- Google has handed bidding, budgets, and even campaign setup to AI agents. The one lever still yours is the value signal you feed it.
- Most lead-gen accounts tell Google “a lead happened” and count every form fill the same. So Smart Bidding does exactly what you asked: it finds the cheapest possible form fills, which are usually junk.
- Value-based bidding means reporting what each lead is actually worth (real revenue, or a defensible proxy) so the algorithm optimizes for money instead of volume.
- Google requires two or more different conversion values, four weeks or three conversion cycles of value data before you switch it on, and 15+ conversions in 30 days for target ROAS.
- The hard parts are the plumbing (getting offline outcomes back to Google) and the lag (leads that close weeks later). Get those right and you stop paying for leads that never become revenue.
Quick answers:
- What is value-based bidding?
- How is it different from Maximize conversions?
- What value should I assign to a lead?
- Does value-based bidding need offline conversions?
- How many conversions does value-based bidding need?
In May 2026, Google described its new ad ecosystem as a set of AI agents that run campaigns for you. “Ask Advisor” recommends what to do next and sets up campaigns; a “Business Agent for Leads” format went into open beta in the U.S. (MediaPost). Bidding was automated years ago. Now the campaign itself is getting automated.
Here is the part nobody markets to you: the more Google automates, the more your results come down to one thing it cannot decide for you. What is a conversion worth?
That is the whole job now. Smart Bidding optimizes toward whatever you tell it to value. If every lead counts as one identical conversion, the machine will faithfully hunt for the cheapest leads it can find, and cheap usually means unqualified. Value-based bidding is how you change the instruction from “get me leads” to “get me revenue.” So this guide stays practical: what to feed Google, how to wire it up, and the mistakes that quietly waste budget.
Step 1: Understand what you are actually telling Google
Google defines value-based bidding as “a subset of Google Smart Bidding that allows customers to optimize campaigns based on the value brought to their business” (Google Ads Help). The key word is value. The default lead-gen setup has none.
Most accounts fire one conversion when a form submits or a call connects, and every one of those conversions carries the same weight. To the algorithm, a tire-kicker and a buyer are identical. So when you run Maximize conversions, it does the rational thing: it buys the volume of identical conversions it can get cheapest. That is how you end up with a falling cost per lead and a sales team that says the leads are garbage.
What to measure: the share of your conversions that carry a differentiated value. Google requires you to report two or more different values before value-based bidding can work. If that number is zero today, you are not bidding on value at all.
The common mistake: treating every form fill as the same event. A lead is not an outcome. It is a request. Until you tell Google which requests turned into revenue, it has no way to tell them apart, and neither do your reports.
Step 2: Decide what a lead is worth
You cannot feed a value you have not defined. Start at the end: the actual revenue a closed deal produces. Then work backward through the stages a lead passes on the way there, because most lead-gen accounts do not have enough closed sales to train an algorithm on final revenue alone.
A workable value ladder for a typical lead-gen funnel:
| Stage | What it means | How to value it |
|---|---|---|
| Raw lead | Form or call submitted | Low or zero base value |
| Qualified | Meets your basic criteria | Sale value times the qualify-to-close rate |
| Booked | Appointment or quote set | A higher proxy reflecting closer-to-revenue |
| Closed | Money changed hands | Actual revenue (or bound policy premium) |
If you do not yet know real revenue per lead, use a proxy. The standard method is to take your average sale value and multiply it by the conversion rate between each stage. A booked appointment that closes 40% of the time is worth 40% of the sale. Reporting multiple values per stage gives the algorithm far more to learn from than a single rare “sale” event.
What to measure: a defensible dollar value attached to each stage, derived from your real close rates, not a vibe.
The common mistake: two of them.
First, inventing round numbers with no basis. Garbage values produce garbage optimization just as reliably as no values.
Second, only valuing the final sale. If you close 30 deals a month, that is too sparse a signal. Value the mid-funnel stages so the system has volume to learn on. This is the same discipline behind building a solar program around cost per sat appointment instead of cost per lead, and behind why cheap window-replacement leads quietly burn installers.
Step 3: Wire up the plumbing
This is where most value-based bidding projects die. The value lives in your CRM. The bidding lives in Google. Nothing connects them by default. You have two main bridges.
Enhanced conversions for leads. Google describes this as sending “hashed first-party, user-provided data from your website (e.g., lead forms) together with imported offline lead conversions” (Google Ads Help). The form captures the lead’s email or phone, it is hashed with SHA-256 before it ever leaves your site, and later you upload the outcome (qualified, booked, closed, and its value) keyed to that hashed identifier. It is the lighter-weight path because it does not require you to persist a click ID.
Offline conversion import with GCLID. The older, sturdier method: auto-tagging appends a Google Click ID to each ad click, JavaScript on your site captures it, you store it in a hidden form field and carry it into your CRM next to the prospect. When the deal progresses, you upload the conversion and value with that GCLID, and Google matches it back to the original click (Google Ads Help). One detail that breaks more setups than any other: the GCLID is case sensitive, so a CRM field that mangles it severs the match silently.
For call-driven accounts, the same logic runs through your call platform. If you sell on the phone, your call tracking software is the system of record that carries lead value back to Google, and in regulated verticals the consent proof from TrustedForm or Jornaya travels alongside it.
What to measure: your import match rate. If you upload 1,000 outcomes and Google matches 300, the algorithm is learning on a third of your data.
