Google Ads for B2B and SaaS: 288,900 people a month are looking up what your words mean!

B2B PPC does not leak to entertainment. It leaks to education.

Consumer categories lose money to television shows and job hunters. B2B loses it to homework. The vocabulary of this industry is mostly searched by people who want a definition, a diagram for a presentation, or a passing grade. They click, they read, they leave.

Then there is the second problem, which is worse because it is invisible. B2B deals close in weeks or months, and a default conversion window closes in thirty days. Google therefore scores your best campaigns on results it never sees, and confidently shifts the budget toward whatever produces fast, cheap, low-quality leads.

AdPilot connects to the account and audits it the way a consultant would before writing a strategy deck. Searches that were never in market. Tracking that is quietly lying to the bidding algorithm. Ads that generate clicks and no pipeline. Each finding returns as a card, without the jargon. You approve it or you skip it. Nothing reaches your account until you do.

And you stay in control the whole way:

• You approve every change. The AI proposes; you decide. Every budget, bid, keyword, or ad edit arrives as a plain-language card you approve or skip with one tap. Nothing is ever applied on its own.
• A snapshot before every change. AdPilot saves exactly how things looked before anything is touched, so you always have a clean “before.”
• One-click revert. Don’t like a change once it’s live? Undo it in a single click, straight back to that snapshot.
• A full audit log. Who, what, when, the old value and the new value, every change recorded. Nothing hidden.
• You keep the keys. You connect with one secure Google sign-in, and it only ever makes changes you approve. The account stays yours start to finish - you never transfer ownership, and you can disconnect anytime.

Where B2B and SaaS ad budgets actually leak

Real US search volumes from July 2026. Every line contains a word software companies bid on, and almost none of them is a buyer.

The last two are worth a moment. One is an insurance company’s partner portal. The other is an Indian television soap opera, because “saas” is Hindi for mother-in-law.

US monthly search volume, Ahrefs, July 2026. Around 288,900 searches a month, and hardly a buyer among them.

Negative keywords for B2B and SaaS. Free, no email, no form.

Use it whether or not you ever become a customer. Ten minutes, no cost, and definitional traffic is the one leak we find in nearly every B2B account we look at.

Definitions and coursework

meaning, means, definition, stands for, examples, example of, explained, tutorial, course, certification, exam, pdf, ppt, slideshare, assignment, thesis

Job hunting

software jobs, engineering jobs, sales jobs, job openings, salary, careers, now hiring, we are hiring, recruiter, resume, cv, interview questions, internship, glassdoor

Free and pirated

free download, freeware, open source, crack, cracked, nulled, torrent, license key, serial key, lifetime deal

Analyst and market research

market size, market share, statistics, industry report, gartner, forrester, trends 2026, forecast

Wrong industry entirely

state farm b2b, state farm login, state farm b2b login, kyunki, saas bahu, hindi serial, full episode

Students and templates

powerpoint template, excel template, notion template, free template, sample report, sample dataset, worksheet, cheat sheet, for students, homework

B2B PPC: five things that matter more than your bids

• Send closed deals back to Google. Offline conversion import is the single highest-impact thing in B2B paid search and the thing most accounts skip. Without it Google optimizes toward form fills, and form fills include students, competitors and job applicants.
• Match the conversion window to your sales cycle. A default window of thirty days against a seventy-day cycle means the campaigns producing real revenue look like your worst performers. Fixing the window sometimes reverses the entire ranking of your campaigns overnight.
• Bid on alternatives, and expect to be bid on. “[Competitor] alternative” is cheap, high intent and full of people already unhappy. Your competitors are doing it to you right now, which is also why your own brand terms deserve a budget.
• Exclude the people you already have. Upload a customer list as an exclusion. Otherwise a meaningful slice of your budget goes to existing customers looking for the login page, and they convert beautifully, which makes the waste invisible.
• Judge lead quality, not lead count. Cost per lead is the metric that most often flatters a failing B2B account. Cheap leads are cheap for a reason. Cost per qualified opportunity is harder to get at and the only one worth acting on.

What one demo request is worth at your contract value

Four numbers your pipeline already has, and the one most B2B accounts are run without. The click price is measured: “crm software” is 285,000 searches a month at about $4.00 and “project management software” 142,000 at about $0.70, from the same pull as the table above - which is itself the lesson, because those two sit in the same buyer's head. The contract value and the close rate are a starting point only. Nothing leaves your browser.

The build, in the order it has to happen

Below is the same order AdPilot's agent works through, set out so a founder with a laptop can run it without us. We publish it because you should get to check the method before you trust software with a budget, and because most people who read it decide they would rather approve cards than make forty judgment calls every other week.

