“Leads” Is a Label. Your Conversions Tell Google Where to Spend.

Pick Leads in Google Ads, then mark every form submission as a primary conversion. If your quote request and newsletter signup fire the same event, you have told the bidder they are the same outcome. The campaign objective you clicked did not make that decision. You did.

That distinction gets lost in a real Google Ads Search Certification question: a business wants prospects through form submissions, so which objective should it choose? The certified answer is Lead generation. Correct. It takes ten seconds. It also tells you remarkably little about where the budget will go.

I want to build the answer one layer at a time, using a home-services account with two forms: a quote request and a newsletter signup. Both initially fire one undifferentiated form_submit event. First, what does the objective set? Then, which conversions reach the bidder? Finally, what tells it that one lead is worth more than another?

Layer 1: The objective sets up the menu

Choose Leads, and Google proposes conversion goals that look like leads: form submissions, calls and signups. It also surfaces campaign types and bid strategies that fit. Choose another objective, and the proposals change. Google’s objectives documentation describes this as a setup choice. The objective shapes what you are offered; it does not place a bid.

That is why the certification answer is useful but incomplete. It gets you to a sensible starting screen. It does not tell Google whether the homeowner requesting a furnace quote matters more than someone downloading a maintenance guide.

An agency running lead generation for fitness studios described testing the same campaigns and creatives under different objectives, without finding a delivery difference worth writing home about. That is one account’s observation, not a rule for every campaign. The practical point is narrower: do not mistake the objective selector for the place where you define a valuable lead.

Click Leads for the menu you want. Then inspect the goals behind it.

Layer 2: Primary actions tell the bidder what counts

A conversion goal groups conversion actions. A quote request and a newsletter signup can be separate actions; the goal determines how those actions are made available to a campaign. Within that setup, each action can be marked primary or secondary. As the Google Ads API documentation puts it, a primary action is used for bidding optimization; a secondary action is not. Secondary is still useful for reporting. It just is not an instruction to chase more of that event.

There is another choice at the goal level. An account-default goal is available across the account; a campaign can instead use campaign-specific goals. For an action to feed bidding through an account-default goal, the goal must be selected and the action must be primary. Google’s account-default goals documentation explains those settings. The second checkbox does not rescue a badly chosen first one.

Back to the home-services site. If both forms fire only form_submit, Google Ads cannot look inside that event and decide which visitor requested a quote. Changing the status of the shared action would change the status of both forms. Separate the tracking first. Give the quote request and newsletter signup distinct conversion actions, then make the quote request primary for the campaign selling heating and cooling work. Keep the signup secondary if you still want to see it in reports.

Campaign-specific goals give you another level of control. Google’s example uses a shoe seller who wants one campaign to optimize for purchases rather than every available action. In this account, the quote-request campaign should optimize for quote requests. A campaign promoting the maintenance guide can have a different goal if signups are genuinely its purpose. The business decides what success means for each campaign; the bidder should not have to guess.

The takeaway is mechanical: one shared event cannot carry two different instructions. Separate it before you tune anything else.

Layer 3: Every primary conversion teaches a pattern

Smart Bidding works from the conversions you send it. It sees the signals around people who converted and bids for more opportunities that resemble them. It does not know that one form-filler has a broken furnace and another has no intention of buying. If both fire the same primary event, both become examples of success.

That is the problem with a cheap, easy event. A newsletter signup can arrive without the friction of requesting a quote, answering a call or booking a job. Give the bidder a pile of those signups and fewer genuine enquiries, and the easier pattern can dominate. What looks like improving cost per conversion may be a growing supply of people who were never likely to buy. The bidder is following the label, not judging the lead.

Volume matters, which makes this tempting. Google’s high-quality-leads playbook points to roughly 15 conversions in the last 30 days at account level as a useful signal threshold. When quote requests are scarce, counting every signup can make the numbers look healthier. It can also teach the wrong lesson faster.

One advertiser asking whether to remove form submissions as a conversion action altogether got a useful diagnosis in this PPC discussion: Maximize Conversions with low-quality submits marked primary can create a feedback loop because those conversions are easier to acquire. Removing every form event is not necessarily the fix. Deciding which event deserves to train the bidder is.

So before blaming an algorithm for finding junk leads, check what the account calls a win.

Layer 4: Values tell it which wins matter more

A primary action answers what counts? A value answers how much does it count? Google’s value-based bidding documentation allows real economic values, such as revenue, or proxy values, such as lead scores. For example, you might assign a raw form fill $50, a sales-qualified lead $500 and a booked demo $1,500. Those are illustrative weights, not a claim that every business earns those amounts. Their job is to preserve the relative ranking.

For the home-services account, separating the two forms stops newsletter signups masquerading as quote requests. Value can take the next step: distinguish a request that never goes anywhere from one that becomes a booked job. That requires information from beyond the form.

Google offers Qualified leads and Converted leads goal types for passing later outcomes back. The workable pattern is to track stages such as form submit, qualified lead, booked meeting and closed-won as separate conversion actions, then decide which stages should be primary. Do not collapse the whole journey into one vague “offline lead” event. Do not switch the primary action to closed-won on day one if there are too few sales to give bidding a usable signal. There is no prize for switching early.

