Guide · Signal selection
Which lead stage should your ads optimise for?
Form fills teach ad platforms to find form-fillers. Closed deals arrive too late and too rarely to teach them anything. The right stage sits in between, and it is often a different stage on Google, Meta and LinkedIn for the same business.
Should B2B ads optimise for leads, qualified leads or revenue?
Short answer
Optimise for the deepest stage that still arrives often and fast enough for the platform to learn. For most B2B accounts that’s a qualified lead or a booked meeting, not a raw form fill and not a closed deal. Form fills train platforms to find form-fillers; closed deals are usually too few and too slow to steer bidding.
“Optimise for revenue” is the advice most agencies give, and in an online shop it is right: the purchase happens minutes after the click, hundreds of times a week. A B2B company has a different shape. A manufacturer might get 60 enquiries a month and two orders. A SaaS company might close eight deals a month, four months after the first demo request. Neither pattern gives an ad platform what it needs to learn.
A deeper outcome is not automatically a better signal. The choice is a trade-off between how truthful a stage is about value and how usable it is for learning. This guide gives you a test for that trade-off, and shows why the answer usually differs by platform.
How do ad platforms learn from the stage you choose?
Short answer
Bidding systems look at who completed the conversion you nominated, predict who else is likely to, and bid more for them. They need enough of those conversions, reported soon enough after the ad interaction, to find a pattern. A shallow stage gives plenty of fast data about the wrong thing; a deep stage gives the right thing too slowly.
Three consequences follow.
- Optimising to form fills finds form-fillers. The platform can’t see that a student, a job-seeker or a company too small to buy never became a customer. It only sees that they converted cheaply, so it finds more of them.
- Sparse stages stall learning. Google describes a learning period that can take up to around 50 conversion events or three conversion cycles (Google Ads Help). Meta shows ad sets as “learning limited” when they are unlikely to get around 50 optimisation events in the week after a significant edit (Meta Business Help). Eight deals a month never gets there.
- Late stages fall outside the window. Uploads for Google’s enhanced conversions for leads must arrive within 63 days of the click, and click-ID imports within 90 days (Google Ads Help). Meta’s Conversion Leads needs the stage within 28 days, and LinkedIn’s qualified leads goal within 30. A deal that closes on day 120 can’t train any of them.
The wider pattern is mapped in the signal gap. This guide turns it into a decision.
Original framework
The Signal Readiness Test
Score every candidate stage, per platform, on four checks. Pick the deepest stage that passes all four. Shallower stages stay as secondary actions for reporting.
Volume
Does the stage happen often enough per campaign or account for this platform? Use the platform’s own figure: at least 15 conversions in 30 days on Google, 200 leads a month on Meta Conversion Leads, five or more qualified leads in two weeks on LinkedIn.
Latency
How many days from click or lead to this stage, typically? Compare the median with the platform’s window: Google ideally within about 7 days for timely optimisation, Meta 28 days, LinkedIn 30 days.
Predictiveness
Of leads reaching this stage, what share later become opportunities or wins, compared with the stage above? If a stage is barely more predictive than a raw lead, it adds delay without adding information.
Stability
Is the stage defined in writing, recorded the same way by every salesperson and channel, and updated within an agreed time? A stage whose meaning drifts month to month teaches the platform noise.
| Stage | Events a month | Median days to stage | Share later won | Defined and updated on time? | Result |
|---|---|---|---|---|---|
| Lead, spam removed | 0 | Fallback | |||
| Qualified lead or meeting booked | |||||
| Opportunity, quote or proposal | |||||
| Closed-won | Usually reporting only |
Reference
Platform by platform: what each can use
Each platform’s published rules for lead-stage optimisation, last checked 15 September 2026. Platforms change these; check the source before relying on a figure.
| Google Ads | Meta | ||
|---|---|---|---|
| Leads that can feed it | Website leads matched by click ID or hashed user data; Google lead forms | Instant Form leads for Conversion Leads optimisation | Lead Generation objective with Lead Gen Forms |
| Volume guidance | At least 15 conversions in the last 30 days; a single funnel stage for bid optimisation | At least 200 leads a month; stage reached by 1–40% of leads | Five or more qualified leads within two weeks to speed learning |
| Timing | Ideally within 7 days of the ad interaction; uploads within 63 days (enhanced conversions for leads) or 90 days (click-ID import) | Stage within 28 days of the lead; upload at least daily | Share qualified leads within 30 days |
| How it’s sent | Data Manager API since 15 June 2026 | Conversions API, with Meta’s lead ID recommended | Conversions API or CRM Sync, not both |
| Source | High-quality leads, value-based bidding, upload windows, Data Manager | Conversion Leads integration | Qualified leads optimisation |
Google gives the most room. Its guidance is to select a single stage of the lead-to-sale funnel as the bid optimisation goal, with a relatively short conversion delay and at least 15 monthly conversions, and to avoid selecting goals from multiple stages for optimisation. Google also offers a shortcut for Search campaigns: qualifying responses on lead forms automatically create a “Lead form – Response qualified (Google-hosted)” conversion (Google Ads Help). That is a self-declared answer, not a sales-confirmed stage, but it can be a useful step up from raw leads.