Two evenings and a Saturday gets it built. After that it wants an hour every other week, forever. Building it is the easy part. The hard part in B2B is that the verdict on what you built lands a quarter after you built it, so you spend three months steering on numbers that are not revenue.

Building the same thing for another industry?

1. 1

   Sort the buyers out of 288,900 searches that never buy

   Every month, 288,900 searches in the US use the words your industry runs on, and almost none of them come from somebody who will ever buy software. 222,700 of those are people circling the word SaaS itself, mostly looking for a definition. 37,600 want a B2B definition. 22,000 are trying to reach State Farm's partner portal, and 6,600 want an Indian soap opera, because saas is Hindi for mother-in-law. They click, they read half a paragraph, they leave. Their clicks are cheap, and cheap traffic goes first: the budget is gone by mid-afternoon and the person who was going to book a demo never saw an ad.

   The buying side is a far shorter list, and the prices on it refuse to behave. "CRM software" runs 285,000 searches a month at about $4.00 a click. "Project management software" runs 142,000 at about $0.70. Same buyer, same browser tab, same Tuesday afternoon, and one costs nearly six times the other. That gap is why you price your own terms before you write a budget, instead of borrowing a category rate you read somewhere.

   Six shapes of search are worth paying for and you can tell them apart on sight. The competitor comparison, from somebody who has already decided the thing they own is wrong. The category term with a qualifier bolted on, which is the difference between "crm" and "crm for wholesale distributors". The demo search. The trial search. The integration search, where they name the system they already run, because nobody types the name of an ERP idly. And the job-title shape, "crm for route sales reps", which is half buyers and half people looking for work, so it needs a qualifier and a look at what it actually brings in.

   Then read every term cold, with nobody standing beside you explaining what you sell. Bare "crm" is coursework, certification and "what is a crm". Bare "platform", bare "solution" and bare "tool" are worse. A one-word category term isn't a keyword yet, it's a draft missing the word only your buyer would type: their industry, their team size, the system it has to talk to, the job title of the person who opened the tab. No match type rescues it either. Phrase carries all the other meanings, and broad reads your landing page and guesses.

   One more habit, and it is the least technical thing in this guide. Keep a text file of every term you rejected and why. In month three somebody will ask why "crm" is not in the account, and "because it means eleven things" is a better answer with the eleven written down.

   • Before a term goes in: who typed this, and what is on their screen twenty minutes later?
   • Words that mean a budget: alternative, vs, pricing, demo, trial, migration, integrates with, for [their industry], for [a job title].
   • Words that mean homework: meaning, what is a, stands for, examples, template, tutorial, certification, market size, salary.
   • A term with a qualifier and 40 searches a month beats a bare category word with 40,000.
   • Search volume is a bill, not a forecast. The 41,000 searches a month for "saas meaning" will happily spend your quarter.

   Five worth paying for, five to block, with the measured numbers where we have them

   • crm software - 285,000/mo, ~$4.00 - BUY, and the most expensive room in the building.
   • project management software - 142,000/mo, ~$0.70 - BUY, and six times cheaper for a buyer sitting in the same chair.
   • [competitor] alternative - COMPARE. Somebody already unhappy with what they own, and step three separates this from the expensive kind of competitor bidding.
   • crm that integrates with [the system they already run] - BUY. Nobody types the name of an ERP for fun.
   • crm for route sales reps - BUY, mostly. The job-title shape also pulls in people looking for work, so watch what it brings back.
   • saas - 144,000/mo - READER. The single largest leak in this industry.
   • saas meaning - 41,000/mo - READER. Block it.
   • b2b meaning - 22,000/mo - READER. Block it.
   • state farm b2b - 22,000/mo - an insurance company's partner portal. Block the portal phrases, never the brand on its own, and read step five before you paste anything.
   • kyunki saas bhi kabhi bahu thi - 6,600/mo - a television soap opera. Block it and enjoy telling someone why.
2. 2

   The page has to answer the question they typed

   One ad, one page, and the page is where a $4.00 click either becomes a calendar invite or doesn't. A homepage answers "who are you". Somebody typing "[competitor] alternative" is asking something narrower and more urgent: why should I move, what breaks if I do, and how long does it take. A B2B account needs two landing pages before it needs a fourth keyword - the industry or use-case page, and the page where a demo actually gets booked. A third, the comparison page, arrives the day you decide to bid on somebody's name, and it arrives before the campaign does, not after.

