Ryze AI Pricing in 2026: The Costs Beyond the Tier Table
Ryze AI’s monthly fee is only part of the cost. Review contract terms, implementation time, landing-page gaps, and a two-week test to run before you renew.
August 2, 2026
•
min read
.png)
Most negative keyword advice gives you a list: free, cheap, jobs, salary, DIY, Reddit, plus 200 more. That list catches obvious waste in week one. By month four, the queries draining your budget are usually specific to your offer: the wrong product size, the wrong city, a competitor model number that resembles yours, or someone looking for a repair manual for the product you sell new.
The last account I inherited from an agency had 3,100 unique search terms in the previous 30 days and 14 negative keywords across the account. Eleven had been added during onboarding two years earlier. Nobody was lazy. The account manager had 19 other clients and a monthly reporting deck to build. Search terms arrive every day. Humans review them when the calendar allows it. That mismatch is the problem. No keyword list fixes it.
Google's help documentation on negative keywords and the search terms report is accurate and short. It tells you what the buttons do. It does not tell you which of 3,100 terms deserves 40 seconds of attention, whether a negative belongs at ad group or account level, or whether an apparently irrelevant query is simply early-funnel. That judgment is the work. Here is the workflow I use, the thresholds behind it, the mistakes that quietly throttle volume, and the point where manual review stops scaling.
Positive keywords expand. Negative keywords do not. Operators often assume the systems mirror each other. They do not.
A positive broad match keyword for running shoes can match "trainers for jogging," "sneakers for marathon training," and misspellings. Add running shoes as a broad negative, and Google blocks queries containing those two words. Nothing else. No synonyms. No invented plurals. No close variants.
Say you sell running shoes and never want to pay for bargain hunters. Add the negative cheap running shoes in each match type:
cheap running shoes): Blocks searches containing all three words in any order. "Cheap running shoes for men" is blocked. "Running shoes cheap women's" is blocked. "Cheap shoes" is not, because one word is missing."cheap running shoes"): Blocks searches containing that exact word sequence. "Best cheap running shoes 2026" is blocked. "Cheap mens running shoes" is not, because a word appears in the middle.[cheap running shoes]): Blocks only that query, word for word. Everything else can run.My default is phrase match for concepts and exact match for individual queries I want gone without collateral damage. I use broad negatives sparingly, and only when the word combination is unambiguous.
Two limits matter before you start pasting: a negative keyword can contain up to 10 words and 80 characters, and case is ignored. You do not need both Cheap and cheap. You do need every spelling variant you care about. Google can match a positive keyword to a typoed query, but it will not block that typo with a correctly spelled negative. If runing shoes cheap costs you money, add the misspelling as its own negative.
Negatives beat positives. If a search query matches both a keyword you bid on and a negative in that campaign or ad group, your ad does not serve. There is no scoring or tiebreaker. Exact match does not win.
I watched an account manager add installation as a campaign-level broad negative to filter DIY researchers. Then they spent three weeks wondering why the ad group targeting [commercial hvac installation cost] had stopped getting impressions. It had not stopped. One word had switched it off.
Single-word broad negatives cause this most often:
Free blocks "free shipping," which can be a buying signal.Used blocks "used by dentists."Cost blocks "cost of hiring a plumber," a query I would pay a premium for.Before adding a one-word negative, filter your search terms report by that word and read the queries it would have blocked over the last 90 days. If any converted, you have your answer: use a phrase negative instead.
Relevance decides first. Spend decides second.
Read the query and ask one question: could the person who typed this buy what I sell? Not, "Is this my keyword?" Not, "Does this look tidy?" Could they buy?
That creates three buckets:
The numeric trigger is for the ambiguous middle, where you genuinely cannot tell whether a query belongs. I use 1.5 times target CPA with zero conversions. With a $120 target CPA, a term that spends $180 without converting gets a decision. With a $30 target CPA, that number is $45.
The usual advice, "three clicks and no conversion," works for a $4-CPC lead-gen account. It is actively destructive for anything with a long consideration cycle. Scale the trigger to your economics and write it down. Otherwise, you will relitigate it every week at 11pm.
