Skip to content

Manual CPC vs smart bidding for shopping

By cartscaling Reviewed

Use manual bids while you cannot trust your conversion tracking or while orders are too rare for a machine to learn from, and switch to automated bidding as soon as both of those are fixed. Nothing else about the account should decide this, because automated bidding is only as good as the conversion data feeding it.

The practical test has two parts. First, does every order record once, with the correct value, within a few seconds of checkout. Second, does the campaign produce conversions in most weeks rather than a scattering across a quarter. Fail the first and automated bidding will optimise toward a broken signal. Fail the second and it has too little history to price an auction, so it falls back on averages and your cost per click wanders.

Choose

Manual CPC

when

  • Conversion tracking is new, or you have found duplicate or missing purchase events
  • The campaign sees only a handful of orders a month and nothing stable to learn from
  • Margin is thin enough that a maximum click price is a hard commercial limit
  • You are opening a new country and want to buy clicks cheaply while you gather data
  • You need to hold a low bid on a product you stock but do not want to push

Choose

Smart bidding

when

  • Purchase tracking is verified against store orders for at least a full month
  • The campaign produces conversions in most weeks, not in occasional bursts
  • Your catalogue is large enough that per product bid maintenance is unrealistic
  • Prices, stock and promotions change often and bids need to follow without you
  • You have found that the same query converts very differently by device or time of day
CriterionManual CPCSmart bidding
CostYou cap the click price, so spend is predictable but opportunities are missedClick prices vary widely, including high bids on auctions likely to convert
Setup effortOngoing: bids need reviewing as products and prices changeOne time, once conversion tracking and values are verified
Time to signalImmediate, since you see the effect of a bid change the same dayTwo to three weeks before performance settles after the change
ControlFull control of the maximum you pay for a clickYou set the objective, the platform sets each bid
Best forSmall accounts, new tracking, single product lines with strict limitsCatalogues with steady order flow and reliable purchase data
What can go wrongBids stay flat while the auction moves, and impressions quietly disappearA tracking break teaches it the wrong lesson and it takes weeks to unlearn
How we measure itImpression share and average position cost against a fixed ceilingCost per order and conversion value over a full purchase window

When both apply

Treat manual bidding as a stage rather than a rival. Run it while the account is being fixed, use the period to build a negative list, verify order values and find which products convert, then hand the same structure to automated bidding with that history attached.

If you need both at once, split by risk. Keep automated bidding on the products with order history, and hold manual bids on a small test group of new or restricted lines until they produce enough conversions to join the main campaign.

Questions

Compare recorded conversions with store orders for the same period, using the order date rather than the click date. If the counts are within a small margin and the revenue values match, the signal is good enough. A gap of more than a few orders usually means duplicate tags or a missing purchase event.

Often yes, and that is not automatically a problem. The bidding pays more for auctions it expects to convert and less for the rest, so the average click gets dearer while the cost per order can fall. Judge the change on orders, not on click price.

Not directly in most shopping setups. If a hard ceiling is a commercial requirement, keep those products in a manual campaign, or set the objective so the campaign cannot spend more per order than your margin allows and monitor it weekly.

Each switch restarts the learning period, so an account that changes strategy every fortnight never settles. Decide the change, give it three weeks with the budget untouched, and only then read the result.

Related

Your plan is free.

A person reads it. You keep the plan. No logins, no call first.

Store, name, email and phone. A person replies from a real inbox. Privacy

Get your free growth plan