   Then the argument you are going to have with somebody internally. Put a price on the page. A number, a range, or the model, per seat or per user or per order, whatever yours is. Plenty of B2B software hides pricing behind "Contact sales" and then pays $4.00 a click for people whose first and only question is what it costs. A price filters those people for free. If sales insists on hiding it, fine, but that is a decision to buy those clicks, and it should be made out loud with the number attached.

   Now count the fields on the demo form. Eleven of them, including "how did you hear about us" and a company-size dropdown that starts at 500 employees. Each field is a percentage of the people whose search you just paid for. Ask for the two things a rep genuinely cannot start without and get the rest on the call. Better again, let them pick a slot on a calendar. "We'll be in touch" is where a hot lead goes cold over a weekend.

   Before you close the tab, make a second list: who you don't serve and what you don't do. No free tier. Nothing under five seats. Not for freelancers. No on-premise version. Write that list down, because it turns into the block list in step five, and the money it saves is bigger than anything you will ever save on bids.

   • The first heading names the category and the buyer, not your positioning line.
   • A price, a range, or the pricing model, visible without filling in a form.
   • The demo booked on a calendar in the page, not promised by email.
   • Proof a stranger can check: a named customer in their industry, a number, a case study.
   • The form's field count, written down. Then cut it.
   • One action per page. A trial button and a demo button on the same page split the traffic you paid for.

   The eleven-field demo form, cut to four

   • KEEP: work email. The one field a rep genuinely cannot start without.
   • KEEP: company name. It is how the rep arrives at the call already knowing something.
   • KEEP: what you are trying to fix, one line, optional. The best qualifying field on the form and almost nobody has it.
   • KEEP: a calendar with real slots on it, in the page. Not a promise of contact.
   • CUT: phone number. Ask on the call, or ask after the slot is booked.
   • CUT: company size dropdown starting at 500. Your buyer has thirty people and just learned they are too small for you.
   • CUT: job title, industry, country, number of reps. Your CRM enriches all four from the email domain.
   • CUT: "how did you hear about us". You have auto-tagging for that, and it is in step six.
   • CUT: the checkbox that says a rep will be in touch within two business days. That is not a form field, that is a warning.
3. 3

   The trial and the demo never share a budget

   A campaign is a budget with a goal attached. That is the only reason to open a second one. A trial and a sales-assisted demo are two goals, and not slightly: different conversion action, different value, different payback, different humans on the other end. Share a budget between them and watch what happens. Google pools the conversions, sees that one of them converts in ninety seconds and the other needs somebody to open a calendar, and pours the money at the fast one. Every time. Two campaigns, two budgets, two conversion actions. Not two accounts, whatever you read - running one business out of two Ads accounts against overlapping keywords is what Google's unfair advantage policy is for, and you meet it as a warning followed by a suspension. Never two ad groups.

   Now do the arithmetic out loud, before the campaign exists. Two numbers have to clear at once. Enough clicks a day to read anything, which is about ten for a campaign with no history and never fewer than five. And a daily budget that is a decent multiple of what one conversion costs: Google's own guidance says at least twice your target, everyone I know who runs these accounts wants three times so the algorithm can afford the expensive auction as well as the cheap one, and three is what the sums below use. At $4.00 a click, $40 a day is exactly ten clicks. It scrapes the first number and fails the second badly. You either pay the real price of the expensive room or you go and find a cheaper one. More keywords is not a third door.

   Everything starts as phrase match. When a specific search has actually produced demo requests, promote it to exact and give it a headline and a page of its own. That is where a mediocre account turns into a decent one. Leave broad match off. It only behaves when the bidding is conversion-based, the tracking counts something real, the negative list has an owner, and the account sees enough conversions a month to feed all three. Miss any of those and it spends your quarter proving it cannot tell a buyer from a student. And whatever anyone told you, exact match has matched close variants and same-meaning queries since 2018. There is no setting to turn that off.

   Two different things get called competitor bidding, and only one of them is expensive. "[Competitor] alternative" and "[competitor] vs" are comparison searches. Somebody has already decided what they own is wrong and is shopping. Those belong in the demo campaign from day one, pointed at a comparison page, and they are among the cheaper buyers you will meet. Bidding on the bare brand name is the other thing. You are the ad nobody searched for, so the click-through rate is poor, the quality score follows it down and the cost per lead goes up, and you cannot print their trademark in the ad, which leaves you the weaker ad on their own name. Nobody can tell you the multiplier in advance, including us. Run it small, measure it against your own non-brand cost per lead, and kill it if it has not closed the gap inside sixty days.

   Your own name gets its own campaign, always. Brand clicks cost a fraction of everything else and close at rates non-brand will never touch. Blend the two and the average flatters you: the dashboard reads healthy while the non-brand half quietly loses money underneath, which is the most common reason a B2B account looks fine on paper and isn't. Before you build it, search your own name and look at what comes back. Top of the free listings with no ad above you, and this campaign is mostly buying clicks you already had.