One rule I keep without exception: never negate a search term that has converted, no matter how bad the CPA looks. A term at $340 against a $120 target is a bidding problem or a landing-page problem. Negating it throws away demonstrated intent to fix a bid. Pull it into its own ad group, point it at a page that matches the query, cap the bid, and give it another 30 days.
I have recovered more revenue from that move than from any negative list I have built.
Sorting by cost finds expensive queries. Most wasted spend is not expensive per query. It is distributed: 60 searches that each cost $9 and share one word. None will appear near the top of your report. Together, they cost $540 a month.
Stop reading queries one by one. Start counting words.
Export 90 days of search terms with cost and conversions. Split each query into individual words and two-word pairs, then sum cost and conversions by word. That is an n-gram analysis. It takes about 15 minutes in a spreadsheet with a split function and pivot table.
The first time I ran one properly was on a replacement-parts store. Nothing in the top-cost view looked wrong. The n-gram view showed diagram across 47 queries: roughly $600 in 90 days and one conversion. People wanted an exploded parts schematic, found the product page, and left. Adding diagram as a phrase negative took nine seconds.
The same pivot flagged how to replace, which I did not negate. Half of those people eventually needed the part. That traffic went to a separate campaign at one-third of the bid.
Read the n-gram table by cost per conversion, not cost alone. A word with $900 in spend and 12 conversions is a high-performing theme. Build ad groups around it. A word with $200 in spend and zero conversions across 30 distinct queries is a negative candidate. The 30 distinct queries make it safer: one zero-conversion query is noise; 30 sharing a word is a pattern.
Run this once a quarter, even if you skip everything else in this article.
Placement is simpler when you ask what mistake you are preventing.
running shoes and trail running shoes, add trail terms as negatives to the general ad group. That is traffic sculpting, not waste prevention.Shared negative keyword lists live under Tools in the Shared library. They are the highest-leverage 20 minutes in a new account because you build once and apply them to every campaign, including future campaigns.
You get up to 20 lists per account and 5,000 keywords per list. That is more room than any sane account needs. I run three:
Three lists you can explain out loud beat 11 lists nobody remembers applying. The failure mode is not overuse. It is building a beautiful list and forgetting to attach it to the campaign launched last Tuesday.
Account-level negative keywords are newer and deserve more caution. They apply everywhere, reach campaign types that never accepted campaign-level Search negatives, and cap at 1,000 keywords. That reach is the problem. An account-level negative is invisible when you troubleshoot a campaign three months later, because nothing in that campaign's keyword screen tells you it exists.
Use account-level negatives for only two categories:
Everything else belongs in a shared list, where you can see it and remove it.
Most negative keyword processes die because "review the search terms report" is not a task. It is a mood. It has no start, finish, or defined output, so it loses to anything that does.
Give it a fixed shape and a timer. This is the weekly workflow I used on every account I managed, on the same day and in the same order.
Cadence scales with spend, not ambition:
Add the quarterly n-gram pass to whichever cadence you choose. Put both on the calendar as appointments with an end time.
Negatives feel safe because every addition looks like saved money in the report. What the report does not show is the auctions you stopped entering.
With Smart Bidding, this matters more than it did in the manual era. The bidding model learns from conversions across the query space it can see. Every negative narrows that pool. Trim obvious waste and the model gets cleaner signal. Trim aggressively and you give it a smaller, thinner dataset. CPA drifts up while cost per click looks fine.
I learned this the expensive way. I spent an afternoon adding around 600 negatives to a $30k-a-month account after a large n-gram export. I felt extremely productive. Impressions fell roughly 40% over the next week, and conversions fell with them.
Some negatives were right. The problem was volume and timing: 600 changes at once, immediately after a restructure, with no way to isolate the cause. It took three weeks of adding terms back to find the eight that mattered. I used to tell clients aggressive sculpting was the mark of a well-run account. I was wrong. It is often the mark of someone optimizing for the appearance of work.
My rule since then: no single week's negatives should remove more than about 10% of a campaign's impressions. Stage larger cleanups across three weeks and check search impression share between stages.
Two smaller habits help:
Run the arithmetic on your account and the ceiling appears quickly. If campaigns generate 3,000 unique search terms a month and your weekly audit covers the top 150 by cost plus a low-CTR skim, you are making deliberate decisions on 20% of the query space. The rest rides.
