Market Watch
Who should I sell to?
Buying companies in the market are found every week, their websites are read, and each is ranked with a 0–100 score that shows its reasons.
Weekly market watch for manufacturers and exporters
Turn marketsignals intoopportunities
Every week Personalitic scans your target market, finds the companies that could buy what you sell, scores each one’s fit and shows the evidence, and flags “time to offer” when one adds a product group close to yours.
TÜBİTAK-backed R&DHosted in the EUData processing addendum available
Shown with sample data.
Observed data, from official sources
- EU27 · monthly
- UK · monthly
- US · monthly
- Other 25 markets · annual
- Company websites · product groups
- 57markets with official trade data
- 31markets on a monthly series
- 1,228HS4 product headings
- Mon 06:00weekly report and e-mail, Istanbul time
You always know what is observed, what is inferred and what is simulated.
- Observedofficial trade data and company websites
- Inferredscores and signals computed from them
- Simulatedsynthetic buyer response
Synthetic personas are not real people or survey respondents; they are artificial profiles built to simulate how behaviour might shift under different conditions.
Why now
Markets change every week. Your team can’t check every week.
Buyers add products, open capacity and switch suppliers. Catching it today means hours lost between scattered spreadsheets, company websites and customs statistics.
Market Watch
When should I make an offer?
When a watched company adds a product group related to yours, a “Time to offer” flag appears with the reason and an evidence link.
Opportunity File
What should I invest in?
For each product × market, the official trade series, buying companies and synthetic buyer response come together in one file; rules decide, not the model.
Inside the product
See the product at work.
Short recordings from the application itself: a market watch through the week, a “time to offer” signal, the weekly report and the Opportunity File.
Recordings use sample data; the product interface is shown in Turkish.
The time-to-offer showcase, score distribution, sales funnel and latest changes on one screen.
The five score components, score history, product groups, company announcements and timeline.
A week-by-week change feed: time to offer, new product groups, new entries, competitor moves.
The report that freezes on Monday at 06:00: executive summary, numbers, funnel, changes and method.
The system’s recommendation, two threshold bars, market chart, drivers, buying companies and scenarios.
Market Watch
One week inside Personalitic.
You pick a market and a product. The system does the rest every week: it finds buying companies, reads their websites, scores them and on Monday morning shows you the ones flagged “time to offer”.
Choose the market, the product and the buying sector.
A four-step setup: market, one to five products, the sector that buys your product, and a watch threshold. The first scan starts the moment the watch is created.
- 57 markets
- Threshold 50–90 · 70 recommended
New buyer candidates arrive every week.
Each discovery searches for about 25 candidates; your customers and companies you dismissed are never suggested. The strategy alternates week to week: by buying sector, or lookalikes of your best companies.
- ≈25 candidates · 10 credits
- Empty discovery is refunded
Each company’s own website is read.
Watched companies are visited every week, pooled ones every four weeks. Products from the sitemap are grouped into HS4 product groups; the company’s own press page is assessed with a dictionary in eight languages.
- Respects robots.txt
- Visits cost 0 credits
Every company gets a 0–100 score, with reasons.
The model only gives judgements; code computes the total: product fit, relationship type, demand signal, scale and strength of evidence. Companies above the threshold enter the watch list, the rest wait in the pool.
- Five components, with reasons
- Competitors stay out of the sales list
The week a need appears, with its reason.
When a watched company adds a product group related to yours, a “Time to offer” flag appears: why now, which group, an evidence link. A first-contact draft is written in the market’s language; the system never sends it, you approve.
- 6-week window
- Silent for 30 days after an offer
The week’s report and e-mail are ready.
The week’s numbers, time-to-offer flags, new entries, competitor moves and sector demand freeze into one report at 06:00 Istanbul time (UTC+3) and go by e-mail to the people you choose. Met, offered, won: the sales funnel fills with your own data.
- Report · e-mail · PDF
- End-to-end win rate
Shown with sample data.
Use cases
The same weekly data, six different decisions.
Teams use Personalitic to answer their own question from the same data.