   Run the sums before you open the campaign, not after

   • $40/day at ~$4.00 a click, which is "crm software" money, buys 10 clicks a day and 304 a month.
   • At 3 demo requests per 100 clicks - assume it now, then check it against your own form in a month - that is 9 requests a month at about $133 each.
   • Three times one conversion is about $400/day, over $12,000 a month, before that campaign has the budget the guidance asks for at that click price.
   • The same $40 at ~$0.70, which is "project management software" money, buys 57 clicks a day and about 52 requests a month at roughly $23 each. Three times $23 is $70/day, and that is a floor you can clear.
   • $70/day at $4.00 is 17 clicks a day and about 16 demo requests a month, still at $133 each. Say the ugly part out loud: $70 clears the click floor and comes in short of the three-times-a-conversion guidance by a factor of nearly six. You are buying 16 monthly rows to read, not a bid strategy that works. So run maximize conversions with no target and judge the campaign on those 16 rows. A target CPA typed into a campaign with no history is a wish, and it strangles delivery to almost nothing.
   • Put those two together and the shape of the account falls out. $70/day clears the floor at $23 a request and misses it by a mile at $133, which is exactly why the build below sits on qualified distributor phrases rather than the bare category word: the longer phrase is less contested, the click is cheaper, and the floor comes back within reach. If your own terms really do price at $4.00, do not run $70/day and hope. Narrow the keywords until the click comes down, or fund one campaign properly and build nothing else.
   • The daily figure is an average, not a cap. Google will overspend a single day by up to double it, then hold the month to 30.4 times the number you typed. $70/day is a worst case of $2,128, not a promise of $70 on Tuesday.
   • $40 split across a demo campaign, a trial campaign and a competitor campaign is $13 a day each, which at $4.00 is three clicks. Thirty days later you have about a hundred clicks per campaign, no conversions worth reading, and three verdicts you are not entitled to.

4. None of this is difficult. It is a long run of small decisions, and each one quietly costs money when it goes the wrong way. The agent makes them for you and shows you every single one before it touches the account.

   Start free - no credit cardFree for 7 days. No card, no call, cancel whenever.
5. 4

   Twelve headlines, and the three that decide the ad

   Google will take 3 to 15 headlines at 30 characters and 2 to 4 descriptions at 90. Ten or twelve headlines is the working number, and each one has to earn its slot. Two headlines saying the same thing in different words is a slot you paid for that taught the system nothing. Count every character yourself, commas and spaces included, because 31 gets refused and you lose the line you liked. Spell the search phrase into a headline the way buyers actually type it. Google bolds the matching words wherever that headline lands on the page, which is the whole reason to hand them their own wording back.

   Then the part almost nobody does. Google assembles any three of your headlines, in an order you don't pick, and you pay full price for every trio it serves. B2B is the worst offender in advertising for headlines that name nothing at all. "Built For Modern Teams", "One Source Of Truth", "Scale With Confidence" is a finished, fully paid ad for a data warehouse, a payroll product, a shipping API or a note-taking app. So find the three of yours that name the product least, read those three together, and get somebody outside the company to tell you what is being sold. If they can't, you have twelve headlines and no ad.

   Say the subject in every headline. Say the same thing about it twice and you have wasted a slot. "CRM For Distributors", "Book A 20 Min CRM Demo" and "CRM From $29 Per Seat" repeat the word CRM and nothing else: the first is the search term, the second is the offer, the third is the price. Sets die when five headlines make one claim in five outfits. In this industry that claim is security and credibility: "SOC 2 Compliant", "Enterprise-Grade Security", "Bank-Level Encryption" and "Trusted By 400+ Teams" are one claim. It gets one slot, two if you're proud.

   The trademark rule catches everybody once. You may bid on a competitor's brand name as a keyword. You may not print it in your ad text once they have filed a complaint with Google, and any competitor worth bidding on has. Assume disapproval and write around it, so the switching promise does its work without the name: "Switch Your CRM In A Week", "Import Your Pipeline In A Day". The comparison page you send them to can name them all it likes.

   Do not write twelve headlines of identical length. At 29 and 30 across the board the page fits two, and the third one you counted so carefully never appears. Two or three genuinely short lines, still naming the product, keeps all three slots in play.

