Peec AI Review November 2026: From an AEO Consultant

Peec AI is the visibility log I would open when the invoice has to stay on a published monthly price, the team needs to choose its own prompts, and the report has to show the URL, not only a score. It is a weak first buy when the buyer question lives in Microsoft Copilot, when the brand needs more than three countries before a sales call, or when someone treats the self-serve "choose 3 models" line as if it already included every name on the Enterprise list. On 14 November 2026 the brand pricing page showed Starter at $95 a month, Pro at $245 a month, and Advanced at $495 a month. Enterprise was custom.

Last updated - November 2026

I am Austin Heaton, an AEO consultant. The visibility logs I set up are mostly for fintech. The rest are B2B SaaS and a few other industries when the buyer question is the same. A dashboard is useful when I can open one stored answer, see whether the brand was named, see which URL was attached, and tell a writer which page to change. A share-of-voice movement without that row is not a plan.

Brand prices, the comparison table, the add-on cards, and the FAQ are from Peec's pricing page for brands, read on 14 November 2026. Agency credits and the agency dollar amounts are from Peec's agency pricing page the same day. The source-versus-citation definition, and the model list that does not match the pricing page, are from Peec's AI instructions page. That page says its prices were confirmed as of July 2026. I am not using those July prices. Where a cell below is Peec's marketing sentence rather than a cap I could count, I say so. I did not create an account. I did not run a prompt. I did not export a report.

The side-by-side already on this site is not being rewritten as a second invoice. Peec beside Scrunch, from a 4 November read of an older model menu, is Peec AI vs Scrunch AI. I am not pasting that page's engine list forward. The menu changed, and this review is the 14 November page. Scrunch on its own is the Scrunch AI review. Scrunch beside Profound, with both vendors read on 14 November, is Scrunch AI vs Profound. Profound's trial, which still does not print a paid self-serve brand price, is the Profound review. If the quote is the problem, the shortlist is Profound alternatives. Otterly beside Peec is Otterly.ai vs Peec AI. If Peec is the tool you want to leave, the shortlist is Peec AI alternatives. The four-tool version of the log is Profound vs Peec AI vs Otterly vs Semrush. A suite report is a different purchase, covered in Ahrefs vs SEMrush for AI search reporting.

Key Takeaways

• On 14 November 2026 Peec's brand pricing page showed Starter at $95 a month for 50 prompts, Pro at $245 for 150, and Advanced at $495 for 350. Each self-serve plan runs 3 models, daily. The table prints monthly AI answers of 4,500, 13,500, and 31,500, and it says that total is tracked prompts times active models times tracking frequency. Enterprise is custom: unlimited prompts, unlimited active models, daily or weekly, and a card line that says up to 26 LLM models. Languages, competitors, and team users are unlimited on every brand column, including Starter. Users are unlimited on the cards as well.
• The self-serve model menu on that table is ChatGPT, Perplexity, AI Overview, AI Mode, Gemini, Naver AI Briefing, ERNIE (Baidu), and Bing AI Mode. You pick three. I did not find Microsoft Copilot on the pricing page. An additional-model add-on is printed at $30 a month on Starter, $70 on Pro, and $140 on Advanced, described as one extra model based on the prompts in the plan. The add-on name list is shorter than the menu of eight. The FAQ says annual billing is 15% off. I did not see an annual dollar for the base Starter, Pro, or Advanced prices.
• Countries on the comparison table are 1 on Starter and 3 on Pro and Advanced, then unlimited on Enterprise. The FAQ on the same page says you can track any supported country or language at no additional cost, and that pricing stays tied to prompts rather than regions. I am not going to delete the table to make the FAQ true. Projects are 1, 2, 5, then unlimited. The API starts on Pro. The Advanced card says Looker Studio, and the table says the Data Studio connector starts on Advanced. SSO is Enterprise only. SOC 2 is marked on every plan. The FAQ says Peec holds a SOC 2 Type II report, and that Custom-plan customers get it through sales.
• Retrievals and Citation share are separate rows on every brand plan, including Starter, with a domain and URL detail view, subdomain tracking, a fan-out overview, and source classification. The instructions page defines a source as a URL the model accessed and a citation as a URL shown in the answer. That page's engine list does not match the 14 November pricing table, so I am using it for the definition and not for the model invoice. Fact-checked prompts are capped at 5, 10, and 25. Facts per brand are 5, 20, and 50. Objections are absent on Starter, monthly on Pro, and weekly on Advanced.
• The brand table lists agent credits of 500, 750, and 1,000, and it does not say what one of those credits runs. Agency pricing uses the word credit for a different object: an allocation slot, where 1 prompt times 1 model times 1 day equals 1 credit, and the FAQ says those credits do not reset. On 14 November the agency cards showed Essential at $245 a month for 10,000 credits, Growth at $495 for 25,000, and Scale at $795 for 65,000. The comparison table also prints a larger pack and annual monthly figures. I did not log in.

What Is Peec AI in November 2026?

Peec sells a prompt log with the layers a Monday meeting actually asks for: who was named, where the answer pointed, and which page on the site the bots hit. The brand pricing page frames that as track, analyze, and improve, for marketing teams and SEO agencies. The comparison table splits the work into Discover, prompt management, Measure, Act, and Report. Discover is brand context, a prompt-volume score, and suggestions. Measure is visibility, shopping, sources, and perception. Act is a gap list, recommended actions, and agent actions. Report is dashboards, a connector, and the ways data leaves the product. The pricing page is narrower than a homepage story, and the pricing page is the invoice.