Meta sets the highest bar. Below 200 Instant Form leads a month, Conversion Leads optimisation isn’t available on a CRM stage, so most B2B advertisers optimise Meta for leads, tighten the form, and judge quality in their own reporting.
LinkedIn sets a low volume bar but a firm 30-day window, and it warns against sending qualified leads through both the Conversions API and CRM Sync, to prevent duplication.
Platform setup is covered on the Google Ads, Meta Ads and LinkedIn Ads pages.
Worked numbers
Worked scenarios: a US SaaS company and an Indian EPC contractor
The same four checks, applied to two very different journeys.
| Stage | Per month | Median days | Share later won | Readiness |
|---|---|---|---|---|
| Demo request | 300 | 0 | 8 ÷ 300, about 2.7% | Plenty of volume, weak prediction |
| Qualified | 110 | 3 | 8 ÷ 110, about 7.3% | Passes volume and latency; about 2.7 times as predictive as a demo request |
| Opportunity | 35 | 18 | 8 ÷ 35, about 22.9% | Strong prediction, but 18 days misses Google’s 7-day ideal |
| Closed-won | 8 | 120 | 100% | Fails volume and every platform window |
| Platform | Leads | Qualified | Opportunities | Stage to optimise |
|---|---|---|---|---|
| Google Search | 130 | 75 | 22 | Qualified as the single bidding stage; opportunity as a secondary action for reporting |
| 80 | 20, all within 30 days | 9 | Qualified leads goal: about ten a fortnight clears the five-in-two-weeks guidance | |
| Meta | 90 | 15 | 4 | Leads: 90 is below the 200 needed for Conversion Leads. Tighten the Instant Form and judge Meta on qualified leads in your own reporting |
Platform rows add up to the company totals: 300 leads, 110 qualified and 35 opportunities.
| Stage | Per month | Median days | Readiness on Google |
|---|---|---|---|
| Enquiry | 60 | 0 | Passes volume; includes vendors, job-seekers and residential enquiries |
| Site visit booked | 24 | 3 | Passes volume (15+ in 30 days) and latency; sales books a visit only for real projects |
| Quote submitted | 9 | 30 | Below 15 a month and slow: review in your own reporting |
| Contract awarded | 2 | 90–180 | Outside every window: validate quarterly |
Steer Google on “site visit booked”. Each quarter, check that campaigns booking more visits also produce more quotes. If a monthly figure drops below 15 during a slow season, keep the goal and review rather than switching back and forth.
When should you move to a deeper stage?
Short answer
Switch when the deeper stage passes four checks: enough monthly events per campaign, typically reached within the platform’s window, clearly more predictive of revenue than the current stage, and consistently recorded by sales. After switching, allow the learning period to finish before judging results or making further changes.
Think of it as a graduation path, one step at a time:
- Leads, with spam removed. The starting point for new accounts and low-volume channels. Make sure spam, duplicates and test submissions never count as conversions.
- Qualified lead or meeting booked. Graduate when sales records the stage within a few days, consistently, and it passes the volume check.
- Opportunity, quote or proposal. Graduate when volume holds up and the stage lands inside the window. This is rarer than teams expect.
- Values by tier. Google suggests tailoring conversion values to how much different actions are worth, including particular lead form answers that indicate higher quality (Google Ads Help). Assign values only after the tiers have been checked against real wins.
Every switch restarts learning. Treat the weeks after a switch as a Wait period: hold bids, budgets and creative steady, and don’t judge the new goal on its first days. For mid-market and enterprise teams, a change to the bidding stage should need approval, because it affects every campaign using it. This is how Zephra frames decisions as Act, Review, Investigate or Wait, with learning-phase protection built into its guardrails; see how it works.
How do you check the choice was right?
Short answer
Compare campaigns on the later stages you didn’t optimise for. If campaigns sending more of your chosen stage also produce more opportunities and wins at a lower cost, the stage predicts value. If the rankings diverge, the stage definition has drifted, sales is qualifying differently, or the stage was never predictive.
