outricher/barcelona-tech-hiring-map/seed-match
Research · round 1 data · 2,825 seed companies · Barcelona · 25 October 2026
2,073 of the 2,825 seed companies have a Barcelona team in the people data — and 27,312 more employers the tech flag was missing
The joint city-map project sent a seed list of 2,825 companies with a Barcelona HQ, office or team, built from public sources, and asked three things: match it to our company ids, return per-company aggregates, and use it to widen the universe beyond the industry-label tech flag. This page is the result. Aggregates and company names only — the archive with every field goes to the project team; a sample is linked at the end.
seed companies found with at least one person in Barcelona — 73% of 2,825; 1,450 of them with five or more
without a Barcelona team in the data: 401 have a record under that name but nobody in Barcelona on it, 351 have no usable record
found with one to four people in Barcelona — the "fewer than five" flag the brief asks for
companies in the archive: the seed companies plus every Barcelona employer with ≥ 5 people and the 2023+ AI founders — 29,242 with a career history read
01 · Coverage check (brief §5)
How many of each segment we found, and how many of those have a team in Barcelona
Found means a company record met the seed name (or the evidence domain) and at least one person in Barcelona sits on it, today or at the snapshot. "Record, nobody in BCN" means the name met a record no Barcelona person is linked to — often a namesake abroad; the id goes back as a candidate, not as an answer. "People in Barcelona" is the sizing's count: a profile whose locality names Barcelona and whose current position is at the company.
| Segment | Seeds | Found, people in BCN | Share | of them < 5 | Record, nobody in BCN | Not found | People in BCN |
|---|---|---|---|---|---|---|---|
| Agency / studio | 570 | 357 | 63% | 162 | 125 | 88 | 7,920 |
| Barcelona-born | 525 | 431 | 82% | 84 | 53 | 41 | 19,984 |
| Tech Barcelona member (company) | 465 | 307 | 66% | 146 | 97 | 61 | 17,318 |
| Deeptech / health | 320 | 257 | 80% | 65 | 26 | 37 | 6,488 |
| Tech Barcelona member (unclassified) | 186 | 111 | 60% | 34 | 33 | 42 | 7,803 |
| International hub | 150 | 135 | 90% | 15 | 6 | 9 | 14,032 |
| Gaming | 132 | 96 | 73% | 26 | 12 | 24 | 4,513 |
| AI | 128 | 105 | 82% | 33 | 11 | 12 | 1,661 |
| Ecosystem | 116 | 75 | 65% | 28 | 20 | 21 | 4,248 |
| Big tech / corporate hub | 96 | 82 | 85% | 8 | 5 | 9 | 52,261 |
| Martech / SaaS lookalike | 90 | 72 | 80% | 20 | 12 | 6 | 2,439 |
| IT services & BPO | 42 | 40 | 95% | 2 | 1 | 1 | 20,830 |
| Barcelona HQ (founded abroad) | 5 | 5 | 100% | 0 | 0 | 0 | 386 |
| All segments | 2,825 | 2,073 | 73% | 623 | 401 | 351 | 159,883 |
Confidence of the 2,825 matches
four classes · the archive carries the reasons and the alternatives per rowConfidence grades the name match, not the team. High: the name or the evidence domain met the record and people sit there. Medium: the presence is thin or a second candidate is close. Low: alias, stem, prefix or the people's own spelling only. 401 medium and low matches have nobody in Barcelona and count as "record, nobody in BCN" above.
What met the record
2,474 matched rows · a row can meet in several ways186 rows had a second candidate within half a point of the best — they carry ambiguous = true and the runner-up. 5 company ids are claimed by two seed rows (a brand and its parent, a hub and its company); both rows say so.
02 · Matching
How a seed name meets a company record, and where it does not
Four forms are tried together and scored, never first-hit-wins: the name as written (diacritics kept and stripped), the name without spaces, punctuation and legal suffix, the name people wrote as their employer, and the domain of the client's evidence URL when it is the company's own. When none of them finds anyone in Barcelona, a record that adds one or two words to the name is tried ("EPAM" → "EPAM Systems"), and a record with nobody in Barcelona loses to a close one with people there. Pages of one company fold into one line.
DisambiguationA plain name meets several records (ten "Miro", four "Framer", thirteen "Perk"). People in Barcelona, HQ against the seed's segment, listed size against the client's S/M/L and the evidence domain decide; the runner-ups stay in the row. Records that share the winner's website, or its exact name and city, are folded into one company — company_ids lists them, people_now_by_page splits the count.
