Why Your New-to-Brand Rate Drops When You Scale Retargeting

The mechanism, not a mystery

New-to-brand percentage at the account level is a weighted average of every campaign's own NTB rate, weighted by how much each campaign spent. Retargeting audiences — people who viewed a product, added to cart, or bought before — are, by construction, mostly people who already know the brand. Prospecting audiences are built the opposite way. Neither rate changing at all, just the proportion of the budget sitting in each bucket, is enough to move the blended number substantially.

The arithmetic, worked through

Say an account runs $60,000 a month split evenly: $30,000 in prospecting at 46.6% new-to-brand, $30,000 in retargeting at 29.2% new-to-brand — both figures close to what we've measured on our own book, where prospecting ran 46.6% NTB against 29.2% for retargeting on roughly equal spend. Blended NTB on that 50/50 split is (30,000 × 0.466 + 30,000 × 0.292) / 60,000 = 37.9%.

Now hold both campaign-level rates exactly constant and just shift the split to 20% prospecting / 80% retargeting — a completely normal move once retargeting starts returning a stronger ROAS and someone reallocates budget toward the winner. Blended NTB becomes (12,000 × 0.466 + 48,000 × 0.292) / 60,000 = 33.0%. Nothing about targeting quality changed. The account-level number dropped nearly five points purely from the mix shift, and a report that only shows the blended figure will read this as new-to-brand "getting worse."

Why this shift happens on its own, without anyone deciding it

It rarely takes a deliberate reallocation. Retargeting audiences tend to post a stronger return multiple because they're converting demand that already exists — in our own data, retargeting line items returned 7.27x against 3.28x for prospecting on comparable spend. Any automated bidding system, or any human optimising toward ROAS without a secondary new-to-brand target, will naturally push more budget toward the campaign type with the better headline return. That's retargeting, almost every time, which means the mix shift described above is the default direction an account drifts unless someone actively holds prospecting spend in place.

Read as a cost-per-acquisition question rather than a ROAS question, the picture flips. On that same book, prospecting cost $12.13 per new customer against $4.47 for retargeting — clearly more expensive per head, but a $12 new customer is a perfectly reasonable price in a lot of categories, particularly subscription or consumable ones with room to earn that back over a repeat-purchase lifetime. Whether prospecting is worth defending depends on that repeat math, not on which line has the prettier ROAS this month.

How to tell a real problem from the arithmetic

Before assuming targeting broke, pull the campaign-level new-to-brand rate, not just the account blend, for the same period. If prospecting and retargeting have each held roughly steady at their own historical rate and only the spend split moved, the blended number is doing exactly what the arithmetic above predicts, and there's nothing to fix — the account is simply spending differently, which may or may not be the right call on its own merits, but it isn't a targeting failure. If the prospecting rate itself has fallen, that's the actual signal: check audience freshness, frequency capping, and whether an exclusion list has quietly grown to cover people it shouldn't.

Build the check as a simple two-column table before drawing a conclusion: campaign-level NTB rate this period against last period, side by side, for prospecting and retargeting separately. If both columns are flat and only the total spend split shifted, you're looking at mix. If either column moved on its own, that's the row to investigate — and it's worth investigating that row specifically rather than the account as a whole, because the fix for a stale prospecting audience is nothing like the fix for a mix shift.

The common mistake, including ours

The mistake is optimising an account toward a single blended ROAS or NTB number without separate targets for prospecting and retargeting written down anywhere. We've watched a client's own team do exactly this — respond to a ROAS improvement by manually shifting budget toward retargeting, then ask two months later why acquisition had stalled, without connecting the two. The honest answer wasn't that the campaigns had gotten worse at recruiting; it was that recruiting had simply been defunded a little at a time, in a way no single week's decision looked wrong.

We've made a milder version of the same mistake reporting a client's headline NTB trend without flagging that the underlying spend mix had shifted in the same window — a technically accurate chart that told the wrong story on its own.

Setting it up so it doesn't happen quietly again

Set a floor on prospecting spend as a percentage of the account, independent of its trailing ROAS, and review it on a fixed schedule rather than letting an automated bid strategy or a well-meaning optimisation pass erode it gradually. Report new-to-brand at the campaign-type level alongside ROAS every time, not blended. And if retargeting is genuinely outperforming to the point where reallocating makes sense, do it as a deliberate, sized decision with a stated new blended-NTB expectation attached — not as the silent byproduct of chasing last week's best return.

Write the floor down somewhere more durable than a Slack message — a shared budget doc, a quarterly plan, whatever your team actually reopens. The reallocation that erodes prospecting almost never happens as one visible decision; it happens as six small ones over six weeks, each individually defensible, and the only real defence against that pattern is a rule that exists before the first small decision gets made, not a retrospective explanation after the ninth one.