   Seven drafts with the count, and the two that had to go back

   • CRM For Distributors (20) - the term, in the words they type it.
   • Book A 20 Min CRM Demo (22) - the offer, and how long it takes.
   • CRM From $29 Per Seat (21) - the price, which filters harder than any negative keyword you can write.
   • 410 Distributors Use Our CRM (28) - a count somebody can go and check.
   • CRM For Building Suppliers (26) - the buyer's own word for themselves, not yours.
   • WEAK: Built For Modern Teams (22) - modern teams of what, doing what, with what?
   • FIXED: CRM For Wholesale (17)
   • WEAK: One Source Of Truth (19) - equally true of a spreadsheet, a wiki and a warehouse.
   • FIXED: CRM That Syncs To Your ERP (26)
6. 5

   The block list, and the words that must never be on it

   The block list is not maintenance you get around to later. It goes in the night before the campaign is allowed to serve, because hour one of a software campaign is definitions, coursework and somebody hunting an insurance portal. Build it at campaign level, and start where nobody can help you: what you do not sell and who you will not take money from. That part is worth more than the rest of the list combined.

   Now the six words that must never go on the list on their own, each with a paying customer standing behind it. Never bare "free", because it kills "free trial", which is the offer you are advertising. Never bare "trial" either, the same problem from the other direction - it takes "crm free trial" down with it, and that is the campaign you just funded. Never bare "cheap": "cheap crm for small business" is a five-person team with a company card. Never "alternative", which is the strongest buying signal in this industry and the most expensive negative anybody has ever pasted into an account.

   Never bare "open source" if your buyers write code. "Open source alternative to [the thing they hate]" is a comparison shopper with a budget and a preference, not a freeloader. The free list higher up this page blocks it outright, which is the right call for a finance or HR product and the wrong one for a developer tool. If that is you, take it off and block the phrases instead: "open source projects", "best open source", "open source list", "self hosted open source". And mind which form of the last word you touch. Bare "job" must never go on the list. "Crm for job shops" is a machine shop with thirty people and a purchase order, and job costing is an entire software category with its own buyers. "Jobs" is the safer form - by the plural rule below it will not even reach "job shop" - but block it as phrases anyway, because a bare negative catches searches you have not thought of yet: "crm jobs", "sales jobs", "job openings".

   Here is the property that makes people swear negatives are broken. A negative sits still. Your keywords stretch and it does not: plurals aren't matched and synonyms aren't matched, so "template" leaves "templates" running all month and "salary" never touches "salaries". Add both forms every time, then read the list back and find the ones you only wrote once. Misspellings are the exception now, since Google started matching obvious ones on negatives in June 2024, so you can stop spelling things wrong on purpose.

   Use the smallest tool that stops the bleeding. Negative broad catches your words in any order anywhere in the query, which is how somebody blocks "free" and then wonders where the trial signups went. Negative phrase wants those words in that order, and that is the right setting nine times out of ten. One more warning before you paste: at least one line below is not a decision you can copy. If you sell HR or payroll software, "salary" is your category, not your leak.

   The day-one list, written as phrases, with the lines nobody can write for you

   • The free list higher up this page is this thinking without the reasoning attached. What follows is the same ground in phrase form, which is the form that survives contact with a live account.
   • what is a, what is an, what is the meaning, what does a, meaning of, means in, stands for, definition of, explained simply, examples of
   • free template, template download, template excel, template google sheets, template pdf, sample template, checklist pdf, cheat sheet, ppt, powerpoint, slideshare, thesis, assignment, homework
   • certification course, training course, online course, exam questions, interview questions, udemy, coursera
   • salary, salaries, crm jobs, sales jobs, software jobs, job openings, now hiring, glassdoor jobs, indeed jobs, indeed com, internship
   • free download, free downloads, crack, cracked, nulled, torrent, license key, serial key, lifetime deal, free forever plan
   • market size, market share, industry report, industry statistics, market statistics, forecast 2026, magic quadrant, forrester wave
   • state farm b2b, state farm login, state farm b2b login, kyunki, saas bahu, hindi serial, full episode
   • Whoever you don't serve, as phrases: for freelancers, for students, for personal use, single user, on premise, self hosted
   • Your own support traffic, on every campaign including brand: login, log in, sign in, cancel subscription, customer service number, status page. An existing customer hunting the sign-in page converts beautifully and buys you nothing, and on the brand campaign that is exactly where the flattery comes from.
   • Careful, and this is the part people paste past: never bare free, never bare trial, never bare cheap, never bare alternative, never bare job, and never bare open source if you sell to engineers. Never bare "what is" or bare "best" either - "what is a crm" is homework, "what is the best crm for small business" is somebody about to buy one, and a single negative phrase takes them both. Never bare "template" if your product ships them, because "crm with email templates" and "quote template software" are feature searches from people with a card. Never bare "state farm" if you sell into insurance, since "crm for state farm agents" is probably the best keyword you own - block the portal phrases instead. Take "indeed" and "glassdoor" off entirely if you sell recruiting or HR software, because "ats that posts to indeed" is your category. Take "salary" off if you sell HR or payroll. And take "magic quadrant" off the day you are actually in one, because then that search is somebody looking for you.