- Before and after, on matured cohorts. Compare cost per opportunity by campaign for leads created before and after the switch, once both cohorts have had time to progress. The cost per qualified lead calculator handles the arithmetic.
- Rank agreement each quarter. Rank campaigns by the optimised stage and by wins. Broad agreement is the goal; perfect agreement is unrealistic with small numbers.
- Separate measurement from demand. If you fixed tracking at the same time, reported conversions will jump. That is recovered reporting, not new business.
Failure modes
- Switching the bidding stage every few weeks. Each switch resets learning, so the campaign never settles.
- Optimising to closed-won with five events a month. The platform either ignores the goal or swings on every new deal.
- Counting disqualified leads as conversions. If the “qualified” event also fires for rejected leads, you are still optimising to leads.
- Different definitions by channel. If LinkedIn’s “qualified” is looser than Google’s, cross-channel comparisons mislead.
- Judging inside the learning period. Early swings are expected; cutting budget during them locks in the worst reading.
- Inflated values. Values set from hopes rather than won-deal data push bidding to overpay for the leads that look best on paper.
What changes by company size
- SMB. Low volumes mean staying on qualified leads for a long time is normal. A spreadsheet status updated by sales is enough to start; no CRM is needed.
- Mid-market. Qualified as the bidding stage, opportunity as a secondary action, and a quarterly check of both against wins.
- Enterprise. Buying committees create several leads per account, so count qualified accounts as well as contacts, and put changes to the bidding stage under formal approval.
Zephra uses lead quality and business outcomes from CRM data or a spreadsheet, so no CRM is needed to start, and sends CRM lead stages to Google through the Data Manager API and to Meta and LinkedIn through their Conversions APIs where configured. See which stage to send and lead quality optimisation. How stages are scored or valued is covered in the lead scoring framework.
What to measure next
- Monthly events and median days to each stage, per platform.
- Share of each stage later won, compared with the stage above.
- Days from a status change in sales to the upload reaching the platform.
- Quarterly rank agreement between the optimised stage and wins.
How to structure the stages themselves is in CRM lead stages for ad platforms. To keep statuses accurate and on time, set up a sales feedback loop; for what a qualified lead should mean, see lead quality.
Questions, answered
Lead stage optimisation FAQ
01Can Google Ads optimise for closed deals with a long sales cycle?
It can import closed deals, but they’re rarely a good primary bidding signal when cycles run months and deals are few. Use an earlier stage that predicts revenue, such as qualified lead or opportunity, as the stage bidding optimises to, keep closed-won for reporting and values, and validate the earlier stage against wins.
02Can I use a different stage on each platform?
Yes, and it is often necessary. Meta’s Conversion Leads needs 200 leads a month, while LinkedIn recommends five qualified leads in two weeks. The same business can optimise LinkedIn and Google to qualified leads and Meta to leads, as long as your own reporting compares all three on the same later stage.
03We get 40 leads a month. Which conversion should we optimise for?
Probably leads, with spam and duplicates removed, plus qualifying questions on the form. Check whether a qualified stage reaches 15 in 30 days across the account on Google, or five in two weeks on LinkedIn. If not, stay on leads, pool campaigns, and judge them on qualified leads in your reporting.
04What is the difference between primary and secondary conversions here?
On Google, primary actions are reported in the Conversions column and used for bidding; secondary actions are for observation only, reported in All conversions but not used for bidding (Google Ads Help). Because Google advises a single funnel stage for bid optimisation, make one stage primary and keep the others secondary.
05Should I send conversion values or just counts?
Start with counts. Add values once you have checked, against real wins, that the tiers you would value differently really do convert differently. Values based on guesses push bidding towards the wrong leads.
06How long should I wait after changing the optimisation stage?
At least until the platform’s learning period ends. Google describes up to around 50 conversion events or three conversion cycles; Meta looks for around 50 optimisation events in the week after a significant edit. With low B2B volumes that can take weeks, so avoid other changes meanwhile.
07Do I need a CRM to optimise for qualified leads?
No. You need a record of each lead’s identifier, source, status and status date, which a spreadsheet can hold, and a way to send qualified events to each platform. A CRM makes this faster and less error-prone, but the stage definition matters more than the tool.
Your accounts
Find out which signal your campaigns can learn from today.
The free ad account assessment checks what your conversions count and where measurement gaps sit across Google, Meta and LinkedIn. 10–15 minutes, no CRM required, no campaign changes.