The short answer
Use Maximize Conversions while you do not yet know what a conversion really costs: new campaigns, new goals, big budget changes. Add a Target CPA once conversions are steady (around 30 in 30 days as a rule of thumb), set at your real 30-day cost, then lower it in small steps. On Meta: highest volume first, cost per result goal later. Both strategies chase more people like the ones who converted, so what you send them matters more than which one you pick.
DataCops is a tool that controls what your bidding learns from: it keeps bots out, sends the real sale from your CRM back to the click, counts each conversion once, and warms up new campaigns with the customers you already have.
How DataCops does it:
- Start warm with Ads Warmup. Upload your existing customers (up to 20,000 rows), see a 0 to 10 match score per person, and send them to Google Ads, Meta and TikTok, so Maximize Conversions starts from real buyers instead of guessing.
- Real people only. Every visit gets a bot verdict against 360+ billion IPs, and every form email is checked (disposable providers, domains with no mail server and an email risk score), so bots never become conversions your target is built on.
- Optimise for the sale. CRM stages go back as offline conversions when they happen: HighLevel natively, any CRM by webhook, matched by click ID or hashed email and phone. Your Target CPA becomes a cost per customer.
- Collected on your own domain. One script and one DNS record, or the free Cloudflare Worker at the edge, so blockers and Safari do not hide real buyers.
- Counted once, logged every time. Pixel and server share an event ID, and a delivery log shows every send as sent, held, skipped or failed.
Best for: lead gen, B2B and SaaS, home services, clinics and HighLevel agencies whose Smart Bidding learns from form fills, and anyone launching new campaigns.
The target that killed a campaign
An example. A home services company has run Maximize Conversions for two months.
Three days later, impressions have fallen off a cliff. Leads are down to a trickle. The verdict in the Monday meeting: Target CPA does not work for us. Switch it back.
Target CPA worked exactly as designed. The strategy was fine. The number was fiction.
And there is a second problem the meeting never got to. Some were bots, some were people who would never buy.
A Target CPA is an instruction, not a wish. Give it a number your account has never hit and it simply stops spending.
Tools that clean what bidding learns from
| Tool | Best for | Watch out for |
|---|---|---|
| DataCops | Filtering bots before conversions are sent, the real CRM sale back to the click, Ads Warmup for new campaigns, counted once, logged; Google Ads, Meta, TikTok and LinkedIn | Needs one script and one DNS record, and a match key at the sale |
| Google's own enhanced conversions and offline import | Google-only accounts with clean data | No bot filter; manual or file-based CRM imports go stale |
| Click fraud tools | Blocking bad clicks on the ad side | Usually do not touch what your CRM sends back |
| Attribution dashboards (Triple Whale, Northbeam, Hyros, Cometly) | Modelling which channel gets credit | Sit downstream; they read the same conversions, junk included. See the Triple Whale and Northbeam comparisons |
| Server-side GTM you build yourself | Teams with tag engineers | Moves data, does not filter it unless you build that too |
Dashboards and data are different jobs. A dashboard is a better gauge; clean data fixes the pipe. If budget forces one, fix the pipe first.
When not to use DataCops
- Your conversions are already clean and Google is your only channel. Google's own enhanced conversions and offline import may be enough.
- You have tag engineers who want to own everything. A server-side GTM container on hosting such as Stape gives full control, if you also build filtering.
- Your question is credit across channels, not data quality. An attribution platform answers that; it still needs clean data underneath.
- Very low volume with no forms. If you get a handful of phone sales a month and no form leads, a simple manual import may do.
What is the difference between Target CPA and Maximize Conversions?
Maximize Conversions spends your daily budget to get as many conversions as it can, with no limit on what each one costs. Target CPA aims for an average cost per conversion that you set. Some conversions will cost more, some less; the target is an average across the campaign.
In Google Ads they are no longer two separate strategies. Target CPA is an optional field inside Maximize Conversions. Leave it blank and you get pure volume. Fill it in and you get volume with a cost goal. Under the hood it is the same Smart Bidding, using signals such as device, location, time, audience and search context to set a bid for each auction.
