When we audited our own Google Ads account, the problem was obvious in retrospect — but it had been sitting there unnoticed for months. Every conversion was valued at £1. The algorithm was doing exactly what we told it to do: win as many conversions as possible, at the lowest possible cost, regardless of what those conversions were actually worth.

That’s the default trap. And it’s remarkably easy to fall into, because Google makes flat conversion values the path of least resistance. The result is a Smart Bidding algorithm that looks productive on a dashboard — conversion volume up, cost-per-conversion holding steady — while quietly optimising for the cheapest, lowest-value outcomes it can find.

We fixed it by making Google bid on revenue instead of clicks. This is how.

The Default Trap: What Flat Values Actually Tell the Algorithm

Most Google Ads accounts assign a flat value — often £1, or the same number for every conversion type — when they set up conversion tracking. The intent is harmless: it’s a placeholder so conversions show up in reporting. The consequence is not.

When every conversion carries the same value signal, Maximize Conversion Value and Target ROAS (tROAS) behave identically to strategies optimising for volume. The algorithm has no way to know that a booked sales call is worth ten times a newsletter sign-up, that a returning customer is worth twice a first-time buyer, or that a lead from a specific campaign converts to revenue at a higher rate. It simply maximises the number it can get.

This matters because Smart Bidding operates on auction-level signals — device, time of day, audience membership, search query semantics, location, and dozens of other real-time inputs. When you give it accurate value data, it can use those signals to bid higher for the profiles that produce high-value outcomes. When you give it a flat signal, it optimises for volume and leaves the value dimension entirely untouched.

What Value-Based Bidding Actually Is

Two bidding strategies in Google Ads are genuinely value-based:

Maximize Conversion Value — Google spends your budget trying to generate the highest total conversion value possible. No return-on-ad-spend target, just maximum value within budget. Best used when you’re in a learning or growth phase and want volume at the high-value end, without constraining the algorithm with a ROAS floor it can’t yet reliably hit.

Target ROAS (tROAS) — Google targets a specified return on ad spend (e.g. 400% = £4 in conversion value for every £1 spent). This is the mature version: you’re telling the algorithm not just to chase value, but to chase it at a specific efficiency. The algorithm will under-bid on auctions where it predicts the conversion value won’t hit your target, and over-bid on the ones where it predicts it will.

Both strategies are useless without real conversion values feeding them. The strategy toggle is the last step, not the first.

Assigning Real Conversion Values

This is where the actual work is, and where most accounts still cut corners.

For ecommerce, the answer is straightforward: pass the actual transaction revenue into the conversion event. Google’s global site tag and the GA4 integration both support dynamic revenue values. Every purchase sends its real pound value. The algorithm has exactly what it needs.

For lead generation — the harder case — you have a few options:

Lead tier values. Assign static values based on what each lead type is worth to your pipeline. A demo request might be worth £200, a content download worth £10, a free trial sign-up worth £80. These should be rough proxies for downstream revenue contribution, not arbitrary numbers. Pull your close rates and average deal values from your CRM and work backwards.

Offline conversion imports. If you have a CRM and a reasonable sales cycle, this is the highest-signal option. You record which Google clicks converted into qualified leads, which leads converted to opportunities, and which opportunities closed — and you import each stage back into Google Ads with the actual (or estimated) revenue value attached. Google’s Enhanced Conversions for Leads and the standard offline import API both support this. The latency is a consideration — imported conversions typically appear with a 24–48 hour lag — but the signal quality is incomparably better than any static proxy.

Value rules. For accounts that can’t get CRM imports working, Google Ads value rules let you apply multipliers by audience, device, or location. A returning customer audience might carry a 1.5× value multiplier because they close at a higher rate. A mobile visitor on a B2B campaign might carry a 0.7× multiplier because they’re less likely to be a decision-maker. These are coarse instruments compared to direct value imports, but they’re far better than a flat signal.

Why Clean Conversion Tracking Has to Come First

Value-based bidding amplifies whatever signal you feed it. If your conversion tracking is broken — double-counting, missing view-through attributions that inflate value, firing on the wrong events — you are not just wasting budget. You are actively training the algorithm on bad data, and it will optimise confidently toward outcomes that don’t exist.

Before touching bidding strategy, verify that your conversion tracking is measuring what you think it’s measuring. Check that primary conversion actions are firing once per genuine conversion. Check that cross-device attribution is consistent. Check that your value numbers reflect real commercial logic, not arbitrary placeholders.

If you’re not sure whether your tracking is clean, the /insights/fix-google-ads-conversion-tracking post covers the diagnostic we run before touching bidding on any account.

The Data Threshold: When Smart Bidding Actually Works

Google’s published minimum for tROAS is 15 conversions per campaign in the past 30 days. In practice, the strategy tends to stabilise meaningfully around 30–50 conversions per month. Below that, the algorithm doesn’t have enough signal to distinguish reliable patterns from noise.

The right bootstrapping sequence is:

  1. Manual CPC or Maximize Clicks — build conversion history. Track everything. Fix your tracking. Let the account accumulate data.
  2. Maximize Conversion Value (uncapped) — once you have 15+ conversions per month with real values assigned, switch to Maximize Conversion Value without a ROAS target. Watch performance for 3–4 weeks. Don’t make major changes during the learning period.
  3. Target ROAS — once Maximize Conversion Value has stabilised and you have a sense of what ROAS the algorithm is naturally achieving, set a target slightly below that natural rate and tighten over time as confidence builds.

Starting with an aggressive tROAS target on a thin account is one of the most reliable ways to trigger the learning period repeatedly and see spend collapse as the algorithm struggles to find auctions that meet an efficiency target it can’t model yet.

Common Mistakes

Setting the ROAS target too high, too early. The algorithm restricts spend to auctions it’s confident will hit the target. On a thin account with a high target, it often can’t find enough volume and underdelivers badly.

Mixing conversion actions with wildly different values in the same campaign. If a purchase (£500) and a newsletter sign-up (£5) are both primary conversion actions in the same campaign, the algorithm’s value signal is muddled. Segment by intent or use conversion sets to isolate the primary commercial action.

Ignoring the learning period. Every strategy change restarts the algorithm’s learning cycle. Changing targets, pausing and resuming campaigns, or making large bid adjustments mid-learning will prolong instability. Budget for a 2–4 week stabilisation window and leave it alone.

Not re-evaluating values periodically. Lead values assigned twelve months ago may not reflect current close rates or deal sizes. Stale values silently degrade the algorithm’s accuracy over time. Review your conversion values each quarter against actual CRM data.

What Changes When You Get This Right

When we moved our own account from flat values to a tiered lead value model with offline conversion imports, the algorithm’s behaviour changed within three weeks. It began deprioritising the campaigns and audiences that were generating cheap, low-intent form fills and allocating more budget toward the signals that predicted high-quality pipeline. Cost-per-lead increased. Revenue from those leads increased faster.

That is exactly the outcome value-based bidding is designed to produce — and it’s not available to any account still running on a flat £1 signal.

If you want to know whether your account is set up to benefit from value-based bidding, our AI Paid-Ads Audit covers conversion tracking integrity, bidding strategy configuration, and value signal quality as part of a structured review. Start with the Account Health Scorecard for a faster read on where your account stands.