Three practical details determine whether the feedback reaches Google Ads:

1. Keep the click identifier. The form needs to store the GCLID with the submission so a later outcome can be matched to the ad click.
2. Watch the attribution window. Imported conversions have a 90-day window from click to conversion. A longer sales cycle creates a measurement problem before it creates a bidding problem.
3. Send updates regularly. Daily uploads keep the bidder closer to the outcomes your sales team is seeing; delayed backfills leave it learning from older information.

That is plumbing, not a new objective. It matters because the bidder cannot value a booked job it never hears about. A travel brand that returned CRM values daily, with distinct values for funnel stages, reported a 28% lower cost per booking at unchanged ad spend. Zoe Financial also fed offline events back across its lead-to-sales journey; its most valuable client segment rose to 60% of sales. Neither example means that copying a value table will produce the same result. The mechanism is the useful part: later outcomes replace a crude form-fill label with a better instruction.

If nearly every lead is worth the same, or the sales team does not update the CRM, one well-chosen primary action and a static value may be enough. If your two forms differ sharply in worth, separate them first. Weighting and offline imports can follow when the sales data supports them.

AI Max can widen the reach, not fix the signal

AI Max is neither a campaign type nor the objective you chose at setup. Google describes it as an optimization layer inside existing Search campaigns. Its search term matching, text customization and final URL expansion can put ads in front of searches or on landing pages beyond the ones you set up directly. The conversion goals and bidding strategy underneath still matter.

Broader reach plus a weak conversion signal creates more room to spend on the wrong people. If the home-services account rewards every newsletter signup as though it were a quote request, expanding the searches it can pursue does not teach it the difference. It gives the existing instruction more places to run.

AI Max also needs enough conversion information to work from. A comparison of AI Max readiness puts the conversion-bidding threshold around 15 to 30 conversions in a 30-day period. Google says AI Max will not be effective in budget-limited campaigns, where expansion has little room to operate. On the upside, Google reports that advertisers activating it typically see 14% more conversions or conversion value at similar CPA or ROAS, based on its 2025 internal data for non-retail advertisers. That is a reason to test it after fixing the signal, not a forecast for this account.

A toggle can extend a sound instruction. It cannot make an unsound one sound.

Put the two-form account through the model

Say the heating and cooling business spends $20k a month across two metros. Its quote request and newsletter signup both fire form_submit. The account counts that shared action as primary, with no value attached. Cost per conversion looks healthy. Booked jobs stay flat. The bidder has no separate quote-request signal to favor, so a rise in easy signups can make the campaign look better while the sales team sees more tire-kickers.

The repair has an order. Skipping the first step makes the second impossible:

1. Split the tracking. Create distinct conversion actions for quote requests and newsletter signups. Confirm each form fires the right one.
2. Set the bidding role. Make quote requests primary for the service campaign and newsletter signups secondary. The latter still appear in reporting without training that campaign’s bidder.
3. Add value when you can support it. Start with a consistent static weight if that is all you have. Feed qualified or closed-won outcomes back when the CRM can report them reliably.

The objective remains Leads throughout. What changes is the bidder’s job description: from find anyone who submits either form to find people who request quotes. That is the decision the setup screen could not make for you.

Test the model on an account you have never seen

The same three questions work outside home services:

1. What did the objective make easy to set up? Check which goals Google proposed. If the action the business actually cares about is missing, set it up rather than assuming the objective has handled it.
2. Which actions are primary? Check whether they represent meaningful progress toward revenue. Keep useful but weaker observations secondary when they should not steer that campaign.
3. What value does each primary action carry? If two actions are worth different amounts to the business, do not leave the bidder treating them as interchangeable.

Now apply that to a new case: B2B SaaS, a long sales cycle, a demo-request form and a gated ebook download. Leads is still the right objective. If both forms share a primary event, the setup repeats the home-services mistake. Separate them. If demos are too scarce to provide enough signal on their own, simply counting ebook downloads as equivalent wins is not the answer. Track stages up the funnel, give early signals smaller values and qualified outcomes larger ones, and pass later results back when the CRM can do it consistently.

A B2B SaaS team took the closed-revenue version of that approach with us, shifting bidding toward deals that actually closed and scaling qualified pipeline by 80%. The point is not to borrow its number. It is to follow the sequence: define the event, establish its worth, then let bidding pursue it.

Set the objective in ten seconds. Spend the hour on the goals.

The certification answer is Lead generation. Choose it. Then spend your attention where the budget actually takes direction: which actions are primary, whether distinct outcomes have distinct values, and whether closed deals ever make it back to the bidder.

Keep checking. Forms get redesigned and stop passing click IDs. CRM updates slip. Values stop reflecting what the business sells. That continuous work is what runs at groas: specialized models working on bids, keywords and budgets within your guardrails, a named senior strategist accountable for direction, logged changes with their reasoning, and a flat monthly fee that does not rise with ad spend. If you want to see what your account is currently telling the bidder, start with groas and find out what is wasting your budget.