Maximize Conversions
Spends the full budget. Explores widely. Learns fast on thin data. Cost per conversion moves around, sometimes a lot.
With a Target CPA
Bids only where it expects to hit your number. Cost is steadier. Volume drops if the target is tighter than reality.
| Maximize Conversions | With a Target CPA | |
|---|---|---|
| Goal | Most conversions for the budget | Conversions at an average cost |
| Spends full budget? | Nearly always | Not if the target is too low |
| Needs history? | No | Yes, a stable cost to aim at |
| Best phase | Launch, testing, scaling | Steady state, protecting margin |
| Main risk | Paying too much per conversion | Starving delivery |
| With junk conversions | Chases junk at full budget | Finds junk efficiently, on a fake cost |
When should you use Maximize Conversions?
Use it when you do not yet know your real cost per conversion. That covers new campaigns, new conversion actions, big changes in targeting, and large budget increases.
In those moments the algorithm needs room to test auctions, audiences and times of day. A cost target cuts that testing short before it finds anything. Let it run for a few weeks, watch the cost settle, and resist fiddling. Every big edit restarts the learning.
Where it goes wrong: leaving a mature campaign on Maximize Conversions forever, especially with a generous budget. With no cost limit it will spend the whole budget even on a weak day, and cost per conversion creeps up. If the daily budget is more than you would happily lose on poor traffic, lower the budget before you reach for a target.
The quieter failure: because it chases volume with no guardrail, Maximize Conversions finds polluted conversions fastest. If bots are converting, it goes where the bots are, at full budget speed.
When should you set a Target CPA?
Set one when the campaign produces conversions steadily and the cost has stopped jumping around. A common rule of thumb is around 30 conversions in the last 30 days in that campaign. Google does not enforce a minimum, but a target built on eight conversions is a guess.
Then set it where you actually are. Open the campaign, add the Cost / conv. column, set the date range to the last 30 days, and use that number, or 5 to 10 percent above it, as your first target. Google usually suggests something close.
Lower it in steps of roughly 10 to 15 percent, and wait a couple of weeks between changes. Watch impressions and conversions, not just cost. If delivery falls hard, you went too far. Go back up.
Set the target where you are, then walk it down.
What about Maximize Conversion Value and Target ROAS?
The same pair exists for value. Maximize Conversion Value spends the budget for the most revenue; adding a Target ROAS aims for a return on ad spend you set. Use it when conversions have different values, such as an online store or a sales team closing deals of very different sizes. The same rules apply: run without a target until the numbers settle, set the target at what you actually achieve, and send real values. A value from a refunded or fraudulent order teaches the strategy the wrong thing; see refunds and offline conversions.
What is the Meta version of this choice?
Meta has the same split with different names. Highest volume is its Maximize Conversions: spend the budget, get the most results. The cost per result goal, formerly cost cap, is its Target CPA. Bid cap is a hard limit on each auction bid, for advertisers who know their numbers very well, often in competitive or high-value auctions.
Meta says an ad set leaves the learning phase after about 50 optimisation events in 7 days. That is per week, not per month. Run highest volume until the ad set is out of learning and the cost is steady, then add a cost per result goal at the real average. If you cannot reach that volume, stay on highest volume and work on creative and audience size instead. Sending the real sale helps here too: see Meta offline conversions.
Decision guide: six situations
| Situation | Do this |
|---|---|
| New campaign, no history | Maximize Conversions, no target, conservative budget, 30 to 45 days. Do not judge cost per conversion yet. Warm it up with your existing customers if you can. |
| Steady conversions, stable 30-day cost | Add a Target CPA at your actual 30-day average or slightly above. Wait two weeks before touching it. |
| Scaling the budget on a mature campaign | Add a Target CPA first as a guardrail, then raise budget, then tighten the target in steps. |
| Meta ad set | Highest volume until about 50 results a week, then a cost per result goal at the real average. |
| Bots or junk in your conversions | Neither strategy works. Fix the data first, then choose. |
The part every comparison skips
Every guide treats this as a settings question: pick the mode, hit the threshold, switch at the right time. All of that matters. But it sits on top of something bigger. Both strategies do one thing: find more people like the ones who converted. They trust every conversion you send.
