Impression share lost to budget vs. rank: which to fix first

Lost impression share to budget and to rank need opposite fixes. Here is the rule for which to act on first, and how to price it before you spend.

Published September 21, 2026 · 8 minute read

Neither, until you check one thing. Act on whichever loss sits on a campaign that already hits its efficiency target, and leave the other alone until that campaign does.

Most advice says to fix rank first, on the logic that better Ad Rank stretches the budget you already have. The logic is sound. The prioritization built on it fails often enough to cost you a quarter, because impression share lost to budget and impression share lost to rank are not really competing for the same decision. They are two different constraints, and which one deserves money depends on a number that appears in neither metric.

What the two metrics measure

Google defines search impression share as "the impressions you've received on the Search Network divided by the estimated number of impressions you were eligible to receive." The two loss metrics split what's left:

• Search lost IS (budget): "the percentage of time that your ads weren't shown on the Search Network due to insufficient budget."
• Search lost IS (rank): "the percentage of time that your ads weren't shown on the Search Network due to poor Ad Rank in the auction."

Both definitions are Google's own, from the impression share metrics documentation. The three figures sum to 100%.

Two details in that documentation matter more than most people notice. Google reports lost IS (budget) at campaign level only, never for ad groups or keywords. And the denominator is a modeled estimate of the impressions you were eligible to receive. You are reading an approximation, which is fine, but precision past a point or two is false comfort.

Why "fix rank first" is the wrong default

The case for rank-first goes like this: improving Ad Rank lowers your effective CPC, so each dollar of existing budget buys more auctions, which creates headroom without new spend. Google supports the mechanism. It says outright that "even if your competition has higher bids than yours, you can still win a higher position at a lower price by using highly relevant keywords and ads," and it lists bid, ad and landing page quality, Ad Rank thresholds, auction competitiveness, and search context as the inputs to Ad Rank.

The mechanism is real. The prioritization built on it still breaks, for a boring operational reason: rank work is slow and budget work is instant.

Rank work means rewriting ads, restructuring match types, fixing landing pages, then waiting weeks for the auction to re-price you. A budget change is one field edit that shows up the next day. If a campaign already converts under target and loses half its eligible impressions to budget, sending the operator off to improve Quality Score first leaves known profitable volume on the table for six weeks in exchange for an uncertain efficiency gain.

The opposite error costs more. If a campaign runs 60% over its CPA target and loses impressions to budget, raising the budget buys more of a losing trade, faster.

The question that actually decides it

Stop sorting by loss type. Sort by whether the campaign earns money at its current efficiency.

The top-left cell has a name in our finding taxonomy: a budget-limited winner. Most accounts have one or two, and most audits never name them, because the audit reports lost IS (budget) as a problem instead of as a priced opportunity.

Turn the percentage into a multiplier

Here is the part that is almost always missing. A raw lost IS percentage is not an opportunity size. Convert it to a volume multiplier:

Multiplier = (impression share + lost IS budget) ÷ impression share

That is the factor your impressions could rise by if the budget constraint disappeared and your rank held. Take two campaigns.

Campaign A. 30-day spend $6,000, 120 conversions, $50 CPA against a $65 target. Impression share 38%, lost IS (budget) 47%, lost IS (rank) 15%.

Multiplier = (38 + 47) ÷ 38 = 2.24×

Campaign B. Same spend, same CPA, same target. Impression share 72%, lost IS (budget) 20%, lost IS (rank) 8%.

Multiplier = (72 + 20) ÷ 72 = 1.28×

Both campaigns "have a budget problem." Campaign A can roughly double. Campaign B has 28% left in it. A dashboard showing 47% and 20% in red presents these as one finding ranked by severity, and the raw percentage is exactly what hides the difference.

Now price Campaign A. At the current $50 CPA, the ceiling is 120 × 2.24 ≈ 269 conversions on roughly $13,400 of spend. Treat that as an upper bound and nothing more, because it assumes marginal auctions convert like your current ones. They will not. The auctions you are not winning today are the ones you bid least aggressively on.

