Research · Q4 2026 edition
Baltic LinkedIn Substance Index 2026
We read 2,638 LinkedIn company-page posts from 111 Baltic B2B companies over 90 days to 26 October 2026. We set aside 540 hiring and culture posts and scored the remaining 2,098 for specificity, expertise and how generic they read.
- Companies checked
- 111
- Companies ranked
- 91
- Posts read
- 2,638
- Data window
- 90 days to 26 October 2026
Published 2 November 2026 by Byro · Read the full report in HTML · Download the full report (PDF)
Key findings
The Baltic median company score is 60.8. Scores run from 0 to 100. Likes, comments and follower counts are reported for context and do not affect any score.
- r = 0.02
Likes and substance are unrelated.
The correlation between a post's likes and its substance score is 0.02. A popular post is no more likely to be specific than an unpopular one.
- 42 vs 13
Celebratory posts get the most likes and score lowest.
Celebratory posts have a median of 42 likes and an average score of 48. Expert posts have a median of 13 likes and score 75.
- 7 of 18
Energy and deep tech companies are overrepresented among Leaders.
Energy and deep tech and manufacturing companies are 18 of the 91 ranked companies and 7 of the 18 Leaders. 39% of them are Leaders, compared with 15% of all other ranked companies.
- 20%
One post in five is about hiring or company culture.
540 of the 2,638 posts (20%) were job ads, team events or employer branding. We show them separately and do not score them.
- 74 vs 45
Opening with a statistic helps. Text-only posts score lowest.
Posts that open with a statistic score 74 on average. Text-only posts score 45, the lowest of the main formats we compared.
- 67% vs 23%
Expert posts attract better conversation.
67% of comments on expert posts are substantive, compared with 23% on celebratory posts.
- 3.7 vs 0.9
Smaller companies get more engagement per follower.
The median small company gets 3.7 engagements per 1,000 followers. The median large company gets 0.9.
- 53 of 97
Most company leaders are quiet.
Of 97 confirmed leaders at the ranked companies, 53 posted nothing in the 90 days and 22 posted five or more times.
The Leaders tier
The Leaders tier is the top 20% of the 91 ranked companies. These 18 companies are the only ones we name. Other tiers are shown as counts.
| Rank | Company | Country | Sector | Score | Hit rate | Posts with numbers | Posts per week |
|---|---|---|---|---|---|---|---|
| 1 | Repsense | Lithuania | Software and SaaS | 81.9 | 82% | 79% | 2.8 |
| 2 | Light Conversion | Lithuania | Deep tech and manufacturing | 79.3 | 81% | 95% | 2.0 |
| 3 | Elering | Estonia | Energy | 75.9 | 89% | 78% | 2.3 |
| 4 | Ekspla | Lithuania | Deep tech and manufacturing | 73.6 | 85% | 92% | 1.3 |
| 5 | Nortal | Estonia | IT services | 73.4 | 64% | 50% | 3.5 |
| 6 | Salv | Estonia | Finance and banking | 73.1 | 100% | 67% | 0.5 |
| 7 | Primekss | Latvia | Deep tech and manufacturing | 73.1 | 100% | 100% | 0.6 |
| 8 | Bolt | Estonia | Software and SaaS | 72.3 | 81% | 69% | 3.0 |
| 9 | Milrem Robotics | Estonia | Deep tech and manufacturing | 70.7 | 53% | 28% | 2.8 |
| 10 | Edurio* | Latvia | Software and SaaS | 70.3 | 80% | 80% | 1.2 |
| 11 | Cybernetica | Estonia | IT services | 69.6 | 50% | 71% | 1.2 |
| 12 | Surfshark | Lithuania | IT services | 69.0 | 43% | 63% | 4.1 |
| 13 | Latvenergo | Latvia | Energy | 68.8 | 57% | 70% | 1.8 |
| 14 | Veriff | Estonia | Software and SaaS | 68.3 | 54% | 58% | 5.3 |
| 15 | SAF Tehnika | Latvia | Telecoms | 67.8 | 48% | 71% | 2.8 |
| 16 | Scoro* | Estonia | Software and SaaS | 67.8 | 48% | 43% | 1.8 |
| 17 | Sympower* | Estonia | Energy | 67.6 | 83% | 100% | 0.5 |
| 18 | Joom* | Latvia | IT services | 67.4 | 60% | 90% | 1.9 |
* Baltic-founded, headquartered abroad: Edurio (United Kingdom), Scoro (United Kingdom), Sympower (Netherlands), Joom (Portugal). Country is the country of founding. Hit rate is the share of a company's scored posts that reached 7 or more out of 9. Score is after small-sample adjustment (see Method), so a company with few posts can have a high hit rate and a moderate score.