The known defectTravelPerk's company id carries a "Perk" (Boston) record in the company dimension — an upstream collision found in the sizing; the client's list already spells the row "Perk (TravelPerk)". Positions carry the right name. The pilot's eleven hand-resolved page groups (Adevinta's four pages, Typeform's three, Paack's three …) are a fixed override table.
Why 752 have no Barcelona team here401 meet a record under their name that no Barcelona person is linked to; 351 have no usable record at all. One structural reason: 37% of the current positions of Barcelona people (693,291 of 1,882,747) carry neither a company id nor a company name — a small company can be invisible here while its people are on the platform. Both groups go back in not-found.csv with the candidate and the reason; they are the client's name-mismatch review list.
Found, but fewer than five623 seed rows have one to four people in Barcelona today. Agencies (162) and Tech Barcelona members (180) dominate: many are real but small, some list their people under a different page. The brief's own caveat ("S" = under 50, size is an estimate) holds in the other direction too.
The named tests
The brief names companies the widened universe should catch (remote-first teams with a small Barcelona cluster), seven it believes have no Barcelona office, two it excluded as closed and four it flagged. Each resolved the same way as a seed row.
| List | Company | Match | Matched record | People in BCN | Snapshot | Hires 2022–25 | HQ | Founded |
|---|---|---|---|---|---|---|---|---|
| remote first | Bïrch | high | birch (ex Revealbot) | 17 | 19 | 21 | Barcelona, ES | 2016 |
| remote first | Revealbot | high | birch (ex Revealbot) | 17 | 19 | 21 | Barcelona, ES | 2016 |
| remote first | Manychat | high | Manychat | 142 | 140 | 163 | Austin, TX, US | 2016 |
| remote first | Plata | medium | Plata Card | 8 | 2 | 2 | Mexico City, MX | 2022 |
| remote first | Framer | high | Framer | 13 | 12 | 22 | Amsterdam, North Holland, NL | — |
| remote first | Miro | medium | Miro | 16 | 12 | 14 | San Francisco, California, US | 2011 |
| remote first | Pipedrive | low | Pipedrive | 0 | 0 | 0 | — | — |
| remote first | Wrike | medium | Wrike | 8 | 5 | 13 | San Diego, California, US | — |
| remote first | JetBrains | medium | JetBrains | 6 | 5 | 4 | Amsterdam, NL | 2000 |
| remote first | Palta | medium | Palta | 0 | 2 | 3 | — | — |
| remote first | Exness | medium | Exness | 9 | 7 | 6 | Limassol, CY | 2008 |
| remote first | inDrive | medium | inDrive | 8 | 5 | 7 | Mountain View, California, US | — |
| remote first | Social Discovery Group | medium | Social Discovery Group | 14 | 9 | 13 | Malta, MT | 1991 |
| no office found | Xsolla | medium | Xsolla | 5 | 4 | 6 | Sherman Oaks, CA, US | 2005 |
| no office found | Bolt | medium | Bolt | 9 | 7 | 9 | Tallinn, EE | 2013 |
| no office found | Pleo | high | Pleo | 34 | 29 | 45 | København N, Hovedstaden, DK | 2015 |
| no office found | EPAM | medium | EPAM Systems | 75 | 47 | 55 | Newtown, PA, US | 1993 |
| no office found | Booksy | high | Booksy | 28 | 24 | 34 | France, Poland, Spain, UK & US, US | 2014 |
| no office found | Gismart | low | Gismart | 0 | 0 | 0 | — | — |
| no office found | Toloka | medium | Toloka | 9 | 9 | 11 | Amsterdam, NL | 2014 |
| closed excluded | Smilegate Barcelona | high | Smilegate Barcelona | 24 | 32 | 23 | Barcelona, Cataluña, ES | — |
| closed excluded | Stenn | high | Stenn | 18 | 15 | 57 | London, GB | 2015 |
| flagged | Viaplay | medium | Viaplay Group | 32 | 44 | 70 | Stockholm, SE | — |
| flagged | Paradigm Shift | high | Paradigm Shift | 6 | 5 | 5 | Barcelona, ES | 2024 |
| flagged | Epsilon | medium | Epsilon Technologies | 21 | 31 | 21 | Madrid, ES | 1993 |
| flagged | Defense Prime | not found | — | 0 | — | — | — | — |
"People in BCN" is today's count by profile locality; "Snapshot" is the positions-based headcount at the spring 2026 snapshot; "Hires 2022–25" counts Barcelona positions started in those years. On the "no office found" list, Xsolla (5), Bolt (9), Pleo (34), EPAM (75, as "EPAM Systems"), Booksy (28), Toloka (9) have five or more people in Barcelona today by profile; whether that is an office or distributed hires is for the client's sources to say.
