RSA pinned headlines: lock positions or let the model mix them?

Pinning gets recommended a lot, and I'm not sure it helps as much as people assume. Pin a headline to position 1 and the system has fewer combinations to work with, so you get predictable copy but a much smaller pool of tested pairings. Say a dental clinic pins its brand name and a "Same-Day Appointments" line to position 1. That can work fine, but the other headlines may barely get impressions, so the test you thought you were running never really happens.

My view is to pin only when there's a legal or brand requirement, and otherwise look at asset-level performance after a few weeks. How do you handle low-volume accounts, where the asset ratings tend to sit at Low for a long time?

Agreed on the legal and brand case. On low-volume accounts, though, a "Low" rating mostly means the asset got little traffic, not that it's bad. Ratings only compare assets inside the same ad, so a headline that served 40 impressions can sit at Low even if it converts well whenever it does show. Before cutting anything based on ratings, look at impressions and conversions per asset side by side. If the impressions are tiny, the rating isn't telling you anything yet.

For a small clinic doing maybe 15 conversions a month, I'd skip the in-ad test entirely. Duplicate the RSA in the same ad group, pin the brand line in one copy and leave the other unpinned, and give both the same budget for a fixed window. That gives you a comparison you can actually read, even if it's slow.

I don't think the duplicate-RSA test makes sense at 15 conversions a month. Split across two ads, that's maybe seven conversions each, and any gap between them will mostly be noise. Same budget doesn't guarantee an even split either, since delivery leans toward whichever ad the auction thinks will do better, so the two copies won't get equal traffic.

For a clinic that size, I'd keep the brand line pinned as a guardrail and judge the unpinned version on cost per conversion over a couple of months.

Mira, I agree that a split at seven conversions an ad won't tell you much. A rule I'd use: before you start, write down what you'd change for each possible result. If you'd keep the pin either way, skip the test. Pinning a brand line is cheap to reverse, so for a clinic I'd just run it, watch cost per booking over two months, and switch back if it drifts up. Two months also smooths out the seasonal booking spikes clinics get, which a short window can mistake for a real difference.