What makes the decision safe is the gap between the $50 actual and the $65 target. You have 30% of CPA degradation available before you breach it. So the real question is narrow and testable: does marginal CPA stay under $65 as you scale?

Test it by stepping. Raise the budget 20% to 25%, let the campaign clear its conversion lag, read the marginal CPA, repeat. That stepping discipline is our own operating recommendation; Google publishes no such rule. The reason for it is that jumping straight to the theoretical ceiling changes the bid strategy's operating conditions, and the results stay unreadable for weeks afterward.

The thresholds we fire on

Where TrailMap surfaces impression-share decomposition as a finding:

• Multiplier at or above 1.25 with CPA at or below target: budget-limited winner, ranked by absolute conversion headroom.
• Multiplier at or above 1.25 with CPA above target: routed to the efficiency findings instead. This is not a budget finding and we do not present it as one.
• Lost IS (rank) at or above 20%: rank investigation, cross-referenced against Quality Score and match-type distribution.
• Below those: not surfaced.

The 1.25 cutoff is the one worth defending. Below 25% additional volume, the expected gain rarely justifies disturbing a bid strategy that is working, and at typical account volumes the result is hard to separate from normal week-to-week variance inside any reasonable measurement window. Treat it as a judgment call about signal, and tune it to your own volume. If your account is large enough to read a 10% move cleanly, move the cutoff down.

Fixing one moves the other

These two metrics are coupled, and the coupling breaks naive before-and-after reads.

Improve Ad Rank and you start winning auctions you previously lost. You now spend the same daily budget faster. Loss shifts from rank to budget, and if you were measuring the rank fix by watching lost IS (rank) fall, it looks like a clean win while the budget cap quietly becomes the binding constraint and total volume barely moves.

It runs the other way too. Raise a budget and you buy into auctions further down your bid curve, where your rank is weaker. Lost IS (rank) rises. That is the arithmetic of buying more, not a decline in ad quality.

So change one at a time, and measure total conversions and marginal CPA rather than the impression share metrics themselves. The IS metrics are good at telling you where the constraint sits and poor at telling you whether your fix worked.

Where this breaks down

Lost IS (budget) being campaign-level means you cannot decompose it any further. In a campaign holding both a strong and a weak ad group, the metric averages them, and the matrix above can point you at the wrong cell. Split the campaign before trusting the read.

A seasonal demand spike produces high lost IS (budget) that is not a structural finding. If query volume doubled last week, your budget did not become inadequate. Demand became temporarily larger. Check the trend before funding a step change, or you will permanently fund a peak.

Performance Max and Demand Gen do not decompose the same way, and Google notes that search exact match IS is unavailable for Shopping campaigns. And if your conversion tracking is wrong, every CPA in the matrix is wrong, which makes the whole framework confidently wrong. Fix measurement first.

Frequently asked questions

What is a good lost impression share to budget?

There is no universal number, and any post that gives you one is guessing. What matters is the multiplier, impression share plus lost IS (budget) divided by impression share, and whether the campaign hits its CPA target. A campaign at 20% lost IS (budget) with a strong CPA is a better opportunity than one at 50% that loses money.

Should I fix impression share lost to rank or budget first?

Whichever sits on a campaign already performing at or under its efficiency target. If the campaign is over target, neither, because both remedies scale a losing trade. Between two efficient campaigns, budget usually goes first, since it acts in a day while rank work takes weeks.

Why did my lost IS (rank) go up after I raised budgets?

Because you bought into auctions further down your bid curve, where your Ad Rank is weaker relative to competitors. This is expected and does not indicate a drop in ad quality. Judge the change on total conversions and marginal CPA.

Can I see lost impression share to budget by ad group?

No. Google reports lost IS (budget) at campaign level only. If you need that detail, the campaign has to be split so the budget boundary matches the segment you want to read.

Is impression share an exact measurement?

No. The denominator is Google's estimate of the impressions you were eligible to receive rather than a count of them. Treat movements of a point or two as noise, and act on direction and magnitude.