7. Everything past this point is the work that never finishes: reading search terms, adding negatives, checking what the budget did overnight. That is exactly the part the agent takes off your hands.

   Start free - no credit cardFree for 7 days. No card, no call, cancel whenever.
8. 6

   Bid on the meeting, not on the form

   The default conversion window is 30 days and your deals take 74. That one mismatch does more damage in B2B than every bad keyword put together, because Google is scoring your campaigns on results it never sees and moving money toward whatever produced a fast, cheap, unqualified lead. Open the conversion action and widen the click-through window to cover your real median. Google allows up to 90 days. Do that before you touch a single bid.

   Then decide what actually counts, and be mean about it. The form fill is the default and it is the worst thing you can bid on, for the reason in the FAQ above. What follows is the ladder, and you take the highest rung your stack can genuinely reach, not the highest one your CRM vendor claims. Top rung: the qualified opportunity, pushed back as an offline conversion import, which means your CRM stores the click ID that arrived with the lead and hands back which of them turned real. Middle rung: the booked meeting, fired by the scheduler when a slot is confirmed, not when a form is submitted. Bottom rung: the demo request with a value on it, so at least the count is weighted.

   The trial side has its own trap, and it is the one that makes a campaign look brilliant while it loses money. Cheap signups are easy to buy. If a signup costs $12 and one in twenty-five ever pays, you are $300 into acquiring a customer before a human has touched the account, and signups that arrived from a "free crm" search convert to paid at a fraction of the rate of the ones that came off a comparison page. Google cannot know that unless you tell it. Either count the upgrade instead of the signup, once you have enough upgrades a month to bid on, or put a value on the signup: first-year revenue times the share that pays. And check the split before you argue about it. Pull last quarter's signups by the search that produced them and set the paid conversion rate of each next to the other. If the gap is not obvious, you don't have a trial problem and you can skip this whole paragraph.

   Two things are probably already broken in the account, and both are switches rather than projects. Auto-tagging: with it off, nothing connects a closed deal back to the click that started it, and campaigns that are working look dead and get paused for it. And conversions Google already built from your Analytics and left switched off, sitting there waiting for somebody to notice. Neither one needs a developer.

   Last, the thing nobody says out loud. Google counts from the day you turn tracking on and never fills in the past. With a 74-day cycle, your first honest read on revenue is next quarter, not next month. What you steer on in the meantime is the leading stuff: cost per demo request, the share of demos that actually show up, the share a rep marks qualified. Anyone offering you a revenue verdict inside 30 days has not looked at your sales cycle.

   • Widen the conversion window to your real median sales cycle, up to the 90 days Google allows.
   • Turn auto-tagging on if it is off, so the click ID reaches your CRM at all.
   • Look for conversions Google imported from Analytics and left disabled. Switch them on.
   • One primary conversion per campaign: the booked meeting on the demo side, the paid upgrade or a valued signup on the trial side.
   • Put a number on it. First-year contract value times your close rate, your figures, not a benchmark from a blog post.
   • Write down today's date. It is the first day of data you will ever have, and the months before it are not coming back.
9. 7

   Two weeks of noise, then the read that pays

   Days one and two, two questions only. Is money leaving the account, and did the ad clear review? A campaign that physically cannot serve files exactly the same empty report as one that is simply new, so silence on day two is a fault you go and find, not a start you wait out. A disapproved ad. A target somebody should never have typed. A location setting that quietly excludes half your market.

   Day seven is a negatives-only visit. Budgets stay put, targets stay put, nothing gets harvested and nothing gets concluded. Week one is noise, and in B2B it is noise twice over, because nothing has had time to close and half your form fills have not been called back yet. A bidding strategy that is still learning will punish you for fiddling with it, and that status usually clears in about a week, sometimes two. Don't confuse it with the other number people quote at you: roughly 30 conversions in 30 days before a target is worth typing. That one is about whether the strategy is any good, not about whether it has finished learning.

   Day fourteen to thirty is the first real read, and only on leading indicators. Cost per demo request by ad group. The share of requests a rep marks qualified, which is the number that separates a campaign that works from one that is merely cheap. Then search terms again, this time for waste and for gold: any search that produced a request gets promoted to exact match with a headline written for it. One hour of harvesting pays for more than everything else you do in the account put together, and it is the same hour every other week for as long as the campaign runs.

