Why Last Click Is Especially Wrong for Glaucoma

Glaucoma is the textbook case for why last-click attribution fails. The consideration window between a patient first searching for information about elevated pressure or visual-field loss and actually scheduling a specialist consultation routinely runs 4 to 12 weeks. The patient researches the disease, reads about treatment options, looks at SLT and MIGS comparisons, gets a referral letter from the OD, then finally clicks a branded query for the specialist’s name and books. Last Click attribution credits 100 percent of the conversion to that final branded click and zero percent to the educational searches that did the actual conviction work.

The downstream effect on Smart Bidding is consequential. The bid algorithm sees brand campaigns producing all the conversions and awareness-stage campaigns producing none, and reallocates budget accordingly. Awareness-stage campaigns get starved over time, the educational content stops appearing for the early-funnel queries, and the patients who would have entered the funnel through the awareness-stage entry point never enter it at all. The brand campaign keeps reporting strong CPA because it is harvesting demand that the awareness campaigns are no longer being funded to create. The pattern is invisible from a Last Click report and obvious from any path-level view.

How Data-Driven Attribution Actually Distributes Credit

DDA uses machine learning to assign fractional conversion credit across every ad interaction in the conversion path, based on each touchpoint’s measured contribution. The model compares paths that included a given touchpoint against paths that did not, and assigns credit in proportion to the measured lift the touchpoint produced. Last Click gives 100 percent credit to the final click. Rule-based models, first click, linear, time decay, and position-based, were removed from Google Ads entirely in 2024.

For a glaucoma practice the redistribution looks like this. A path that includes a symptom search, a treatment-comparison search, and a final branded search no longer credits the branded search alone. The model assigns some fraction (often 20 to 35 percent) to the symptom search, some fraction to the treatment-comparison search, and the remainder to the brand search. Smart Bidding sees the awareness-stage campaigns producing real, fractional conversion credit rather than zero, and reallocates budget upward toward the auctions that actually contribute to the conversion path.

The credit redistribution is the point of the migration. The brand campaign’s reported CPA may look slightly worse after migration because some of the credit it was harvesting now flows to the campaigns that did the real conviction work. Practice leads accustomed to evaluating brand campaign efficiency in isolation often misread the change as a regression; the right read is that the report is now describing the funnel honestly. The companion offline conversion imports piece covers the upstream EMR data that gives DDA richer paths to optimize against.

The Default Change and the Documented Lift

DDA is the default for all new conversion actions in Google Ads as of 2026 (ALM Corp, March 2026). Google removed rule-based models in 2024 (foundational). Healthcare accounts migrating from Last Click to DDA typically see 10 to 20 percent credit redistribution to top-of-funnel campaigns and a small but real improvement in Smart Bidding efficiency, usually in the 5 to 15 percent range on CPA at constant volume.

The lift size varies with the structural state of the account. Accounts with already-strong awareness-stage campaigns see modest gains because the campaigns were already getting some signal. Accounts that had let awareness-stage campaigns wither over years of Last Click attribution see larger gains because DDA reactivates funding to the parts of the funnel that had been starved. Glaucoma accounts skew toward the second category, which is why the migration tends to produce noticeable rather than marginal improvement. For glaucoma practices running 30 to 50 conversions per month per campaign, DDA reaches statistical reliability within 4 to 6 weeks; below that threshold the model falls back to position-based weighting and the lift compresses to the 3 to 6 percent range. The companion lead quality blind spot piece covers the parallel quality-side audit that should accompany an attribution migration.

Five Red Flags That Attribution Is Out of Date

The following patterns indicate the attribution layer is misconfigured and is distorting Smart Bidding. First, conversion actions created before 2023 are still on Last Click. The default migration to DDA only applies to new conversion actions, not to legacy ones, which means most established accounts have a mix of attribution models in active use, with the legacy actions dragging on bid decisions. Second, no attribution setting review has happened in the last 18 months. Attribution lives in a screen most agencies and practice leads never visit, which is how 24-month-old settings persist.

Third, brand campaigns receive most of the credit, which is the canonical Last Click artifact in a long-consideration funnel like glaucoma. The brand campaign sits at the bottom of the funnel and harvests every conversion, while the awareness-stage campaigns that produced the demand show zero credit. Fourth, top-of-funnel campaigns (awareness, symptom search, treatment comparison) are undervalued in budget and impression share because the report says they do not convert. They do convert; the report is wrong.

Fifth, the agency cannot explain the current attribution setting in one sentence. If the answer is a 90-second story about historical reasons and forthcoming reviews, the agency does not have a current view of the setting. The right answer is a single sentence: “We are on Data-Driven Attribution for all conversion actions, migrated on this date, and the brand-versus-non-brand split has been reviewed in the last 30 days.” The companion server-side tagging piece covers the upstream conversion-data integrity work that DDA depends on.

The Concrete Migration Sequence

Open the conversion actions screen in Google Ads. Check the attribution model for each conversion action. Most accounts will find a mix: newer actions on DDA, older actions on Last Click, and possibly a few on rule-based models that were grandfathered. Migrate every Last Click and rule-based action to DDA using Google’s migration tool, which compares both models for 30 days before committing the change. The 30-day comparison gives the practice a side-by-side view of how credit is redistributing before the bid algorithm starts spending against the new model.

After the comparison window, commit the migration. Smart Bidding recalibrates over 3 to 4 weeks. During that recalibration window, expect CPA volatility and resist the urge to intervene. The algorithm is relearning bid patterns against the new credit distribution, and any manual adjustment during the learning window slows the relearning process. Annotate the migration date in Change History so subsequent performance reviews can cleanly bridge before-and-after.

Review Smart Bidding performance over the 30 to 90 day post-migration window. Expect a 5 to 15 percent efficiency improvement in accounts with multi-touch conversion paths, which most glaucoma accounts are. If the improvement does not materialize, the issue is usually upstream of attribution, conversion data integrity, missing offline imports, or thin-volume campaigns that DDA cannot help. The companion PPC benchmarks piece covers the post-migration baselines.

Specialty Vision’s Take on Attribution as a Bidding Input

Our view, attribution is not a report; it is a bidding input. The setting determines which conversions get credited where, and Smart Bidding uses that allocation for every bid decision. Last Click systematically starves awareness-stage campaigns of credit and budget, pulling them out of rotation while the bid algorithm reallocates toward brand campaigns that harvest the resulting demand.

In glaucoma, where consideration windows run 4 to 12 weeks, Last Click is a decade-old default that no longer matches buying behavior. The patient who clicks the brand campaign has often touched the practice’s content five or six times in the preceding two months. Over a 12-month horizon the awareness campaigns shrink and the brand campaign hits diminishing returns because the demand source has been defunded.

Migrate to DDA and document the change. For broader audit context, see our 2026 PPC audit playbook.