finding 05

    Landed cost and margin decide.

    By our estimate, a Swedish shop taking card payments keeps about €0.72 of every euro a shopper pays, before goods, freight and delivery.

    How we worked it out

    Of every €100 a shopper pays:

    1. −€20.00VAT, at 25% on top of the net price.€80.00 left
    2. −€6.80Refunds and lost parcels, assumed at 8% and 0.5% of net sales.€73.20 left
    3. −€1.60The card fee, at 1.6% of the price.€71.60 left

    That is about €0.72 of every euro.

    Each order also carries about €0.27 that does not depend on the price: the fixed part of the card fee, €0.17, and a small allowance for disputes. Customs duty is not in the €0.72: the EU charges a flat €3 customs duty per tariff category on parcels worth €150 or less sent to consumers from outside the EU. On a €39.90 item we work that out as about 9% of the price before VAT.

    Sources: TrendFalcon margin formula per market, 1 November 2026 (2.98 SEK an order); the Swedish Tax Agency for the VAT rate; Shopify’s Swedish price list, read 11 October 2026, for the card fee of 1.6% plus 1.85 SEK; European Commission, “Ensuring fairness and safety: €3 customs duty on low-value parcels”, 29 June 2026. Limits: the sources give the fee and the allowance in kronor, and the euro amounts convert them at this page’s fixed illustrative rate; refunds, lost parcels and disputes are assumptions that a shop’s own figures replace; the duty is temporary; the 9% is our own arithmetic.

    Amounts in euro. Fixed illustrative rate: 1 EUR = 11 SEK.

  6. Freeis what a basic list of trends costs today

    Finding 06

    Discovery is free. Outcomes are not.

    Lists of what is trending are given away: TikTok’s Creative Center costs nothing, and neither does Meta’s Ad Library, where anyone can see which ads are running. Paid tools sell deeper lists. What our searches did not find is a tool that keeps a dated record linking the evidence before a test to its stop rule and its result, rejected candidates included. That record is the asset TrendFalcon is built to grow, from our own tests first and then from customers who consent. It is never sold.

    Source: TrendFalcon strategy research, 7 October 2026. Limits: the gap rests on what our searches did not find, which is weaker than a count.

The honest-numbers rule.

Our first pre-registered test on Wikipedia attention passed: a rise that still held two weeks later went on to hold in 59.1% of cases in German and 53.9% in Swedish, against 31.0% and 31.5% for a rise read on a single day. That is attention, not sales, from one run. Predicting the next rising trend is the goal we are building toward, and each claim is tested in public first. These are the rules the product is built to.

  1. Numbers come from code.A language model may write the sentence. It never does the arithmetic.
  2. Three readings, never one score.Attention, commercial readiness and the quality of the evidence are shown side by side and never merged.
  3. An estimate is labelled as an estimate.Where a figure is missing the product gives its best estimate, says so, and never passes it off as a measurement.
  4. Every figure has a source and a date.If we may not show a source, we do not show it.

Where it goes

Our goal: every shop. Every market. A record that compounds.

We are building TrendFalcon to be where a shop decides before it buys: the next product chosen on evidence, costed line by line, tested by a plan and signed by a person.

  1. Today

    The first pieces are running.

    Daily collection is running, the cost arithmetic exists as code and a private workspace is in use.

  2. Next

    One workspace for the whole decision.

    From first signal to signed decision, in the shop’s own language.

  3. Then

    Every market, every category.

    And a record of outcomes that compounds with every test, for every shop that shares in it.

Our goal: no shop has to guess what to sell next.

Get early access for your shop.

The early-access list is open. Access dates and pricing are not set yet.

We store your email and when you joined, only to tell you about early access. Joining does not create an account or place an order. Privacy