A first buyer list in a new market
A packaging maker entering Germany for the first time scans the market’s companies, ranks those above the threshold by product fit, website profile and synthetic buyer response, and adds the best ones to a watch in one click.
Outcome
A prioritised meeting list, decision-maker profile and first message draft
- Customer discovery
- Add to watch
Never miss the time to offer
A watched prepared-meal maker adds a new product group to its website. It arrives in Monday’s e-mail as “time to offer”, with the reason and an evidence link.
Outcome
A reasoned offer to the right company in the right week
- Market Watch
- Weekly e-mail
Evidence before investment
One file per product × market: the system’s recommendation, opportunity score, evidence score, three scenarios and next steps. Watch, hold or reject; the decision is logged.
Outcome
An evidenced, auditable investment decision
- Opportunity File
What should we build?
A customs series, an eight-driver outlook and synthetic buyer validation measure a product idea; without evidence, the idea never climbs the decision ladder.
Outcome
Measured opportunities and a “what could prove us wrong” list
- Trend and product opportunity
What else can an existing customer buy?
Your customer’s website is read regularly (weekly with a sitemap, otherwise every two weeks) and its product groups are mapped to HS4. Cross-sell analysis reads that catalogue together with your order history and ranks candidate products.
Outcome
Expansion candidates per account
- Customers · catalog
- Cross-sell
Watch competitors apart from the sales list
Companies that make the same product are classified as “Competitor”; they never enter the watch list or get a time-to-offer flag. Kept on watch manually, their product-group moves and announcements flow into a separate feed.
Outcome
Competitor portfolio moves in the weekly report
- Market Watch · Competitor moves
Opportunity File
One product, one market, one file.
Market and trend analysis and buyer discovery run together; the result is not a text to read but a board to decide on. Defined rules make the call, not the model.
Mono-material food packaging × Germany · 12 months
System recommendationInferred
VALIDATE
Validate with customer segments
Data support is strong; the synthetic buyer response is mixed, so the hypothesis is not supported yet. Next step: validate with customer segments.
- MONITOR
- VALIDATE
- PROTOTYPE
- PILOT
- INVEST
A score never decides alone: PROTOTYPE needs evidence at least SUPPORTED; PILOT and INVEST need STRONG.
For the next step, PROTOTYPE
- Evidence state
- VALIDATINGSUPPORTED — a positive shift in the synthetic audience
- Opportunity score
- 78≥ 70 · met
Opportunity score (model)Inferred78/100
- 55VALIDATE
- 70PROTOTYPE / PILOT
- 85INVEST
Evidence scoreVALIDATING75/100
- 40SUPPORTED
- 65STRONG
- Demand series45/45
- Last 12 months15/15
- Sourced outlook15/40
- Opportunity78 / 100Inferred
- EvidenceImport series + last 12 monthsOfficial data
- Cohort8 positive · 8 negative shiftsSimulated
- WindowFour quartersInferred
Scenarios
Three stories side by side — no probability is assigned.
Cautious
58/100
VALIDATEBaseline
78/100
VALIDATEAccelerating
86/100
VALIDATE
What could prove us wrong?
If the price of recycled PP granulate rises by more than 20%.
If the PPWR implementation timetable is postponed.
If Polish producers start cutting prices.
Counter-viewthe recycling premium may be a compliance wave, not lasting demand.
Your decision (in the product)
- Watch
- Hold
- Reject
- The cost is shown before the file opens: about 166 credits with default settings.
- Purchase intent, acceptance rate and willingness to pay are not measured — synthetic buyers only show shifts on demand axes.
- This is a sample file, not a real customer’s result.
Data
Not a forecast. A customs declaration.
Demand is measured from official foreign-trade statistics; what companies sell is read from their own websites. Periods that cannot be measured are never filled with estimates — they stay empty.
