Retargeting steals credit from prospecting (and how to prove it) (your best ROAS is the most suspicious)

Chinmay R. · has paused a 9x campaign and watched nothing happen🎯 with a pause test and a pool split, not a lecture

Published September 16, 2026

The 60-second version

  • Retargeting harvests demand prospecting created, then takes 100% of the credit under last-click — the gap is selection, not performance.
  • Expect 10–30% incrementality for retargeting versus 50–80% for prospecting — measured by a 2-week pause test on warehouse totals.
  • The fix: report the pools separately weekly and judge budgets on incremental CAC, never on claimed ROAS.
bottom line: scale creation, cap the harvest

Your retargeting campaign reports 9.2x ROAS. Prospecting reports 2.1x. The team wants to move budget to the winner.

The disagreement is not about creative. It is that retargeting harvests demand prospecting created, then takes the credit for it. Last-click gives the final touch 100% of the sale, and retargeting is — by design — always the final touch.


Why retargeting always looks like a hero

Retargeting only talks to people who already raised their hand: site visitors, cart abandoners, past buyers. Prospecting talks to strangers. One of these audiences was already going to buy.

Think of it like a shop assistant who only approaches customers already standing at the till. Their close rate looks incredible. They did not create the sale.

Three mechanisms inflate the number, all pushing the same way:

Intent bias

Retargeting pools are 5–10x more likely to buy with zero ads. The ad gets credit for intent that existed before the impression.

Journey truncation

A customer sees three prospecting ads, then one retargeting ad, then buys. Last-click gives retargeting 100% and prospecting zero.

Loose windows

A 7-day click plus 1-day view window sweeps in buyers who brushed an ad days ago. Retargeting pools brush constantly, so they sweep most.

This is why every account shows the same pattern: retargeting ROAS of 8–12x sitting above prospecting at 1.5–2.5x. The gap is not performance. It is selection.

Prospecting subsidises harvesting. Cut prospecting to fund retargeting and total orders fall a month later — retargeting has fewer hands left to harvest. The dashboard never warns you; it celebrates the retargeting ROAS right up to the cliff.


The toll booth, not the factory

A toll booth does not build the road. It sits on it and collects. Retargeting is the toll booth; prospecting, search and organic are the road builders.

What the dashboard claims

🧾 Retargeting CAC: ₹476

Spend ÷ platform-claimed orders. ₹2,00,000 over 420 claimed orders — including buyers who would have returned anyway.

  • 420 claimed orders
  • Justifies shifting budget — wrongly
What the pause test measures

🔬 True incremental CAC: ₹2,857

Spend ÷ orders that would not exist without the ad. Only 70 of 420 were truly incremental — six times the reported cost.

  • 70 truly incremental orders
  • 17% incrementality share

The dashboard prices toll booths as factories. The business pays factory prices for toll-booth output.

Put it in bars, because the gap is the whole story:

Same spend, same month — five of six claimed orders would have happened anyway.

The 9.2x campaign that was mostly harvest

Worked read
Retargeting spend₹2,00,000Meta + Google remarketing, one month
Platform-claimed orders4209.2x ROAS at ₹4,400 AOV
Pause-test incremental orders~70Orders that disappeared when retargeting went dark
Reported CAC₹476₹2,00,000 ÷ 420
True incremental CAC₹2,857₹2,00,000 ÷ 70 — 6x the reported figure
Blended CAC that month₹1,120The only number finance should see
At ₹476 the campaign scales forever. At ₹2,857 it fails payback against a ₹900 contribution per order. Same campaign, opposite decision.

The pause test that proves it

You do not need a lab. You need two weeks where retargeting goes quiet and prospecting keeps running, and a warehouse count of total orders.

The retargeting pause test

Process Flow
1

Tag every campaign

Label each campaign prospecting or retargeting in a sheet — by audience, not by name. Lookalikes and broad count as prospecting; site visitors, carts and customer lists count as retargeting.

2

Pause retargeting only, 2 weeks

Keep prospecting, search and everything else untouched. No sale events, no budget jumps mid-test — stillness is the instrument.

3

Count warehouse totals, not platform conversions

Total paid orders per day from your orders table. Platform numbers go blind in the dark cell by construction.

4

Read the gap

Orders during pause vs the prior 4-week baseline, adjusted for weekday mix. The shortfall is retargeting's true incremental contribution.

A footwear brand ran exactly this: retargeting dark for 14 days, prospecting untouched. Daily orders fell from 118 to 113 — five orders a day, roughly 70 over the fortnight. The dashboard had claimed 30 a day. Five were real; twenty-five were harvest.