The other tiers
We publish the other tiers as counts only. A further 20 companies were checked but not ranked, because they had fewer than 5 non-hiring posts in the window.
| Tier | Band | Score range | Companies | Estonia | Latvia | Lithuania |
|---|---|---|---|---|---|---|
| Leaders | Top 20% | 67.4 to 81.9 | 18 | 9 | 5 | 4 |
| Strong | Next 30% | 60.8 to 66.9 | 27 | 13 | 6 | 8 |
| Average | Next 30% | 57.0 to 60.8 | 27 | 8 | 6 | 13 |
| Generic | Bottom 20% | 47.6 to 56.9 | 19 | 6 | 7 | 6 |
Strong: 27. Average: 27. Generic: 19. Companies outside the Leaders tier are not named on this page or in the report.
Badge for Leaders. Companies in the Leaders tier may use the Leader badge on their site or LinkedIn page: SVG or PNG.
Awards
Every named award goes to a company in the Leaders tier and uses the same data as the ranking.
Best by country
- EstoniaEleringHighest company score among companies founded in this country.
- LatviaPrimekssHighest company score among companies founded in this country.
- LithuaniaRepsenseHighest company score among companies founded in this country.
Category awards
- Most evidence-backedPrimekss100% of posts contained a specific, checkable number, with more posts than Sympower, which also reached 100%.
- Most consistentRepsense82% of posts scored 7+/9 while posting 2.8 times a week (highest hit rate among Leaders posting 2.5+ per week).
- Best engagement for its substancePrimekss19.1 engagements per 1,000 followers, the highest in the index, with a top-10 score.
- Loud but emptyNot namedA Generic-tier company whose posts drew more likes than most companies in the index while scoring below the Baltic median.
By country
Country is the country of founding. The averages are close: the gap between the highest and lowest country is under two points.
| Country | Companies ranked | Average score | In Leaders tier | Top company |
|---|---|---|---|---|
| Estonia | 36 | 62.7 | 9 | Elering |
| Lithuania | 31 | 61.3 | 4 | Repsense |
| Latvia | 24 | 60.9 | 5 | Primekss |
By sector
Energy and deep tech and manufacturing companies are 18 of the 91 ranked companies and 7 of the 18 Leaders. 39% of them are in the Leaders tier, compared with 15% of the other 73 ranked companies.
| Sector | Average score | Leaders | Ranked | Posts with numbers | Hiring and culture |
|---|---|---|---|---|---|
| Energy | 66.7 | 3 | 6 | 78% | 28% |
| Deep tech and manufacturing | 65.6 | 4 | 12 | 60% | 16% |
| Transport and travel | 62.8 | 0 | 3 | 65% | 41% |
| Software and SaaS | 62.5 | 5 | 22 | 55% | 17% |
| Telecoms | 62.3 | 1 | 6 | 61% | 21% |
| IT services | 59.8 | 4 | 26 | 55% | 18% |
| Finance and banking | 59.1 | 1 | 15 | 64% | 24% |
Each company's sector comes from the industry on its LinkedIn company page, grouped into 8 sectors. We assigned 7 companies by hand where the industry was missing or unclear.
The hiring and culture percentage uses all posts; the share with specific numbers uses only scored posts. The one-company Consumer and other sector is omitted from this table and chart but included in the other totals.
What a strong post looks like
Two posts from Leaders-tier companies, one at the top of the scale and one near the middle. We link to each post on LinkedIn and do not reproduce its text.
- A top-scoring post
Light ConversionScore 100 of 100
Why it scored this way: names the research, the method and what it enables; specific technical detail; scored 100 by both the five models and the founder check.
Read the post on LinkedIn - An average post
VeriffScore 56 of 100
Why it scored this way: a short informational post that points to an article: relevant topic, little evidence in the post itself.
Read the post on LinkedIn
Method
Data
Public company-page and profile posts collected through third-party LinkedIn data providers (accessed via an API marketplace), not through Byro accounts. The window is 90 days to 26 October 2026. We checked 111 Baltic B2B companies and read 2,638 posts. The ranking uses company pages only.