The common mistake: not capturing the click identifier at form submit, then discovering weeks later there is no way to tie any sale back to a click. Auto-tagging off, GCLID dropped, hashing misconfigured. Test the round trip on real leads before you trust a single value.
Step 4: Handle the lag
In high-ticket and regulated lead gen, the lead and the revenue are separated by weeks. An insurance lead binds a policy a month later. A home-services quote closes after two site visits. Value-based bidding has to live inside Google’s import windows, and those windows are finite.
Google supports a click-to-conversion cycle of “less than 14 days or 90 days, depending on the data source” (Google Ads Help). Offline conversion imports apply data up to 90 days old to the bidding algorithm (Optmyzr). If your sales cycle routinely runs past that, optimizing on final revenue will not work, and no amount of bid tuning fixes it.
The fix is a two-stage value. Send a proxy value you can defend from your close rates when the lead qualifies, inside the window. Then true it up to actual revenue when the deal closes. The proxy keeps the algorithm fed on fresh signal. The true-up keeps it honest.
What to measure: your real click-to-close cycle against the 90-day ceiling. If most revenue lands after 90 days, your trainable signal is a mid-funnel proxy, not the sale.
The common mistake: optimizing on a proxy you never reconcile. A proxy that is never trued up against real outcomes drifts, and the algorithm chases a number that stopped meaning anything. It is the same trap as letting your conversion labels drift from what they represent: the label you send and the outcome it stands for have to stay tied together.
Step 5: Feed it enough, then turn it on
Value-based bidding is a learning system, and a learning system starves on thin data. Google is explicit about the warm-up. Upload values “for 4 weeks or 3 conversion cycles, whichever is longer, before activating value-based bidding.” Target ROAS “should have at least 15 conversions in the last 30 days at the conversion tracking level” (Google Ads Help). Practitioners push the floor higher: aim for 30 to 50 conversions per month before you hand bidding to automation (Optmyzr).
Feed the data daily. Google recommends getting conversion data back as soon as it is available, with daily offline uploads as the optimal cadence. A weekly batch upload teaches the algorithm on stale information.
What to measure: conversions per month against that 30-to-50 floor, and your upload frequency. Below the floor, stay on a simpler strategy until volume catches up.
The common mistake: flipping value-based bidding on with too little value history, then panicking during the learning phase. Give it one to two weeks to stabilize before you read the results. Judging a learning system mid-ramp is how good setups get killed a week early.
Step 6: Measure revenue, then tune slowly
Here is the test of whether any of this worked, and it is not your cost per lead. If value-based bidding is doing its job, your cost per lead may rise while your cost per booked job or per bound policy falls. That is the trade you wanted: fewer, better leads that actually become money.
So change the number on the dashboard. Track revenue per click or return on ad spend, not CPL or raw lead count. The classic failure mode is celebrating a CPL that dropped 30% while booked revenue quietly cratered, because the algorithm did precisely what the old setup asked and bought cheap junk. And watch your spend. Value-based bidding spends to your budget, so Google’s June 2026 budget pacing change can push daily spend up fast.
When you tune, tune slowly. The guidance from practitioners is not to change targets by more than 20%, or more often than every two weeks (Optmyzr). Yank the target around and you reset the learning you just paid for. It is also worth a moment to audit your Smart Bidding strategies after the Target CPA and Target ROAS rename, since the strategy names changed while the underlying mechanics did not.
If you sell on the phone, the same revenue-not-volume discipline shapes how you price and buy media in a pay-per-call program.
Frequently Asked Questions
What is value-based bidding?
Value-based bidding is a form of Google Smart Bidding that optimizes for the value a conversion brings your business rather than the count of conversions. Google offers two strategies: Maximize conversion value, and Maximize conversion value with a target ROAS (Google Ads Help). Instead of buying as many leads as possible, it buys the leads most likely to be worth the most, based on the values you report.
How is it different from Maximize conversions?
Maximize conversions optimizes for the number of conversions, treating each one as equal. So it tends toward the cheapest leads available. Value-based bidding optimizes for total conversion value, so it favors leads that resemble your high-value past conversions even if they cost more. The difference only exists if you actually report differentiated values; without them, value-based bidding has nothing to optimize on.
What value should I assign to a lead?
If you know the real revenue a lead produced, use it. If you do not, use a proxy: take your average sale value and multiply it by the conversion rate from that stage to a close, so a booked appointment that closes 40% of the time is worth 40% of the sale value. Report values at more than one stage so the algorithm has enough signal to learn on, and reconcile proxies against real outcomes as deals close.
Does value-based bidding need offline conversions?
For most lead-gen businesses, yes, because the value of a lead is decided offline after the form fill. You bring that value back to Google with enhanced conversions for leads, which uses hashed first-party data (Google Ads Help), or with offline conversion import using the Google Click ID (Google Ads Help). If your entire sale completes online with a known value at conversion, you can skip the offline import.
How many conversions does value-based bidding need?
Google recommends uploading value data for four weeks or three conversion cycles, whichever is longer, before activating value-based bidding, and target ROAS needs at least 15 conversions in the last 30 days (Google Ads Help). Practitioners suggest a sturdier floor of 30 to 50 conversions per month before handing bidding to automation (Optmyzr). Below that, stay on a simpler strategy until volume builds.