Germany · plastic packaging (HS 3923) imports
monthly · € million
0100200300400
400.62026-06
Source: official trade statistics · Germany imports, world total · updated 15 Sep 2026
- 2025 imports
- €4.30 bn
- world total · CIF
- Last 12 months
- +3.8%
- Jul 2025–Jun 2026 vs. the prior 12 months
- Türkiye’s share
- 2.8%
- of 2025 imports
Supplier share of 2025 imports
- PLPoland12.9%
- NLNetherlands11.6%
- FRFrance8.1%
- CHSwitzerland7.7%
- ATAustria7.1%
- CNChina5.5%
- TRTürkiye2.8%
Share = imports from partner / world total
Official trade data, 57 markets
- EU27Official statistics · including Türkiye, Ukraine and GeorgiaMonthly · EUR · CIF
- United KingdomOfficial statisticsMonthly · GBP · CIF
- United StatesOfficial statistics · where the data connection is enabledMonthly · USD · CIF
- The other 25 marketsHarmonized global data · Gulf, Asia, AmericasAnnual · USD · FOB
What a company sells, from its own website
Up to 10 product groups · 1–3 evidence links per group · checked every one or two weeks at no credit cost
- 01robots.txt
- 02Sitemap
- 03Product pages
- 04HS4 product groups
- HS 2106Prepared meals
- HS 2005Prepared salads
- +HS 2103Sauces & dips
Unmeasured ≠ zero: a period with no data stays a gap on the chart and is never filled with an estimate.
Analyses
One engine. Five decisions.
A real engine stands behind every question. Below you can open a nine-stage example run for who to sell to and what to build.
Prospect simulation
From thousands of companies to the ones worth selling to.
Path to the decision
CompaniesProduct fitPersonasPriority account
Watch an example run · 9 stages
- 01Target parsing. It works out what you sell and who you are looking for. Country, sector vocabulary, buyer profile and the end-customer clue are parsed apart.DE
- 02Company discovery. It finds the prospects that exist in the market. Real companies are found across company, sector, web and other corporate data sources.247
- 03Pre-screening. It focuses on the strongest candidates. Companies weakly related to your target are dropped; strong candidates go to deep analysis.82
- 04Web profiling. It gets to know each company more closely. The website, product portfolio and field of activity are analysed.20
- 05Product match. Does your product genuinely overlap with their business? The commercial fit between your portfolio and the company’s current products and activities is analysed.4
- 06Trend & seasonality. Each company is read against the market’s demand rhythm. Customs trade data, a trend estimate and seasonality enter the same analysis, showing when demand rises.3
- 07Persona simulation. The end customer’s likely reaction is simulated. Synthetic profiles representing the end customer are built; their reaction to a market signal you write can be measured.25
- 08Fit analysis. Each company’s sales potential is scored on several axes. Firmographics, product fit, demand signal and reachability, each on its own.85
- 09Decision briefing. The analysis turns into a sales action. Priority company, decision-maker profile, timing and a personalised message draft are delivered together.3
Example run
Stage 09 / 09
247 real companies · 25 synthetic personas · 3 scenarios · 1 decision
The figures are illustrative, not a measured outcome.
Future trend and product opportunity
From today’s evidence to tomorrow’s product opportunities.
Path to the decision
DemandForesightPersonasProduct opportunity
Watch an example run · 9 stages
- 01Scope parsing. First it defines which future we are looking at. Horizon of 3, 6, 9 or 12 months. Segment B2B, B2C or a custom definition.12 mo
- 02Demand seasonality. It reads the rhythm of demand through the year. Which quarter demand rises and falls in is inferred from a sourced reading; peaks and troughs feed the timing decision.Q3 peak
- 03Evidence measure. Not the strength of the trend — the density of the evidence. The evidence score is computed in code from the demand series, the last 12 months and sourced foresight; an unmeasured part is stated.75 / 100
- 04Demand series. Customs declarations, not inference. Real trade data measures the direction and change of demand; periods with no data are never filled in with estimates.HS 3923
- 05Driver sweep. Eight fundamental forces that could move demand are swept together. Regulation, demand, cost, competition, technology, trade policy, consumer and capital effects are weighed with both supporting and contradicting evidence.8
- 06Future synthesis. Not a single forecast — evidence-backed possible directions. Every future inference shows its direction, confidence level, supporting sources, counter-evidence and the conditions that would invalidate it.Up
- 07Product opportunities. There is no single leap from evidence to product. The measured evidence comes first, then the product features it calls for; only opportunities your production capability supports are kept.4
- 08Hypothesis validation. Personas read only the product hypothesis. The product hypothesis is tested on synthetic personas from identical starting conditions; behaviour change is measured, real purchase intent is never claimed.25
- 09Strategic verdict. The decision is built in code, not asked of the model. The strategic call is never left to model output alone; measured evidence, simulation results and explicit decision rules are weighed together.78
Example run
Stage 09 / 09
The figures are illustrative, not a measured outcome.
