Most accounts optimise toward the conversion action that was easiest to install, which is almost never the one that predicts revenue best. Ranked by predictive value instead, the order runs from a qualified sales outcome down to a page view, and the gap between the top and the bottom is the difference between buying customers and buying form fills.
An automated bid strategy will get you more of whatever you tell it to count. That single sentence decides most of the value of a paid search account, and it means the choice of conversion action matters more than the bid strategy, the match types and the ad copy combined.
Which conversion actions actually predict revenue?
The ones a human has qualified. Anything a visitor can complete without intent predicts traffic quality at best and teaches automated bidding the wrong thing.
Ranked by
How reliably the action predicts money arriving. An action that a visitor can complete on a whim ranks below one that required somebody on your side to judge it worth pursuing.
Closed revenue, imported back with its value
First because it is the outcome itself rather than a proxy for it. Nothing predicts revenue better than revenue, and importing it back with a value is what makes value-based bidding mean anything.
The obstacle is lag, not difficulty. If your sales cycle is three months, the signal arrives too late to steer bidding directly, and the platform cannot learn from an outcome it hears about a quarter later.
For short cycles this is straightforwardly the right answer and is under-used. For long cycles it belongs in the account as a measurement truth even when it is too slow to bid on, because it is the only thing that tells you whether the faster signals above it are honest.
A lead your sales team has qualified
Second, and the right target for most lead generation accounts. It carries a human judgement that the enquiry was real, and it usually arrives days rather than months after the click.
This is the entry that changes accounts. Optimising toward qualified leads rather than raw form fills routinely cuts volume and raises revenue, because the algorithm stops chasing the traffic that fills forms and never answers the phone.
It requires the outcome to travel from the CRM back to the ad platform, which is the piece most accounts never build. It is nearly always the highest-return tracking work available.
A booked appointment or a call that lasted
Third. Nobody sits through a scheduled call or a five-minute phone conversation by accident, so the action carries real intent, and it arrives fast enough for a bid strategy to learn from.
A phone call counted at 30 seconds is measuring a wrong number. The same call counted at 90 seconds or two minutes is measuring a conversation. The threshold is the entire difference between this rank and a much lower one, and it is a single field in the call tracking setup.
For businesses that cannot get CRM data back into the platform, this is usually the best available proxy and it is a good one.
A form submission or a raw enquiry
Fourth, and this is where most accounts sit. It proves somebody typed their details in, which correlates with intent loosely enough that a large share of these are never worth calling.
It is not a bad conversion action. It is the default one, it is easy to measure, and it is the right target for an account that has no volume to spare. The problem is treating it as the destination rather than the starting point.
If you are optimising toward this and your sales team complains about lead quality, those are the same fact described from two sides. The bid strategy is doing exactly what you asked.
A download, a signup, or a page view
Last because a visitor can complete these without wanting anything from you. Counting them tells a bid strategy to find people who browse, which it will do very effectively.
These have a legitimate use as secondary conversions: recorded, watched, not optimised toward. They tell you something about how content is performing and they are a reasonable input for audience building.
Marking them as the primary conversion is the single most common way an account ends up spending well against a target nobody wanted. If your conversion count looks healthy and revenue does not, start here.
Why not just count everything?
Because a bid strategy optimises toward whatever you mark as a conversion, so counting a low-intent action tells it to find more of those.
There is a real distinction between recording an action and optimising toward it, and most accounts collapse the two. Record everything you like. Mark as a primary conversion only what you would genuinely pay to get more of, because that flag is an instruction rather than a report.
The clearest symptom of getting this wrong is a set of accounts where cost per conversion improves month after month while revenue is flat. The algorithm has found the cheapest way to produce the thing you asked for. It is not malfunctioning, and no amount of tuning it will help, because the target is the problem.
How do you get qualified outcomes back into the ad platform?
Offline conversion import, which sends the outcome back with the click identifier. It is the highest-value tracking work most lead generation accounts never do.
The mechanism is simple in principle. The click identifier is captured with the enquiry and stored against the record in your CRM. When somebody marks that record as qualified, or as won, the outcome goes back to the ad platform attached to that identifier, and the bidding can finally learn from what happened rather than from what was submitted.
The work is not in the ad platform, it is in the plumbing: capturing the identifier on the form, keeping it through to the CRM, and getting a reliable feed back out. That is why it is skipped so often, and why the accounts that do it tend to be the ones that are hard to compete with.
Before building any of it, work out what a qualified lead is actually worth to you, because that number decides whether the effort is justified and what you can afford to pay for one. The target CPL calculator does that from your own economics. If your existing conversion data is not trustworthy, fix that first, in the order set out in conversion tracking mistakes ranked by how hard the damage is to undo.
Marketing analytics services
Analytics work has two halves. The first is technical: events firing correctly, deduplicated, with the right parameters, server-side where the browser is unreliable. The second is interpretive: knowing which numbers can carry a decision and which cannot. Most reporting problems are actually the second half, dressed up as the first.
Read about Marketing analytics