03 · Per-company fields (brief §2)
Every field the brief lists, where it comes from, and how often it is filled
The archive carries them per company in companies.jsonl and next to each seed row in seed-match.csv. Fill rates below are for the 29,781 delivered companies and for the 2,469 seed companies among them.
| Field (brief §2) | Where it comes from | Filled, all delivered | Filled, seed companies |
|---|---|---|---|
| People in Barcelona today | profile locality + current position (people table) | 29,781 / 29,781 | 2,469 / 2,469 |
| Headcount 31 Dec 2019 → 2025, hires / departures | dated positions at the company, Barcelona rule | 29,242 / 29,781 (98%) | 2,163 / 2,469 (88%) |
| Function mix · seniority · AI-title share | regex over titles of the current Barcelona team | 29,242 / 29,781 | 2,163 / 2,469 |
| HQ country · local vs hub | company record (HQ country with a confidence score) | 29,717 / 29,781 (100%) | — |
| Founded year | company record | 22,248 / 29,781 (75%) | 1,911 / 2,469 (77%) |
| Last round (type, date, amount) | company record (funding block) | 3,443 / 29,781 (12%) | 825 / 2,469 (33%) |
| Industry labels, raw | company record (new taxonomy + legacy string) | 29,346 / 29,781 (99%) | — |
| Where the Barcelona team came from | previous position of each current team member, country from its location text; share of known and share of all staff | 29,242 / 29,781 | 2,163 / 2,469 |
| Street address · coworking check | company record read by id upstream; address against the hub addresses | 22,546 / 29,781 (76%) | 34 of 224 coworking labels confirmed by address |
| Company page URL | company record (slug) | 29,681 / 29,781 (100%) | — |
| Closed / acquired | operating-status field — empty for every Spanish record | 0 | 0 |
Function mix of the current Barcelona team, seed companies
180,593 positions open at the snapshot · the brief's seven bucketsEleven functions are measured and collapsed to the seven of the brief; both are in the data. A title with no function word lands in "other" (inside the last bucket).
Where the current Barcelona team came from, seed companies
180,593 positions · previous position's locationTop origin countries abroad: GB 1,287 · FR 1,206 · DE 1,018 · AR 1,017 · IT 893 · US 724 · NL 544 · IE 507. Per company in origins.csv; countries listed from 2 people.
Function and origin are measured on the current Barcelona team of the seed companies with a career read (2,220 companies, 180,593 open positions). Per company, the AI-title share is shipped twice: by positions at the snapshot and by the people-today view.
04 · The widened universe (brief §3)
Four lists the industry-label tech flag does not reach
The sizing flagged 5,555 tech employers with five or more people. The brief asked for four more groups, each with its own rule and floor; a company can sit on several lists. Coral bars are companies the tech flag does not carry.
Agencies and studios
2,639 companies · 45,209 people in BarcelonaThe brief's labels (design services, graphic design, advertising, marketing services, online audio and video, animation, computer games, and their legacy spellings) or a digital-marketing / web-design / studio phrase in the description · ≥ 5 people in Barcelona. Neighbouring labels we added (PR, media production, photography, translation, market research, film) are marked label_set = extended.
2,042 under the brief's labels, 597 under the neighbouring ones. 2,161 of 2,639 sit outside the sizing's tech flag — the gap the brief describes. Labels: Advertising Services 1,096 · Design Services 209 · Media Production 195 · Marketing Services 148 · Computer Games 136 · Public Relations and Communications Services 123.
Remote-first and relocated product teams
7,256 companies · 119,764 people in BarcelonaAny HQ · ≥ 5 people whose profile is in Barcelona · founded 2010 or later. The list is wide on purpose: it is the one that catches a foreign company's small Barcelona team.
1,457 have a foreign HQ (shown above), 4,600 are outside the tech flag. The named tests — Manychat, Framer, Miro, Wrike, JetBrains, Exness, inDrive, Social Discovery Group — are in the table below.