   Month two and month three, every campaign gets one verdict: fund it, fix it, or shut it. A campaign sitting at three times your target cost per qualified demo with nothing behind it, after a full month of tracking that works, is not still learning. And know what actually kills these accounts. It isn't the build. It is the quarter where you closed a round, shipped a release, hired two people, and nobody opened the search terms report. The searches arrive whether anybody reads them or not, and so does the invoice.

   Fourteen days of search terms, and the decision each row forces

   • "crm for wholesale distributors" - 22 clicks, 3 demo requests -> add as exact, and give it its own ad.
   • "crm that integrates with our erp" - 7 clicks, 1 request -> exact, and a landing page that names the integration.
   • "crm for job shops" - 5 clicks, 1 request -> exact. Also the reason bare "job" is nowhere on your list. Negatives don't match plurals, so "jobs" would have sailed straight past this one, and "job" would have killed it outright.
   • "how does a crm work" - 41 clicks, 0 -> negative phrases "how does" and "how do". Your day-one list blocked "what is a" and never saw this wording.
   • "crm software free download" - 17 clicks, 2 trial signups, 0 upgrades, and this row is from the trial campaign, not the demo one -> negative phrase "free download". Not bare "free", which would take "crm free trial" down with it.
   • "crm administrator jobs" - 9 clicks, 0 -> negative phrase "administrator jobs". You blocked "crm jobs" and it never reached this one.
   • "crm reviews reddit" - 14 clicks, 0 requests -> leave it one more month. That is a real buyer going off to watch strangers argue, and a fair share of them come back by name, where your brand campaign quietly takes the credit.
   • "state farm b2b login" - 3 clicks, 0 -> negative phrase "state farm b2b". You would never have guessed it, and it is in your report because 22,000 people a month want an insurance portal.
   • Also, that table is a sample, not a ledger. Searches under a privacy threshold are withheld from the report entirely, so the rows will always add up to less than the invoice. Never tell anyone you accounted for every dollar.
   • The row that matters most isn't here yet. Which of those five demo requests became revenue lands 74 days from now, which is the whole reason the offline import in step six is the difference between steering and guessing.

The demo ad, counted character by character

This is one ad group, demo requests for a distributor CRM, written for a software company that does not exist. Twelve headlines, every one of them counted by hand, plus a single description. The preview draws the first three the way a results page would, and the other nine sit underneath as the pool Google picks from: it assembles any three of them, in an order you don't choose, so every headline has to survive being read beside any other two. Google will not save the ad with fewer than two descriptions and will not take more than four, and it shows one or two of them depending on how much room the page has - on a phone, usually one.

• The keyword goes in two headlines, the subject goes in all twelve. "CRM For Distributors" and "Distributor CRM Software" carry the search phrase itself, and Google bolds those words wherever the headline lands. Google asks for three headlines that are not the keyword, and here they are: "Sales Software Demo, No Pitch", "Switch Your CRM In A Week" and "Book A 20 Min CRM Demo" sell the offer rather than repeat the term. Repeating the subject is required. Repeating the phrase is padding.
• Read the weakest trio, not three you picked at random. Sample three at random and they will read fine, because you will unconsciously reach for the good ones. The three here that name the product least are "Sales Software Demo, No Pitch", "CRM Set Up In 3 Days" and "Switch Your CRM In A Week". Read them cold as one ad and a stranger still knows it is sales software, that they can look at it without being sold to, and that the move takes about a week. Compare that with the trio every software account ships by accident, "Built For Modern Teams | Scale With Confidence | Start Free Today", which is a complete, fully paid ad that never says what is for sale.
• There is no free trial in this ad, and that is deliberate. The trial has its own campaign, its own budget and its own conversion action, for the reason in step three. Offer both in one ad and every click takes the cheaper action, the demo campaign learns that its job is collecting signups, and the pipeline thins out while the report gets prettier.
• A competitor's name cannot appear here, however much you want it to. Bidding on their brand as a keyword is allowed. Printing it in the ad text stops being allowed the moment they file a trademark complaint, and any competitor worth bidding on filed one years ago. So the switching promise does the work without the name: "Switch Your CRM In A Week" and "CRM Set Up In 3 Days" are what somebody arriving from "[competitor] alternative" needs to read. Save the naming for the comparison page, where you can put the two side by side and be specific about it.
• The $29 is a qualifying question you never have to ask. Publishing the seat price is the cheapest qualifying you will ever do. A five-person shop with no budget reads $29 a seat, decides it is real money, and does not cost you $4.00 to find that out. If you truly will not print a number, print the model instead, and go in knowing what the alternative costs you per click.
• One description is shown here, and Google will not save the ad with fewer than two. It shows one or two depending on the room the page has, so the box above is the one written to carry the ad by itself. Write the second one around the money and the risk: the seat price, what the twenty minutes actually involves, and that nobody gets a sales call for saying yes to a demo. Add a third only when you have a claim the other two haven't made. A reworded twin just hands Google a weaker version of your own description to spend your budget on.
• Twelve headlines, twelve jobs, and one trust slot. The term, a variant of it, the demo offer, the price, a count somebody can check, the no-pitch reassurance, the ERP integration, the switch, the quote-to-order job, the buyer's own word for themselves, the setup time, and one plain ask. What is not in there is the security paragraph four times over. "SOC 2 Compliant", "Enterprise-Grade Security", "Bank-Level Encryption" and "Trusted By 400+ Teams" are one claim in four suits, and a comparing buyer needs the other questions answered instead.
• The lengths run 18 to 29 characters on purpose. Three headlines get shown only if three will fit. Write every line at 29 or 30 and the page routinely drops you to two, so a third of what you counted never runs at all. "Quote-To-Order CRM" is eighteen characters and still names what is for sale, which is all a short headline has to do.
• Sitelinks: four small links, and most software accounts ship without any. Skip them and your ad occupies visibly less of the results page than the competitor underneath you, at the same click price. These four take the questions that arrive second: what it costs per seat, whether it talks to what we already run, whether anyone like us has done this, and how bad the move is going to be.