The usual advice, "run Maximize Conversions until you have 30 to 50 conversions, then set a target", assumes those conversions are real. Suppose a fifth of them are bots. You need more data to reach the same number of real ones, the learning is skewed, and the target you set is calibrated on a population that includes non-humans.
The report looks steady. Revenue does not follow.
The best bid strategy in the world still learns from whatever you feed it.
How bad data poisons an account
- The platform finds a pocket of traffic that converts cheaply.
- It bids harder to reach more people like them.
- If that pocket was bots or junk form fills, it now finds more bots.
- Cost per conversion stays flat, because bots keep converting.
- Sales get worse leads, revenue drifts down, and every in-platform number says the strategy works.
That is why switching strategy does not fix it. Target CPA finds the junk efficiently; Maximize Conversions finds it at full speed. The fix is upstream. Our click fraud protection guide covers the traffic side.
Three problems before the algorithm sees anything
1. The pixel is blocked before it fires
Ad blockers and privacy browsers stop browser pixels for part of your real visitors, and Safari shortens how long cookies remember a click. Those may be your best customers, and the algorithm has no record of them. Moving collection to your own domain, and sending from a server, brings them back. See server-side tracking.
2. The consent banner never loads
Most consent banners load from the vendor's domain, which some blockers stop. No banner, no choice, and in basic Consent Mode no tracking at all for those visitors. You never see it fail, because the analytics that would record it did not load either. A banner from your own domain avoids that; see the first-party consent manager.
3. Bot conversions go straight into the platforms
Server-side tracking improves delivery, but it does not filter. If a bot submitted the form, a server faithfully forwards that conversion to Google and Meta. The pipe is cleaner; what flows through it is not. Filtering has to happen before the event is sent.
B2B: a form fill is not a customer
A form fill is not a lead. A lead is not an opportunity. An opportunity is not revenue. Yet most B2B accounts optimise bidding toward form fills, and the algorithm gets very good at finding people who fill in forms: researchers, students, job seekers, competitors and bots.
The fix is offline conversions: send the stages that matter, a booked demo, a qualified opportunity, a won deal, from your CRM back to Google Ads and Meta. The bidding strategy stays the same; the event you feed it changes everything. Target CPA on form fills is a cost per form. Target CPA on won deals is a cost per customer. Read Google Ads offline conversions and B2B SaaS tracking.
Sign-ups are a special case. Account registrations are among the most bot-heavy conversion events there are, and "Sign in with Google" also drops the ad click ID in its redirect. SignupCops keeps that click and sends each sign-up with the click that earned it, while risky sign-ups can be held back.
The audit before you choose a bid strategy
- Primary goals. In Google Ads, Goals, Conversions, Summary: only real business actions set as primary? Page views and "every visit" thank-you pages out.
- Counting. One per click for leads, Every for purchases. Is the pixel and server counted once, or twice?
- Match quality. Enhanced conversions on in Google? On Meta, what does Event Match Quality say for your main event? Low means many conversions are not tied to people.
- Reported vs confirmed. Last month's conversions in the platform next to leads your team confirmed. A big gap means fix the data first.
- Traffic quality. Broad match or Performance Max with no checks bring data centre and bot clicks. What share of your paid visits are not people?
- Form patterns. Disposable emails, repeated details, bursts of sign-ups from one place.
The same audit underpins fixing an inactive conversion tag.
How to switch from Maximize Conversions to Target CPA
- Clean the conversions first. Only real business actions as primary goals, counted once, bots kept out.
- Warm up if you are starting cold. Send existing customers as starting signal with Ads Warmup.
- Run Maximize Conversions until conversions are steady, roughly 30 in 30 days as a rule of thumb.
- Read your real cost. Last 30 days, Cost / conv. column, per campaign.
- Add the Target CPA at that number, not below it.
- Wait two weeks. No other big edits while it adjusts.
- Lower it 10 to 15 percent at a time while impressions hold.
- Move the goal to the sale once your CRM data flows back, and reset the target to a cost per customer.
