How alpha beta structure works
Two campaigns run side by side on the same theme. The beta campaign uses looser match types and exists to discover what people actually type. The alpha campaign holds only the search terms that have already proved they convert, usually added back as exact match.
The routing is done with negatives. Every term promoted into alpha is added as an exact negative in beta, so beta can no longer serve it and the proven phrase is bought only through the campaign built for it. Alpha carries the larger budget and the tighter targets; beta gets a smaller exploratory allowance.
Why alpha beta structure was used
It was a sensible answer to a manual problem. When every bid was set by hand, you wanted your money going to phrases with a track record rather than to whatever a broad keyword had decided to match that week. Separating the two gave proven traffic its own budget and its own protection from experiments.
It also imposed a discipline most accounts lacked. Somebody had to read the search terms report on a schedule and decide what had earned promotion. That habit was often worth more than the structure enforcing it.
Where alpha beta structure goes wrong now
Close variant matching broke the routing. An exact negative blocks the phrase itself, but not every close variant of it, so a term you carefully promoted can still be matched in beta by a query that means the same thing. The two campaigns end up bidding against each other for one search.
Automated bidding broke the economics. A bid strategy learns from the conversions inside its own scope, and this structure deliberately divides those conversions across two campaigns while continuously moving the best ones out of one and into the other. Both sides are left learning from an unstable, incomplete sample.
The maintenance was never light either, and it grows with the account: negative lists to keep in step, duplicate ads to update in both places, and reporting that has to be recombined before it means anything.
What to do instead
Keep the habit and drop the scaffolding. Mine the search terms on a fixed rhythm, add genuinely new phrases as keywords where they deserve ads of their own, and put the wasteful ones into a shared negative keyword list. That is the part of alpha beta that still pays for itself.
Where you do want separation, separate by decision rather than by proof: brand against non-brand, one market against another, a high-margin service against a thin one. Each of those carries a different budget or a different target, which is something a campaign can act on. Being a proven phrase is not, because the bid strategy already knows which queries convert. If you inherit an account built this way, consolidate it in stages and watch the search terms closely as the negatives come out, which is the same care any search campaign restructure deserves.