Then add the lag. A query that starts spending on Thursday morning gets its first human review the following Tuesday. At $2,000 a day in spend, that is five days of budget flowing through a filter nobody has inspected.
The audit is not wrong. It is periodic. The auction is not.
Scripts close part of that gap. Google Ads automated rules, the no-code rules in the interface, cannot create negative keywords. They can pause keywords, adjust budgets, and send alerts. Negative creation is not available.
Scripts can do it. A standard script pulls the search terms report on a schedule, flags terms above a cost threshold with zero conversions, then either emails the list or writes negatives into a shared list. Run it daily, cap what it can add, and send yourself the log. That removes the worst offenders within 24 hours instead of seven days.
A script cannot decide relevance. Relevance was the first filter in this process for a reason.
A script sees that exploded parts diagram spent $47 without a conversion and applies your threshold. It does not know that the same searcher converts 8% of the time when they land on a page with the schematic and an add-to-cart button. It also cannot distinguish a genuinely new demand pattern from noise, which is when you want a positive keyword instead of a negative.
Threshold logic is a blunt instrument dressed as automation. It is useful and worth deploying. It is not judgment.
The disclosure first: I work at groas, so weigh this accordingly.
What pulled me away from managing accounts by hand was seeing the cadence problem solved structurally rather than through more discipline. The models evaluate each new search term against the specific offer as it appears, rather than against a cost threshold on a schedule. The relevance and spend questions get asked together instead of a week apart. Irrelevant queries get blocked before they accumulate a meaningful loss.
The part I did not expect to care about was the other side of the decision. When a query is relevant but the page does not answer it, the system builds a version of the landing page that does. The diagram problem gets a page instead of a negative. That is the call I used to make wrong in both directions.
The tradeoffs are real. You lose the tactile sense of the account that comes from reading queries yourself. If you like opening the search terms report on a Sunday, that loss is genuine. Your job changes from adding negatives to reading the weekly log of what was added and why, then pushing back when you disagree.
There is also a structural detail to know before evaluating it: below $25k a month in ad spend, groas runs fully managed with a dedicated account manager and no dashboard to log into. That is deliberate and will irritate some people. Above $25k, you can access the software and run campaigns yourself with a strategist behind you. Neither arrangement removes the need for someone to know why a term was blocked.
There is no number. Anyone who gives you one is guessing.
The limits are generous: 10,000 negatives per campaign and 5,000 per shared list. You will never approach them for good reasons. Watch the ratio of negatives to active keywords instead. An account with 40 keywords and 3,000 negatives is not well tended. It is a broad-match campaign held together by exclusions, and it would likely perform better rebuilt around tighter themes.
Watch direction, too. If you add negatives every week and never remove one, you are accumulating decisions made under conditions that no longer exist. Google does not help here because the interface never shows how much traffic a negative has blocked.
Once a year, read the list. Delete anything you cannot justify out loud. Then watch impressions for a fortnight.
Usually yes. Separately and deliberately, decide whether to bid on them.
Competitor queries leaking into a general campaign are often the worst traffic in the account: high CPC, low CTR, and a searcher who has already picked someone else. Blocking them at shared-list level costs nothing.
What I object to is adding competitor as a broad negative, catching competitor vs [your brand] comparison queries in the same net, and throwing away the highest-intent traffic on the internet. Use phrase negatives. Keep vs, alternative, and versus queries alive. If you want to conquest, build a campaign for it with its own budget and landing page.
Conquesting by accident is expensive. Conquesting as a decision is sometimes the cheapest lead source you have.
Go back to that inherited account: 14 negatives and 3,100 monthly search terms. The agency was not hiding anything, and nobody there was incompetent. The work arrived faster than one person could review it monthly. The gap between those rates is where the money went.
Pick your way to close it: the timed weekly checklist, a daily script with a spend cap, or a system that reads every query as it lands. What I would stop doing is paying someone a percentage of spend to review this monthly. That arrangement produced those 14 negatives in the first place.
Export your last 90 days of search terms this afternoon and run the n-gram pivot. Whatever appears in the second column will tell you which option you need.