The shape a seed-stage B2B account actually runs

Two campaigns doing the work, one small defensive one, five ad groups across the account, and one campaign deliberately left unbuilt. Everything goes in as phrase match. Exact is a promotion, earned by whichever searches actually produce a demo request. The total is $100 a day, a little over $3,000 a month at the ceiling, which is roughly where a seed-stage company can run this without pretending.

Why it is shaped like this. The trial and the demo are separate campaigns because they are separate goals, not because the keywords look different: one counts signups and gets judged in a week, the other counts booked meetings and gets judged next quarter, and on a shared budget the fast one eats the slow one every time. The three demo groups share a budget and a goal but need different ads and different pages, and five ad groups is already near the limit at this volume - a group seeing two demo requests a month teaches Google nothing at all. Your own name gets its own campaign for the reason in step three, and the arithmetic is not close: blend it in and the account average lies to you every month. "Login" is a negative everywhere, brand included, for the reason in step five. The bare-brand competitor campaign is the one missing on purpose. Comparison terms already sit in the demo campaign where they belong; buying a rival's name outright earns a budget once the demo campaign is landing meetings and you can afford sixty days of losing.

PPC management for B2B and SaaS, shown as cards you approve one at a time.

What you actually pay a B2B agency for is the hands-on part. Structuring campaigns by buying stage, digging through search terms for definitional traffic, getting the conversion window and the offline import right, writing ads for people mid-comparison. AdPilot’s AI does that work, with the reasoning attached to every recommendation.

Examples of what the cards look like - not results from your account

A broken conversion window does not wait for the monthly report.

You keep control, and you stop being dependent.

Plenty of founders hand paid search to an agency because it is opaque, and it stays opaque because nobody has a reason to explain it. AdPilot is built the other way round. Every card tells you:

• What it changes - for example, pause a keyword that spent $611 and produced no demo requests.
• Why it helps - the reasoning, not just the instruction.
• What it is based on - the numbers in your own account.

Approve one and a snapshot is saved first. Then a before and after dashboard shows what happened to cost per lead, spend and conversions, in plain words. After a month or two you can hold an agency conversation on equal terms, whether or not you still have one.

AdPilot is new - and we’d rather earn your trust than fake it.

We could plaster this page with five-star quotes and a “trusted by 10,000 businesses” banner. We won’t - because AdPilot is new, and inventing proof would be lying to you. Instead, here’s what actually protects you:

• You risk nothing to try it. 7 days free, no credit card. If it isn’t useful, you walk away having spent nothing.
• Your account is never at risk. Nothing is ever applied without your one-tap approval, a snapshot is saved before every edit, every change has one-click revert, and a full audit log records it. The worst case is a suggestion you don’t like - so you just don’t approve it.
• Guardrails you can’t trip by accident. AdPilot never deletes campaigns and never touches shared budgets. Some doors simply don’t exist.
• Built by an operator, not a VC-funded black box. Software made to give small businesses the Google Ads work an agency charges a retainer for - without agency prices or the loss of control.

Judge us on the work, the math, and the safeguards - on your own numbers, not our adjectives.