Portfolio expansion
What else can I sell?
Path to the decision
AccountCatalogueContextExpansion offer
Demand at the end of the chain
Will my customer’s customer want it?
Path to the decision
ProductDownstream buyersPersonasDemand call
Market entry
Which market should I enter?
Path to the decision
MarketsDemandCompetitionMarket call
Enterprise
For decisions that carry weight.
Tenant isolation
Every record lives in its own organization; every query runs within that scope.
Roles and permissions
Viewer · member · admin · owner; 33 capability-based permissions.
Audit trail
68 action types, with the user or API key, IP and request ID.
API keys
Full, read-only or scoped access; the secret is shown only once.
ERP and CSV import
Customers and orders via API or CSV; 1,000 rows per request.
Secure report sharing
Expiring links, 30 days at most; 3 views for an invitee; revocable at any time.
Hosted in the EU
Frankfurt; a separate or on-premise installation by contract.
KVKK and GDPR
Privacy notices are published (in Turkish); a data processing addendum can be signed.
How long does setup take, and what do I need?
A market, one to five products and the sector that buys them. Setup has four steps; with weekly discovery on, the system looks for new buyer candidates itself (10 credits per discovery), otherwise you add the companies. The first scan starts the moment the watch is created, and the weekly report arrives every Monday at 06:00 (Istanbul time).
Which markets are covered?
Official trade data covers 57 markets: the EU27, Türkiye, Ukraine, Georgia and the UK monthly; the US monthly where the data connection is enabled, otherwise annually; the other 25 markets annually.
How accurate are the results?
A simulation is a scenario, not a measurement. We claim no accuracy percentage; we give you the source of every claim, its epistemic type, and what argues against the recommendation.
Is a synthetic persona real customer data?
No. It is a model construct; it is not a real person, not a survey respondent, and no answer ever comes from a human.
Where is our data held?
Every record is scoped to its own organisation. The platform is hosted in the EU (Frankfurt); for enterprise customers a separate installation or installation on your own infrastructure is available by contract.
How does it work with our systems?
Customers and orders come in via API or CSV (1,000 rows per request). The weekly report arrives by e-mail and its printable version can be saved as a PDF; analysis reports are shared through expiring links.
Does Market Watch collect people or contact details?
No. No names, e-mail addresses or phone numbers are produced; only the products and announcements on the company’s own website are read. The system never sends the first-contact draft; you approve it.
What happens when credits run out?
You buy a credit top-up; the plan is never auto-upgraded. On a failed run the reserved credit is refunded automatically.
Pricing
Pricing fits your work. The credit table is public.
Pricing is shared as a quote.
Your usage volume, user count and data-residency needs set the price. One call is enough for a firm quote.
Get a quote- Opportunity File (market + buying companies)
- ≈166 credits
- Customer discovery (5 companies in depth)
- ≈97 credits
- Company discovery (≈25 candidates)
- 10 credits
- Customer website analysis
- up to 17 credits
- Market watch · weekly company visits
- 0 credits
- Failed run
- 0 credits
With default settings; the estimated cost is shown before every run, and failed runs are refunded.
What signals is your market sending?Let’s look together.
- We reply within two business days.
- In the call we look at your product and target market together.
- We show the weekly report and an Opportunity File on sample data.