Value-based bidding raised our value per lead. It also cost us volume.

Ask Google which bidding strategy to use for lead generation and the answer will usually be Maximise Conversion Value. It is the strategy the platform is actively promoting, and for a mature account with clean value data flowing back into it, the recommendation is sound.

What the recommendation does not carry is the cost of switching before the account is ready for it — or the fact that the platform has an interest in the answer.

What actually happened when we switched

In accounts our team has run, moving to value-based bidding did exactly what it promises on the headline metric. Value per lead went up.

Three other things went down or up in the wrong direction at the same time:

• Lead volume fell. The algorithm became selective, which is the point, but the absolute number of enquiries dropped enough to be felt by the sales team.
• Impressions fell. Fewer auctions entered means less presence in the market, which has second-order effects that do not show up in the campaign report.
• CPA rose — and this was sharpest among smaller clients, where there was simply less value data for the model to work with.

The mistake at this point is to read that as a failed experiment and revert. Value-based bidding was doing its job. The problem was that it was being asked to do the whole job.

The fix: split the funnel deliberately

Rather than choosing one strategy for the account, we ran two and gave each the work it is suited to.

Alongside the value-based Search campaigns we added a Maximise Conversions campaign with a low target CPA against core keywords, to recover the volume and impression presence that value-based bidding had conceded. We supported it with Performance Max.

Within two months that had settled into a deliberate structure:

The point is not that this exact split is correct for every account. It is that “which bidding strategy should we use” is usually the wrong question. The right one is which strategy belongs on which part of the funnel.

When Maximise Conversions is still the better choice

Google’s product direction increasingly favours value-based optimisation. That is a signal about where the platform is heading, not about which strategy suits your account this quarter — and one strategy does not fit every account.

Maximise Conversions remains the better call in four situations we see regularly:

• New markets. An advertiser entering Qatar has no value feedback loop yet. Maximise Conversions remains a sensible choice in new markets and immature accounts when there is not yet enough value data to guide a value-based strategy.
• Immature accounts. Not enough first-party data to guide a value model means the model will optimise confidently toward a number you essentially invented.
• High quality score keywords. Where CPCs are already low, volume is cheap and selectivity buys less than it costs.
• Brand protection. Depending on account structure and what the incrementality data says — see below, because our finding here surprised us.

How to test this without fooling yourself

There is a specific trap in comparing these two strategies, and almost everyone falls into it once.

If you run a Maximise Conversions campaign and a value-based campaign side by side in the same account, Google will label the Maximise Conversions campaign as limited by bidding strategy. It is very easy to read that label as a result. It is not a result — the label reflects how Google Ads is designed and which strategy the system is built to favour.

For the comparison we wanted to make, we used separate accounts with comparable spend and similar algorithm maturity. Anything less and you risk measuring how the system is designed rather than your own performance.

Two further things make this kind of testing work:

Run it on a schedule, not as a reaction. Maintain an A/B test calendar. Optimisation actions get sequenced in advance and evaluated over a horizon long enough to separate signal from noise. Changes made in response to a bad week are not tests.

Measure incrementality, not last click. Incremental ROAS and last-click ROAS answer different questions. Only the first tells you whether the spend caused the revenue.

Feed the algorithm before you ask it to be clever

Every argument above rests on one prerequisite: value-based bidding is only as good as the value data behind it.

Privacy regulation and consent mode have steadily reduced what the platforms can observe unaided, which has made the advertiser’s own data considerably more important in guiding bidding strategy. In practice that means gathering from every source available and organising it by funnel stage before uploading — not as one undifferentiated list:

• CRM records: previous customers, high-value customers, repeat customers
• Engagement from social campaigns
• LinkedIn followers, Facebook followers and page engagement
• Micro-conversion events from the site

Long sales cycles need micro-conversions. Where the conversion window runs to ninety days, optimising only on completed sales starves the model — it learns from a trickle of events months after the clicks that caused them. Record meaningful intermediate signals instead, such as a destination search or a price check. That gives the algorithm something to learn from now, including from the people who did not convert, which is as useful as the positive signal.

Seed new accounts from mature ones. Where a client operates across regions, audience and performance data from strong markets can be fed into new or underperforming accounts to compress the learning period substantially. It is one of the fastest ways to make a new market productive and it is routinely left unused.

Audience uploads should run monthly. A first-party audience decays.

The short version

• Value-based bidding raises value per lead and can cost you volume, impressions and CPA efficiency — most visibly on smaller accounts.
• Do not pick one strategy for the account. Split the funnel: value-based bidding on high-value keywords, Maximise Conversions and Performance Max for volume and discovery.
• Maximise Conversions is still the better opening strategy in new markets and immature accounts, because it does not need a real-time value feedback loop.
• Never A/B test the two inside one account. Google will flag the Maximise Conversions campaign as limited, and it is easy to mistake that label for data.
• None of it works without first-party data, organised by funnel stage and uploaded monthly.