Share Classes Explained: GOOG vs GOOGL and Dual-Class Stock

Why one company trades under two tickers

A share class is a distinct series of a company's stock with its own bundle of rights: votes per share, dividend treatment, conversion terms. Most US companies issue a single class of common stock, so the question never comes up. Dual-class and triple-class structures exist because founders want to raise public capital without surrendering control, and the cleanest way to do that is to sell shares that carry less voting power than the ones you keep. The SEC's investor-education site has a plain overview of how common stock and its variants work.

Alphabet is the canonical example. Its Class A shares trade as GOOGL and carry one vote each. Its Class B shares carry ten votes each, are concentrated with the founders, and are not listed on any exchange. In 2014 the company executed a split that created a third series, Class C, which trades as GOOG and carries no votes at all. The economic claim on Alphabet's earnings is essentially identical across GOOG and GOOGL; the governance claim is not. You can see both classes side by side on the Alphabet stock page.

Berkshire Hathaway runs the other famous version. Class A (BRK.A) has never split and trades at a price most retail accounts cannot touch. Class B (BRK.B) was created in the 1990s as an affordable fraction of an A share and was split further in 2010; each B share now represents a small fixed fraction of an A share's economic interest and an even smaller slice of its vote. A shares convert into B shares at the holder's option, but never the reverse. When you look at Berkshire's own 13F portfolio, remember the filer itself is a two-class company.

One issuer, several CUSIPs

The identifier layer is where share classes stop being a corporate-governance curiosity and start being a data problem. A CUSIP's first six characters identify the issuer and the next two identify the specific issue, so two classes of the same company share a prefix but are different identifiers end to end (the full anatomy is in what is a CUSIP). The SEC's quarterly Official List of Section 13(f) Securities carries each reportable class as its own entry, which means institutional filings treat them as unrelated rows unless you join them back together yourself.

The same pattern repeats across the market: News Corp, Zillow, and Under Armour all list two classes under near-identical tickers, and plenty of companies (Meta, Ford) keep a high-vote class entirely off-exchange. Tickers make the relationship look obvious to a human (GOOG next to GOOGL) and completely opaque to a machine, because nothing in a ticker string encodes "these two are the same issuer." The CUSIP prefix does encode it, which is why serious holdings pipelines resolve identity through identifiers rather than symbols. Note the trap in the other direction too: a shared six-character prefix means same issuer, not same security, so prefix-level grouping is a deliberate aggregation step, never a dedup step.

How trackers double-count or merge wrongly

Share classes break holdings trackers in two opposite ways, and both produce numbers that look plausible.

Double counting. A fund that holds both GOOG and GOOGL files two 13F rows with two CUSIPs. A tracker that counts "funds holding Alphabet" by iterating rows will count that fund twice, inflating breadth-of-ownership stats. Leaderboards of "most widely held stocks" are especially vulnerable: each Alphabet class can rank separately, or worse, a name-level merge done inconsistently can rank "Alphabet" and "GOOG" as if they were different companies. Concentration math suffers the same way; a fund with 4% in each class has an 8% issuer position that never shows up if you only look at per-row weights.

Wrong merges and silent drops. The opposite failure is a pipeline that deduplicates on issuer name and keeps one row, throwing away the other class and understating exposure. String matching on names like "ALPHABET INC" versus "ALPHABET INC CL A" versus a filer's idiosyncratic spelling makes this worse, because 13F issuer names are free text typed by thousands of different filers. And per-share arithmetic across merged classes is quietly wrong: dividing a combined position value by one class's share count yields a fictitious price. Berkshire's extreme A-to-B price ratio makes that bug impossible to miss; Alphabet's near-identical class prices make it nearly invisible, which is more dangerous.

None of this is hypothetical filing noise. It compounds with the other known limitations of quarterly holdings data (how accurate is 13F data covers the rest), and it is the kind of systematic, structural error that survives spot checks because every individual row is correct. The data is right. The aggregation is wrong.