Scoring
Each non-hiring company-page post was scored by a panel of five language models from different providers. Company names were masked before scoring, and each model scored without seeing the others' scores. The models were Claude Sonnet 5.5 (Anthropic); Gemini 3.5 Flash (Google); gpt-oss-120b (OpenAI open weights); GLM 5.3 (Zhipu); Kimi K2.7 (Moonshot).
Each model scored three criteria, S, E and G, from 0 to 3 each. A post's score is the median of the five models' S+E+G totals, scaled from 0 to 9 to 0 to 100. Agreement across the panel was Krippendorff's alpha = 0.74.
| Criterion | Scores 0 | Scores 3 |
|---|---|---|
| S, specificity and evidence | No concrete content, such as slogans or "excited to announce". | Verifiable claims with a source, data or first-hand detail a reader can check. |
| E, expertise | Unrelated to the company's field, such as holiday wishes or general motivation. | Shares real know-how or a point of view that only an expert in the field would write. |
| G, generic style (inverted) | Reads as a template or filler text with no substance. | A distinctive voice. The post would be missed if removed. |
Company score and tiers
A company's score is the average of its non-hiring post scores, pulled toward the overall mean. The overall mean is weighted as if it were 10 posts. This stops a company with a few strong posts from ranking above companies with a long record. Hiring and culture posts are shown separately and are not scored. A company needs at least 5 non-hiring posts to be ranked. Leaders are the top 20%, Strong the next 30%, Average the next 30% and Generic the bottom 20%.
Tone
The panel also labelled each post with one tone.
- expert
- explains how something works, what the team learned, or gives a reasoned view in its field.
- informational
- news or updates without much analysis.
- promotional
- product, event or offer promotion.
- celebratory
- awards, milestones, partnerships, 'proud to announce'.
- personal
- personal story or reflection.
- Hiring and culture
- job ads, team events, employer branding (shown separately, not scored).
Company size
Size is measured by LinkedIn followers. Small companies are those under 9,611 followers (smallest third). Large companies are those over 27,794 followers (largest third). Company substance score and engagement rate had a weak correlation of r = 0.16.
Comments
Comments pulled for every scored post with at least one comment (2,872 comments on 1,188 posts). Gemini 3.5 Flash classified each commenter from their public headline as own company staff, decision-maker at another organisation (exec, founder, director, head, VP, owner), practitioner, vendor or recruiter, or other; and each comment as substantive (adds a point, real question, disagreement, experience), social (congratulations, thanks, tags, emoji), self-promotion, or link only. Classification is automated and not hand-checked.
Company leaders
Up to three leaders per company (CEO or founder, CTO or CPO, head of marketing) found via LinkedIn people search and web search, then confirmed by an AI check of the company profile and search evidence. 97 leaders confirmed (80 high confidence, 17 medium); roles left empty where evidence was unclear. Leader names were masked before scoring. Individual leaders are not named on this page or in the report.
Checks
Panel scores were checked against a sample scored blind by Byro's founder, who did not see the model scores: 10 posts (r = 0.93), and against a separate blind AI judge on 50 posts (r = 0.80).
Post length
Longer posts score higher, for both people and models. The correlation between post length and panel score is r = 0.56. Long posts scored 38 points higher than short posts in the founder-scored sample and 34 points higher in the panel. We kept length in the score, because longer posts have more room for specifics. We also computed a length-adjusted rank. 15 of the 18 Leaders are also in the top 18 on that rank, and none is lower than 24th.
Limitations
- The index measures what a post says. It does not measure business outcomes, reach or leads.
- Impressions, clicks, leads and pipeline are not visible from outside a company, so we cannot say whether substantive posts produce more business.
- The window is 90 days. A company's position can change with a few posts, especially for companies that post rarely.
- The ranking covers company pages only. Posts by individual employees are not part of a company's score.
- Scores come from language models. They agree with each other and with a small sample scored by Byro's founder, but that sample is only 10 posts from one reviewer, and the models may share blind spots.
- Comment and leader classifications are automated and were not checked by hand.
- Data comes from third-party providers. Some posts may be missing, and like and comment counts change after collection.
- The company list is our selection of Baltic B2B companies. It is not a complete census. Some sectors have only a few companies.
- The associations on this page (tone, hooks, format, size, sector) are correlations across the sample. They do not show cause.
Request a correction
If you think a number about your company is wrong, email us with the company name and what you think is wrong. We review every request and publish corrections on this page.
Corrections log
No corrections yet.
We plan a Nordic edition in early 2027. Nominate a Nordic company.