New AI companies, founded 2023 or later
463 companies · 1,910 people in BarcelonaFounded 2023+ · ≥ 2 people in Barcelona (the brief drops the floor) · AI/ML words in the name or description, or ≥ 15 % AI/ML titles.
86 of them already have five or more people.
Relocated-team signal
148 companies · 1,937 people in Barcelona≥ 3 people in Barcelona (the brief's floor for this group) · headcount on 31 Dec 2025 above 2021 · ≥ 30 % of the current Barcelona staff previously worked — in any earlier position — in RU, UA, BY, KZ, AM, GE, CY, IL or RS, and at least two such people (one person is not a team signal).
82 of them have five or more people and at least three such staff (strong). 44 more sit between 20 % and 30 % (not flagged, in the data as relocation_ever.share_of_staff).
05 · Relocated teams
Where a Barcelona team was built from the nine countries the brief names
The brief's wording, as written: a company with three or more people in Barcelona whose team grew after 2021 and where at least 30 % of the current Barcelona staff previously worked in RU, UA, BY, KZ, AM, GE, CY, IL or RS — in any earlier position, the company's own office there included. We add one guard: at least two such people. Countries are read from the location text of each earlier position. The narrower view (only the job right before Barcelona) is shown next to it.
| Company | People in BCN | Staff (snapshot) | 2021 → 2025 | Previously in the nine | Share of staff | Previous job there, of known | HQ | Top origins | Tech flag |
|---|---|---|---|---|---|---|---|---|---|
| Semrush seed #28 | 405 | 410 | 162 → 403 | 193 | 47% | 81 of 250 | US | RU 184, CY 9, RS 9, GE 5, BY 4 | yes |
| Manychat seed #19 | 142 | 140 | 44 → 138 | 76 | 54% | 17 of 76 | US | RU 71, AM 13, UA 3, CY 2, BY 2 | yes |
| EPAM Systems | 75 | 47 | 41 → 48 | 21 | 45% | 7 of 26 | US | RU 12, UA 8, BY 1 | yes |
| Tripledot Studios seed #498 | 73 | 75 | 20 → 74 | 27 | 36% | 11 of 53 | GB | BY 14, RU 11, CY 5, UA 1 | yes |
| Appodeal, Inc. seed #41 | 51 | 60 | 14 → 55 | 26 | 43% | 6 of 43 | US | RU 14, BY 10, CY 2, AM 2, UA 2 | yes |
| Workato seed #35 | 47 | 55 | 41 → 53 | 21 | 38% | 6 of 28 | US | RU 19, CY 4, UA 1, BY 1 | yes |
| hackerone hunterdm | 40 | 36 | 0 → 34 | 30 | 83% | 18 of 24 | MX | RU 28, CY 4, AM 3, GE 2, RS 2 | no |
| Tabby | تابي | 33 | 24 | 0 → 23 | 21 | 88% | 13 of 15 | SA | RU 19, AM 5, GE 1 | yes |
| Avito | 25 | 17 | 6 → 18 | 15 | 88% | 4 of 5 | RU | RU 15, AM 2 | yes |
| Constructor | 22 | 16 | 3 → 14 | 9 | 56% | 1 of 9 | US | RU 9 | yes |
| Plarium | 19 | 27 | 19 → 27 | 13 | 48% | 1 of 14 | IL | UA 13, IL 1 | yes |
| birch (ex Revealbot) seed #45 | 17 | 19 | 7 → 19 | 6 | 32% | 2 of 11 | ES | RU 6, BY 1 | yes |
| Sumsub | 17 | 16 | 2 → 14 | 7 | 44% | 1 of 11 | CY | RU 7, GE 1, BY 1, RS 1 | yes |
