Google's bidding settings make an appealing promise to a busy software founder: tell us what a lead is worth, and we'll handle every auction for you. For a lot of accounts, eventually, that promise holds up. The catch is the word eventually. Google's automation is only as good as the conversion history behind it, and a young software account, above all one in a narrow market, has very little history to give.
A few years ago we reviewed a niche software account that was running Target CPA with almost no conversion data, and almost nothing was happening. The diagnosis we gave at the time was simple: "maybe that's because there's just not that many of these companies." The algorithm wasn't broken. It was starving. There weren't enough qualified buyers in the market to produce the conversions it needed to learn.
This chapter covers how to bid while an account is still earning that data, which columns tell you whether bids are working, and when to hand the job to Google. Chapter 6 split the account in two. Isolation (our label) is the exact match campaign where you type in every bid. Exploration (also ours) is the broad match campaign where Google does the bidding.
Manual Bids Let a New Software Account Pay for What It Has Proven, Not What Google Guesses.
Isolation runs on Manual CPC, with bids set at the ad group level. That sounds old-fashioned next to Google's automated strategies, and it is. It is also the setting that gives you the most direct control over each bid.
Start each ad group's bid from the top-of-page range Keyword Planner showed you in Chapter 5. In our experience, a new account should bid toward the high end at first. Until the account has a record of clicks and conversions, quality signals are thin, and a stronger bid is what gets a new software ad onto the top of the page. As the account builds a record of clicks and conversions, the same position usually costs less, and bids can come down.
Different software categories start from very different places. For scheduling buying searches in our Keyword Planner pulls, the typical high end was about $36. For payroll, it was about $100. Your category's range is the only one that matters, and Chapter 3's payback math tells you how far up that range a click can go before the customer stops paying for it.
A Software Ad at the Top Only Half the Time Hands the Other Half to a Competitor. Hold Top Impression Share Between 75% and 90%.
Once Isolation runs, the first column to add is Search top impression share. Put plainly, it's your attendance record for the paid slots above the results on searches you were in the running for. Its sibling, Search absolute top impression share, measures how often you held the very first position.
For the keywords that bring in qualified leads, we aim for a top impression share between 75% and 90%. Below 75%, you're missing the top of the page on a quarter or more of the searches you've already proven are worth winning. Above 90%, the last few points get expensive, and paying a premium to appear for every single search rarely pays back.
Check these columns weekly in the first months, by ad group where you can, since a campaign average can hide one ad group that never reaches the top and another that wins every auction at a price it doesn't need to pay.
Absolute top is worth watching but not worth chasing everywhere. On a handful of the most valuable searches, holding the first position can be worth the extra cost. On most, a spot among the top ads is enough.
Lost Impression Share Comes From Budget or From Ad Rank. A Bigger Bid Fixes Only One of Them.
When top impression share falls short, two more columns tell you why: Search lost top IS (budget) and Search lost top IS (rank). The same split exists for impressions lost anywhere on the page:
- Search lost IS (budget), the share of the time your ads didn't show because the budget ran out. This is reported at the campaign level.
- Search lost IS (rank), the share of the time your ads didn't show because Ad Rank fell short.
One catch when reading them by ad group: the budget columns exist only at the campaign level, and Google hides the rank column on the ad groups view if the campaign ran out of budget at any point in the period.
The fixes are completely different. A budget loss means the campaign is trying to buy more than you're paying for. Raise the budget if the keywords pay back, or cut the weakest keywords so the money concentrates on the best ones. A higher bid in a campaign that's already out of money just empties it earlier in the day.
A rank loss means your ads are losing auctions. The fix is either a higher bid or a stronger ad and page, which Chapter 4 covered. Check the keyword's Quality Score here. A keyword with a low landing page experience rating will keep losing auctions until the page changes, and a higher bid only makes those losses more expensive.
Some Software Auctions Cost More Than Their Leads Are Worth. Auction Insights Shows Who You're Paying to Beat.
Auction insights is where you see the other advertisers. Six numbers come back for each rival that showed up in your Search auctions. A few are about the rival alone, like how often its ad appeared and how often it reached the top. The rest compare the rival to you, such as how often both ads showed for one search and which ad landed higher.
Read it with one question in mind: who am I actually competing with, and are those auctions worth it? In software, the answer sometimes surprises people. A small CRM can find itself in auction after auction against the two or three largest vendors in the category, companies with far bigger budgets and brands buyers already recognize.
When a rival keeps outbidding you and the cost of matching it would push your cost per customer past what Chapter 3 says a customer is worth, step out of that auction. Cut those bids or cut those keywords, and spend the money on searches you can win at a price Chapter 3 approves. Walking away from an expensive auction isn't losing. It's refusing to pay more for a customer than the customer is worth.
A Niche Software Account Can Go Months Without 30 Qualified Leads a Month. Target CPA Waits Until It Has Them.
At some point, an account earns the right to automated bidding. Google's own guidance on Target CPA offers two markers. It recommends a target equal to your average cost per acquisition over the last 30 days, adjusted for conversion delays, and it suggests judging performance over a period that includes at least 30 conversions. Google also cautions that setting the target below what conversions really cost will lose you some of them.
Our own rule builds on that: Exploration, the broad match campaign, launches when the account reaches about 30 qualified leads a month, with the target set from what qualified leads actually cost in Isolation. For most software accounts that takes two to three months. Bigger accounts get there faster.
A note on names: starting in June 2026, Google is relabeling "Maximize conversions with a Target CPA" as simply "Target CPA," and some accounts may show either name during the switch. It's the same strategy.
Here's the part most bidding advice skips. Some software accounts never get there. If you sell a CRM for funeral homes or scheduling software for climbing gyms, the whole market may not produce 30 qualified leads a month from Google, no matter what you bid. That was the niche account from the start of this chapter: too few companies for the algorithm to learn from.
In that situation, don't force the switch. Keep bidding by hand, because you know more about a small market than an algorithm with a dozen data points does. Or widen the net deliberately: add a closely related audience or a second category your product serves, so the account reaches the volume Google needs. What doesn't work is turning on Target CPA and hoping. A target with nothing to learn from just guesses with your money.
Your Sales Team Can Name the Leads That Were Worth the Bid. The Conversion Column Only Counts Them.
The bid columns tell you how auctions went. They can't tell you which leads were any good. Your sales team can.
Once a month, sit down with whoever works the leads from Google Ads and go through them by keyword and ad group. Which ones became real conversations? Which ones were students, competitors or companies too small to buy? The CRM connection from Chapter 1 makes this easier, because qualified leads come back to Google Ads by themselves, but the conversation still catches what the data doesn't, like a keyword that brings in qualified leads who never buy the paid plan.
Then move bids on what you learn. Raise them on the ad groups that bring buyers, lower them on the ones that bring tire-kickers, and pause the ones that bring neither. In a software account, that monthly review between marketing and sales does more for cost per customer than any bid setting.
Next: Software Buyers Choose on Price and Ease, and They Quote Your Ad Back When It Overpromises. Write Headlines From Their Reviews.
Bids decide where your ads show. Chapter 8 decides what they say, starting from the words buyers already use in their reviews.
Every chapter, from tracking to brand campaigns, competitor campaigns and the rules for free trials, is on the guide's home page. Chapter 6 explains how the exact match and broad match campaigns split the work.
On the software accounts we run through Google Ads management, the first month's bids are set by hand and the sales review starts in week two. Pull your top impression share and lost impression share columns today, and you'll know which problem to fix first.
