How DentalSignal builds the dataset

Sources

The dataset is compiled from a rotating set of public sources, weighted toward those with structured deal reporting:

• Group Dentistry Now, DSO Deal Roundup monthly recap series (2022 to present). Primary source: most deals first appear here, with acquirer / target / location explicit in the paragraph structure.
• Becker's Dental + DSO Review, news section. Used to backfill and cross-verify.
• Direct DSO press releases, pulled when referenced by either of the above.
• Industry publications, e.g. Dental Practice Management News, when a deal is cited.

Every deal in the public preview has at least one source_url field that resolves to the original published reference. Open it and verify directly.

Extraction

Pages are scraped on a regular cadence and parsed with a hybrid pipeline:

1. Paragraph segmentation: each article is split into paragraphs that typically describe one transaction. Most curated roundups (Group Dentistry Now in particular) follow a one-paragraph-per-deal pattern.
2. Verb anchoring: a regex matches against 25+ M&A verbs (acquired, joined, partnered with, welcomed, etc.), then captures the surrounding context.
3. Entity recognition: acquirer and target are extracted from the matched span using positional heuristics (acquirer typically subject, target typically object).
4. Geo + date enrichment: location strings and dates are normalized from the paragraph or the article header.
5. Confidence scoring: see below.

Confidence scoring

Each extracted deal gets a confidence score between 0 and 100. The public preview shows only deals at 70 or higher.

Score is weighted by:

• Acquirer presence in known-DSO list (+25): we maintain a list of 60+ known DSOs and 21 active dental PE firms. Match means high confidence the acquirer is real and dental-active.
• Target name structural match (+15): name follows typical practice-name pattern (proper noun + "Dental", "DDS", "Family Dentistry", etc.).
• Location string parsed (+10): a real US city / state is present.
• Date specificity (+10): exact month captured (not just "recently").
• Deal verb specificity (+5 to +15): "acquired" / "purchased" score higher than "partnered with" / "joined".

A score under 70 typically means a fragment of a paragraph was misread (e.g., "the company also recently added two locations" with no clear acquirer). Those are filtered out of the public dataset but stored internally for re-review.

What we exclude

• Mergers between equals where no acquirer is named.
• Greenfield openings (a DSO opening a new location is not an M&A event).
• Internal leadership changes, board appointments, fundraising rounds (unless tied to a specific practice acquisition).
• Specialty practices outside general dentistry if the source is ambiguous (orthodontic, endodontic, oral surgery practices are included when explicit).
• Deals where only one side is named (mystery seller or mystery buyer).

Known limitations

Honest caveats:

• Public-only bias: deals that never get reported are invisible to us. This skews toward larger / PR-active acquirers.
• Deal size: rarely published. We capture practice count as a proxy when available. Multiples and EBITDA are almost never disclosed.
• Lag: deals typically appear in roundups 30 to 90 days after close. Real-time data is not the promise; comprehensive backward-looking history is.
• Geographic skew: US-centric. International deals (UK, Ireland, Australia, Canada) are captured opportunistically but coverage is partial.
• Source coverage gaps: if Group Dentistry Now misses a month or recategorizes a deal, our pipeline may inherit that.

Reproducibility

If you want to verify a specific deal, click its source ↗ link on the preview page or any acquirer page. The link goes straight to the original publication. If you find a discrepancy, the dataset isn't a black box, the chain back to the source is one click.

For researchers and advisors who want the underlying CSV (deal-by-deal export with all fields), reach out, that's a planned premium-tier feature.