| ID R&D | 17 | 13 | 6 → 13 | 9 | 69% | 2 of 9 | US | RU 9 | yes |
| RingCentral | 16 | 10 | 7 → 10 | 4 | 40% | 0 of 3 | US | RU 2, BY 1, UA 1 | yes |
| Innoget | 16 | 11 | 8 → 11 | 4 | 36% | 0 of 10 | ES | GE 4 | yes |
| CommerceGate | 16 | 24 | 15 → 24 | 11 | 46% | 5 of 17 | ES | UA 11, CY 1 | no |
| BETBY seed #515 | 15 | 7 | 0 → 6 | 4 | 57% | 0 of 2 | MT | RU 4, AM 2 | yes |
| Fundraise Up | 15 | 7 | 2 → 7 | 6 | 86% | 3 of 5 | US | RU 6, GE 1 | yes |
| Emerging Travel Group | 15 | 13 | 4 → 13 | 9 | 69% | 5 of 8 | US | RU 9, CY 1 | yes |
| Luxoft | 15 | 12 | 9 → 12 | 5 | 42% | 2 of 9 | CH | UA 3, RU 2 | yes |
| Social Discovery Group | 14 | 9 | 3 → 9 | 5 | 56% | 2 of 6 | MT | RU 5, RS 2, GE 1 | yes |
| Combination AB | 14 | 10 | 7 → 10 | 3 | 30% | 1 of 10 | SE | UA 2, RU 1 | yes |
| MY.GAMES | 14 | 8 | 5 → 8 | 5 | 63% | 2 of 4 | NL | RU 4, CY 1 | no |
| Raiffeisen Bank Russia | 14 | 10 | 7 → 10 | 8 | 80% | 4 of 5 | RU | RU 8 | no |
| PandaDoc | 14 | 11 | 3 → 10 | 4 | 36% | 0 of 6 | US | BY 3, UA 1 | yes |
| SoftServe | 13 | 5 | 2 → 5 | 2 | 40% | 0 of 2 | US | UA 2, RU 1 | yes |
| Smart Engineering | 13 | 13 | 7 → 13 | 4 | 31% | 0 of 12 | ES | RS 3, RU 1 | no |
| Laduchi Consult | 13 | 13 | 10 → 13 | 9 | 69% | 4 of 9 | ES | RU 5, UA 5, GE 2, KZ 2 | no |
| BidMachine | 12 | 13 | 2 → 14 | 5 | 39% | 0 of 11 | ES | RU 2, BY 1, AM 1, UA 1, CY 1 | yes |
Relocation share across every company with ≥ 3 people and a measured team
26,879 companies · share of current staff who previously worked in the nine countriesCountries among the flagged companies (a person counts once per country): RU 714 · UA 106 · BY 58 · CY 47 · GE 35 · AM 33 · RS 26 · IL 16 · KZ 8.
How to read itThe share divides by the whole current team, so earlier positions without a location count against it — a company below the bar is not "not relocated", it may simply have thin careers (career_located says how many staff have a located earlier position). Russian banks and similar employers whose staff live in Barcelona show up too: their HQ column says so. The full list of 148 with every origin country is widen-relocated.csv; the per-company country table for all delivered companies is origins.csv.
06 · Lookalikes (brief §4)
The thirty Barcelona employers nearest each anchor, by team mix, size, founding year and origin
Weighted Euclidean distance on z-scored features over 22,799 companies with ≥ 5 people and a measured team: the seven function shares (weight 0.6 each), log of the Barcelona team (1.3), log of the listed company size (1), founding year (1.5) and whether it is known (0.8), share hired from abroad (1), share from the nine countries (0.8), share whose previous job was in Barcelona (0.6), AI-title share (0.5). Weights are in the data so the client can re-cut.