What SEC filings report, class by class

Every major ownership disclosure is class-aware, so your pipeline has to be too.

13F holdings are reported security by security, keyed on CUSIP, by institutional managers with over $100 million in covered US equities, due 45 days after each quarter end (for 2026: February 17, May 15, August 14, and November 16). A manager holding both Alphabet classes reports two rows, full stop. Across Arkolith's Q1 2026 dataset that granularity adds up: 1,824 institutional filers reporting 1.87 million long positions worth $53.7 trillion, with multi-class issuers like Alphabet and Berkshire appearing as distinct CUSIP-keyed line items throughout. The SEC's 13F FAQ documents the form's mechanics, and how to read a 13F filing walks the raw XML.

Schedule 13D makes class identity load-bearing: the 5% beneficial-ownership threshold is measured against a single class of registered equity, not the company as a whole, and the filing is due within 5 business days of crossing it. An activist can be economically large in an issuer while staying under 5% of any one class.

Insider forms are class-specific as well. A Form 4, due within 2 business days of the trade, names the exact security transacted, and conversions between classes show up as their own reportable events. Form 3 is due within 10 days of becoming an insider, Form 5 within 45 days of fiscal year end. Arkolith tracks 51,000+ insider transactions with this class detail preserved; the differences between the ownership forms are mapped in 13F vs 13D vs 13G vs Form 4.

Querying share classes without fooling yourself

Arkolith stores holdings the way filers report them: one row per CUSIP, each row carrying provenance back to the SEC EDGAR accession number it came from. Issuer-level rollups are something you do explicitly, not something the pipeline does to you. The practical workflow:

First, resolve the issuer and see every class it trades under:

curl -H "Authorization: Bearer YOUR_KEY" "https://arkolith.com/api/v1/search?q=alphabet"

Then pull a specific fund's holdings by CIK and aggregate classes yourself, keeping or merging them depending on the question:

curl -H "Authorization: Bearer YOUR_KEY" "https://arkolith.com/api/v1/funds/1067983/holdings"
curl -H "Authorization: Bearer YOUR_KEY" "https://arkolith.com/api/v1/funds"

The decision rule is simple. For economic questions (how much Alphabet exposure does this fund have, what is aggregate institutional ownership of the issuer) merge classes by issuer and sum values. For governance questions (who can influence a vote, who crossed a 13D threshold) keep classes separate, because the law does. For per-share math, never cross classes.

This matters double for AI agents. A model asked "how many funds own Google" will happily improvise an answer; grounding it in class-level rows with accession-number provenance is exactly the failure mode provenance-backed data exists to prevent. Setup takes a few minutes via the quickstart.

Frequently asked questions about share classes

Why does Alphabet have both GOOG and GOOGL?

Alphabet's 2014 split created a non-voting Class C share (GOOG) alongside the one-vote Class A (GOOGL), letting the company issue stock for compensation and acquisitions without diluting founder control. The founders' ten-vote Class B shares do not trade publicly. Both listed classes represent essentially the same economic claim on the business.

Are GOOG and GOOGL the same price?

No, they trade independently and the spread between them moves around, though it has generally stayed small. Class A's voting right tends to command a modest premium at times, and buyback activity can differ by class. For most economic analysis the two are interchangeable; for governance analysis they are not.

Does a 13F show GOOG and GOOGL separately?

Yes. 13F positions are reported by CUSIP, and the two classes have different CUSIPs (02079K305 for Class A, 02079K107 for Class C), so a fund holding both files two distinct rows. Any issuer-level view of "Alphabet ownership" is an aggregation you or your data provider performs after the fact.

Which share class should I use when counting institutional ownership?

Neither alone. Sum reported values across all classes of the issuer for economic ownership, and keep classes separate when the question involves voting power or regulatory thresholds like Schedule 13D's 5% per-class test. Counting rows without merging classes overstates the number of holders and understates position sizes.

This article explains public filings and data concepts. It is not investment advice.