How DataCops fixes the input
- Ads Warmup. Upload a CSV of existing customers, past buyers, old leads or booked calls. Only email is required. Each person gets a 0 to 10 match score, you pick the event (Purchase, Lead, Complete registration, Add to cart or Schedule), and up to 20,000 people go server-side to Google Ads, Meta and TikTok, dated when you send (on Google Ads, only people who clicked a Google ad are credited). A new campaign on Maximize Conversions then learns from real buyers on day one. Preview is free; sending needs a paid plan.
- Real people only. Every visit gets a bot verdict checked against an IP database covering 360+ billion IPs and 350+ monitoring points, and a Real people only switch per platform keeps bot visits out of what the bidding learns from.
- Fake leads caught. Each form email is checked by fixed rules (throwaway and disposable providers, domains with no mail server, and an email risk score). With LeadCops (Business and up) a lead that fails is held and never billed.
- The sale goes back. HighLevel natively (lead, booked, showed, won with value, paid; cancelled, no-show and lost never sent), any other CRM through a private webhook, matched by click ID or hashed email and phone. Click IDs are kept on the server for up to 90 days, so a deal that closes weeks later still finds its click.
- Collected on your own domain. One script and one DNS record; with DNS on Cloudflare, the free Cloudflare Worker reads the click at the edge. A signed server-set cookie on your domain lasts up to 400 days where enabled, set after consent and a clean bot check.
- Counted once. Pixel and server share an event ID; the same order ID twice counts once, so your Cost / conv. is not halved by duplicates.
- Refund evidence. On the Organization plan, the fraud refund report exports bot-flagged Google Ads clicks in the format Google's Click Quality form asks for. You attach it; Google decides.
- Every send logged. A delivery log row per conversion: sent, held, skipped or failed, with the reason.
- Shopify and analytics. On Shopify, the DataCops Shopify app sends orders the same way. First-party analytics, a GA4 alternative, shows the same real-people traffic your bidding learns from.
One conversion, on its way to Google AdsThe same conversions go to Meta, TikTok and LinkedIn. Setup details on Google Ads conversion tracking and offline conversions. Pricing.
FAQ
Is Target CPA better than Maximize Conversions?
Neither is better on its own. Maximize Conversions gets volume while the campaign is learning. A Target CPA controls cost once you know what a conversion really costs you. Both learn only from the conversions you send.
Are Target CPA and Maximize Conversions the same thing in Google Ads?
Almost. Google now puts Target CPA inside Maximize Conversions as an optional field. Leave it empty and Google spends the budget for the most conversions. Fill it in and Google aims for that average cost.
How many conversions do I need before setting a Target CPA?
Google does not force a minimum, but a target set on a handful of conversions is a guess. A common rule of thumb is around 30 conversions in the last 30 days in that campaign, with a cost per conversion that has stopped jumping around.
Why did my impressions drop after I set a Target CPA?
Usually the target is below what the campaign actually achieves. Google stops bidding in auctions it thinks cannot hit the number. Set the target at your real 30-day average and lower it in steps of 10 to 15 percent.
What is the Meta equivalent of Target CPA?
The cost per result goal, formerly called cost cap. Highest volume is the Meta version of Maximize Conversions, and bid cap is a hard ceiling per auction. Meta says an ad set leaves learning after about 50 optimisation events in 7 days.
Can fake leads make Smart Bidding worse?
Yes. Smart Bidding goes after more people like the ones who converted. If bots fill your form and the conversion is sent, the bidding goes looking for more of them, and your cost per conversion looks fine while sales fall.
Should I optimise for the lead or the sale?
For the sale, when you can. Send the closed deal from your CRM back to Google Ads and Meta as an offline conversion, so the target becomes a cost per customer, not a cost per form.
How do I give a new campaign enough conversions to learn from?
Use what you already know. DataCops Ads Warmup sends a CSV of existing customers (up to 20,000 rows, each with a 0 to 10 match score) to Google Ads, Meta and TikTok as fresh signal, so Maximize Conversions starts from real buyers instead of guessing.
Does DataCops choose my bid strategy?
No. You set the strategy in Google Ads and Meta. DataCops controls what those strategies learn from: real people only, the sale matched to the click, counted once, and your existing customers as a starting signal.