Manychat
Manychat · 142 in BCN · 140 at the snapshot · founded 2016- Tripledot Studios73 · d 2.34
- ZeptoLab75 · d 2.60
- ID Finance171 · d 2.87
- Workato47 · d 3.17
- EPAM Systems75 · d 3.40
- Semrush405 · d 3.58
- Gaming Innovation Group17 · d 3.69
- REMY Robotics31 · d 3.83
- viaPhoton14 · d 3.93
- Quorum Software21 · d 4.23
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Semrush
Semrush · 405 in BCN · 410 at the snapshot · founded 2008- EPAM Systems75 · d 3.43
- Manychat142 · d 3.58
- ZeptoLab75 · d 3.71
- Tripledot Studios73 · d 4.34
- ID Finance171 · d 4.70
- Social Discovery Group14 · d 4.87
- Workato47 · d 4.88
- CommerceGate16 · d 5.03
- SENSOR2312 · d 5.29
- Antiaging Group Barcelona16 · d 5.36
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Bïrch
birch (ex Revealbot) · 17 in BCN · 19 at the snapshot · founded 2016- Olyn7 · d 1.98
- Tuscany Now & More Ltd.17 · d 2.48
- Mercuryo9 · d 2.86
- Quorum Software21 · d 2.98
- Gaming Innovation Group17 · d 3.11
- illumin EU & LATAM13 · d 3.41
- Amigo Gaming | Game Provider23 · d 3.45
- Biotech Tricopharming Research14 · d 3.47
- ADCE - Art Directors Club of Europe15 · d 3.63
- Bitio8 · d 3.63
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Preply
Preply · 557 in BCN · 564 at the snapshot · founded 2012- Glovo2,249 · d 2.07
- eDreams ODIGEO1,480 · d 2.08
- Allianz Technology913 · d 2.08
- MongoDB265 · d 2.16
- Revolut725 · d 2.19
- TravelPerk1,193 · d 2.23
- Amaris Consulting262 · d 2.23
- EcoVadis284 · d 2.28
- Veeva Systems291 · d 2.30
- N26506 · d 2.32
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Factorial
Factorial · 1,392 in BCN · 1,684 at the snapshot · founded 2016- TravelPerk1,193 · d 1.06
- Autodesk950 · d 1.99
- Revolut725 · d 2.09
- EcoVadis284 · d 2.27
- SAP1,379 · d 2.42
- Gartner†1,100 · d 2.42
- Amenitiz337 · d 2.43
- MongoDB265 · d 2.54
- Qonto232 · d 2.54
- Preply557 · d 2.62
Top 10 of 30; the full list with the feature vector is in lookalikes.csv. † founding year missing on the record (median imputed).
Typeform
Typeform · 198 in BCN · 216 at the snapshot · founded 2012- Bitpanda128 · d 1.33
- Lodgify239 · d 1.38
- Oxigent Technologies145 · d 1.47
- Stark Future SL130 · d 1.49
- Back Market91 · d 1.53
- Skyscanner226 · d 1.57
- Win Systems131 · d 1.59
- Worldsensing133 · d 1.62
- Capitole373 · d 1.62
- BIG - Bjarke Ingels Group104 · d 1.69
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Holded
Holded · 254 in BCN · 273 at the snapshot · founded 2016- SeQura337 · d 1.19
- Exoticca373 · d 1.20
- Qualifyze154 · d 1.24
- The Knot Worldwide391 · d 1.27
- Top Doctors203 · d 1.37
- Innovamat181 · d 1.39
- Holaluz272 · d 1.45
- Voxel, an Amadeus company157 · d 1.47
- Red Points306 · d 1.49
- Dogfy Diet152 · d 1.49
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Appodeal
Appodeal, Inc. · 51 in BCN · 60 at the snapshot · founded 2004- MIGx AG†17 · d 3.13
- Universal Diagnostics S.A.10 · d 3.74
- Mitek Systems81 · d 4.52
- Shalion65 · d 4.56
- zenital†20 · d 4.59
- OLX†46 · d 4.81
- Chartboost†22 · d 4.82
- NDT Global†60 · d 4.82
- Clarity AI27 · d 4.83
- Merkle Spain19 · d 4.84
Top 10 of 30; the full list with the feature vector is in lookalikes.csv. † founding year missing on the record (median imputed).
ZeptoLab
ZeptoLab · 75 in BCN · 105 at the snapshot · founded 2010- Manychat142 · d 2.60
- SENSOR2312 · d 3.48
- Leagues of Code12 · d 3.50
- viaPhoton14 · d 3.54
- Semrush405 · d 3.71
- Tripledot Studios73 · d 3.71
- ID Finance171 · d 3.74
- birch (ex Revealbot)17 · d 3.90
- Mercuryo9 · d 3.91
- Antiaging Group Barcelona16 · d 3.94
Top 10 of 30; the full list with the feature vector is in lookalikes.csv.
Plata: matched (Plata Card, seed #86 (medium)) but the Barcelona team is too small to measure a mix (8 people today, 2 at the snapshot) · Spaceboost: matched (Spaceboost, seed #318 (medium)) but the Barcelona team is too small to measure a mix (0 people today, 0 at the snapshot)
07 · City view from this scope
The seed companies as a group, year by year
A sanity view, not a new sizing: the 2,469 seed companies' Barcelona headcount rebuilt from dated positions, with the hires and departures that moved it. The full scope (29,781 companies) is in the archive per company and per year.
Barcelona headcount at the 2,469 seed companies on 31 December
positions dated and located by the rule · 2,163 companies with a career readThe snapshot column is the spring 2026 snapshot (open positions). Past years are undercounted: a person who left Barcelona and had no location on the position is not visible. Read the trend.