Pause-test rules stillness is the instrument →

  • Pause retargeting, never prospecting — the question is what the harvester adds on top of the creator.
  • Measure total orders, never platform conversions — the platform cannot see the orders it did not touch.
  • Cover one full purchase cycle — 2 weeks suffices at 3-day median lag, 3–4 at 10-day lag.

Do not test during a sale. Festive peaks, price drops and influencer spikes move both groups 3–5x and void the read. A voided fortnight is cheaper than a believed-but-wrong answer steering budgets for a year.


Splitting prospecting from harvesting in SQL

The permanent fix is structural: report the two pools separately, every week, from your own tables. One query ends the argument.

Prospecting vs retargeting CAC from your warehouse

Tag campaigns once, split every week

Same-day CAC still flatters retargeting — buyers it harvests also buy fast. So join the query to your pause-test share: multiply retargeting orders by ~0.17, prospecting by ~0.65, and judge budgets on those. The ratio between the pools matters more than either absolute: when retargeting claims more than 35–40% of total attributed orders, harvesting is eating the mix.

LIVE · DRAG ITWhat is your retargeting really costing?
₹2L
420
70

From your pause test, not the dashboard

Reported CAC
₹476
True incremental CAC
₹2,857
Incrementality share
16.7%
TOLL BOOTH

Under 30% incrementality the pool mostly taxes demand others created. Cap frequency, cut budgets, shift to prospecting.

live mathsestimates are fine — this is a what-if sandbox

Say what to ignore: ignore retargeting ROAS as a scaling signal. Watch it only as a health check — a sudden fall means frequency caps, creative fatigue or feed breaks, not a demand collapse. drag me — your numbers change the verdict!


Quick gut-check

One question. Get it and the whole post clicks. 30 seconds, no maths!

Retargeting claims 420 orders at ₹476 CAC. Pausing it cuts total orders by 70. What is its true incremental CAC on ₹2,00,000 spend?


Frequently asked questions

Should I turn retargeting off entirely?

Rarely. A lean reminder layer — tight frequency caps, short windows, cart-only audiences — usually pays back even at low incrementality. What should go is the bloated version: 30-day pools, uncapped frequency, and budgets scaled on claimed ROAS. Amputate the harvest, keep the reminder.

Why does retargeting ROAS fall when I scale prospecting?

Because attribution is a fixed pie sliced by last touch. More prospecting creates more journeys that end in a retargeting touch, so retargeting claims a larger slice without improving. Judge each pool on its own incrementality share, never on slices of one pie.

What frequency cap should retargeting use?

Start at 3–4 impressions per user per week for carts, 2–3 for site visitors, and watch indexed CTR. Retargeting past frequency 6+ mostly buys ignoring — the same creative to the same person stops registering and starts annoying. Cap first, then judge.

Does this apply to brand search too?

Yes, identically. Brand search sits at the till like retargeting does — navigational clicks from customers who already decided. It typically tests at similar 10–30% incrementality. Report brand and retargeting together as capture, prospecting and generic search as creation.

How often should I re-run the pause test?

After every 2–3x scale-up, every major audience change, and twice a year otherwise. Incrementality decays as spend grows — the first ₹1,00,000 harvests the ripest carts and each additional lakh buys thinner intent.


The summary

  • Retargeting targets the most likely buyers, so last-click makes it look unbeatable. The gap is selection, not performance.
  • A toll booth does not build the road. Prospecting creates demand; retargeting collects a share of it.
  • The pause test — retargeting dark for 2 weeks, warehouse totals as the metric — measures the true share. Expect 10–30% for retargeting, 50–80% for prospecting.
  • Report the pools separately every week, from tagged spend and new-customer orders. Never let one blended ROAS hide the mix.
  • Judge budgets on incremental CAC (spend over incremental orders), not reported CAC. At 17% incrementality a ₹476 CAC is really ₹2,857.
  • Ignore retargeting ROAS as a scaling signal. Keep a lean reminder layer; shift scale budgets to creation.

Takeaways for your next report

  • Retargeting harvests demand prospecting created — last-click gives the final touch 100% of the credit.
  • Expect 10–30% incrementality for retargeting vs 50–80% for prospecting; measure it with a 2-week pause test.
  • True CAC is spend over incremental orders — a ₹476 reported CAC can be ₹2,857 in reality.
  • Report pools separately weekly; cap retargeting frequency and scale prospecting, not the harvest.
stick this on your Monday report
Free tool

Blended CAC Calculator

Total spend across every channel, divided by total new customers — the acquisition cost number that reconciles with what you actually spent.

Chinmay Raibagkar

About author →

Founder of DataLens AI. He helps non-technical teams read their ad and database numbers with confidence — which number to trust, what to do next, and what to ignore.