Hires and departures per year, the same companies
positions started · positions ended · 2025 is a partial year in the dataInk = hires, coral = departures. The 2024 and 2025 columns are the ones the brief asks for per company; they are in headcount-by-year.csv for every delivered company.
Whole scope on 31 Dec 2025: 795,270 positions in Barcelona at 29,242 companies (2019: 605,175). The sizing's city-level series counted 171,715 people today at 5,555 tech employers; this scope is wider (agencies, non-tech employers of the seed list) and the two are not meant to be added.
08 · Honest edges
What the data cannot say, and what it says only with a rule
Snapshot, not liveThe people data is a spring-2026 build. "Today" on this page and in the archive means that snapshot; hires in 2025–26 are undercounted relative to a live read, and a refresh is a re-run.
Past years are undercountedA person who left Barcelona and had no location on the position is invisible for the years before. 80% of the counted Barcelona positions carry their own Barcelona location; the rest are the profile-location fallback. Both counts are per company (loc_rules).
Closed / acquiredThe operating-status field is empty for every Spanish company on the record. The client's acquirer notes stand; nothing here contradicts or confirms them.
OfficesOne street address per company record and no coordinates. The coworking check compares that address with the hub addresses the client's own list spells out, plus Pier01, Parc Científic and Barcelona Activa's two front doors: 34 of 224 coworking labels are confirmed that way. Job postings carry coordinates and were not part of this round.
Catalonia without a cityA position located as "Catalonia" alone is counted as Barcelona (14,124 positions, labelled position_catalonia) — the client can drop them per company.
Rules, not measurementsFunction, seniority, AI titles, the country of a location text, the tech / AI / agency flags and the map score are rules written down in the README and recorded per company. Everything tagged measured is a count.
09 · What the project team receives
24 files, one source of truth
companies.jsonl is the dataset — one nested record per company; every CSV is cut from it by the build script and carries the ids to join back. Your seed # is in every row that has one. The sample below holds 25 full company records, the coverage table, the test-name table and two lookalike tables.
| File | Rows / size | What it is |
|---|---|---|
CLAUDE.md | 4 KB | reading rules for an assistant |
README.md | 13 KB | field guide with the counts |
companies.jsonl | 162.5 MB | one nested record per company — the source of truth |
postings/barcelona-postings.jsonl.gz | 101.7 MB | |
slices/companies.csv | 14.2 MB | flat view |
slices/coverage-by-segment.csv | 1 KB | the coverage check |
slices/function-mix.csv | 2.8 MB | function and seniority counts and shares |
slices/headcount-by-year.csv | 15.2 MB | company × year |
slices/hiring-by-company.csv | 8.6 MB | |
slices/lookalikes.csv | 56 KB | anchor × rank × company × distance × features |
slices/not-found.csv | 187 KB | seed rows without a Barcelona team: not found, or a record nobody in Barcelona sits on — with why |
slices/origins.csv | 5.6 MB | company × previous-position kind / country |
slices/poaching-pairs.csv | 3.8 MB | inbound and outbound employer pairs, from 3 people |
slices/postings-by-company-year.csv | 4.7 MB | |
slices/postings-city-by-month.csv | 5 KB | |
slices/sector-year.csv | 18 KB | |
slices/seed-match.csv | 1.7 MB | one row per seed #, our ids and fields next to it |
slices/segments-scoring.csv | 4.1 MB | segment labels and the map score |
slices/test-names.csv | 4 KB | the named test companies |
slices/widen-agencies.csv | 396 KB | list §3.1 |
slices/widen-new-ai.csv | 76 KB | list §3.4 |
slices/widen-relocated.csv | 37 KB | list §3.3 |
slices/widen-remote-first.csv | 1.3 MB | list §3.2 |
table2-company-year-function-seniority.csv.gz | 20.3 MB | Table 2 in long form |
Next steps(1) The not-found and record-only rows are the name-mismatch review: a corrected spelling or a page URL per row re-runs the match. (2) Table 3 (pseudonymous moves) and Table 4 (postings with coordinates) are the next pull once the aggregates look right. (3) A quarterly refresh is a re-run of the same scripts on a fresh build.
ReproducibilityTen scripts in the task folder (00-seed … 09-dim) rebuild every file and this page; the page asserts the implied sums before it is written. The figures behind each exhibit are in _data/figures.json next to this page.