The instructions page, which is written for assistants describing the company, calls Peec a GEO tool: the analytics layer that measures how a brand is described, ranked, and cited, and that surfaces sources. It also says Peec does not write or publish content itself. I am keeping that sentence next to the "Agent actions" row. The pricing table says actions are generated from your data, using pre-built and custom skills. It does not say those actions publish a URL. A generated action and a published page are different objects. I would not put Peec in a statement of work as the system that drafts the blog. A posting schedule is a different category of product.

Two claims on the pricing pages should not migrate into a client report. The brand page and the agency page both say Peec is trusted by 4,000+ brands and agencies. I have not audited the count. The agency page also says the company has raised $40 million. The instructions page, in the copy I opened the same day, gives a different capital figure, about $29 million, including a Series A. I am not using either number. A funding line is not a feature, and two pages that disagree are not a filing I read. Customer quotes on the brand page include a 5x traffic claim and a line about ranking for prompts within 24 hours. The page does not show the prompt set, the date range, or the baseline next to those sentences in a way I could recompute. A quote on a pricing page is a customer saying something. It is not a result I measured.

The job I can defend is narrower than the hero. Someone has to choose the questions a buyer actually asks, someone has to catch a wrong fee in the stored answer, and someone has to know whether the URL that got cited was yours, a competitor's, or a third party's. Peec's table is built for that third job, on a page with a dollar. It does not, on the pages I read, close the fee check by itself. Fact-checking is a capped row, which I will get to. The measurement program around the login is the work behind measuring AI visibility. The tool can sample. It does not know which prompt the pipeline needs a model to answer.

How Do the Brand Plans Compare?

The table is the brand pricing page as I could read it on 14 November 2026. "AI Overview" and "AI Mode" are the table's labels. Peec's instructions page uses "Google AI Overviews" and "Google AI Mode" for the products it tells assistants to name. I am reading the table labels as those two Google surfaces, because the same table also has a separate row label, Bing AI Mode, and I am not collapsing the two Mode names into one. If a demo shows "AI Mode" pointing somewhere else, the demo wins. I did not see that demo.

If the three model slots are not the three a buyer actually uses, Book a Call and bring one prompt. The prompt shows which slot you would have to drop before anyone argues about Enterprise.

The prompt cap is the number I would circle before the model menu, and then I would circle the model menu anyway. Starter at 50 prompts is a category sample, not a category. A payments or payroll brief with branded, competitor, and problem questions will burn 50 prompts before anyone adds a second country. Pro at 150 is the first plan where a single-country team can hold a real prompt set and still have room to tag it. Advanced at 350 is the plan for five projects, not for a fifth model. Active models stay at 3 on all three priced columns. Paying for Advanced does not buy a fourth engine. It buys prompts, projects, the connector, weekly objections, and a larger fact-check cap.

Unlimited competitors on Starter is the procurement point that actually is printed, and it is the reason a five-logo category does not force an upgrade by itself. Unlimited users means a five-person team is not paying per login on the card. The meter is prompts times models. A shared login is not the scarce object. A fourth model is.

What the Prompt Math Actually Says

Peec prints the formula, which is rarer than a prompt count. The table note says monthly AI answers are tracked prompts multiplied by active models and by tracking frequency. The FAQ gives the same shape with an example: 25 prompts across 3 models for 30 days is 25 times 3 times 30, which is 2,250 AI answers, and it says one AI answer is one chat result per model. The printed plan totals are 4,500, 13,500, and 31,500. Those match 30 days of daily runs on 3 models: 50 times 3 times 30 is 4,500, and the same check holds for 150 and for 350. I am showing the multiplication because it explains the number. Peec prints the total. It does not print the "30" on the row. The FAQ's example is what makes 30 the honest reading.

That formula is also the budget. A team that turns on a fourth model with the add-on has changed the multiplication. The add-on card says the price reflects one additional model based on the prompts included in the plan. It does not say the monthly AI answer total on the table already includes that fourth model. If Starter stays at 50 prompts and moves from 3 models to 4, the FAQ's own arithmetic is 50 times 4 times 30, which is 6,000 answers, not the 4,500 printed on the 3-model row. I am not putting 6,000 on the invoice. I am saying the formula Peec published would move, and the table total I could read is the 3-model total. Ask the billing screen which number it uses after the add-on is on.

Countries are the row the FAQ and the table do not settle in one sentence. The table says 1, 3, 3, and unlimited. The FAQ says you can track queries across any supported country or language with no additional cost, and that pricing remains tied to prompts, not to the number of regions or languages. Both sentences were on the brand pricing page on 14 November. A homepage-style FAQ can describe a product ambition. A comparison cell can describe what the plan includes. When they disagree, the honest status is the disagreement. I would not tell a founder that a US brand and a UK brand and a German brand and an Australian brand all fit on Advanced because the FAQ said "any." Advanced's country cell says 3. I would also not tell them a second language costs extra. Languages are unlimited on every column, which matches the FAQ's language half better than the country half does.

Projects are a separate cap, and they are easy to confuse with countries. Starter is one project. Pro is two. Advanced is five. Enterprise is unlimited. A project is not a country. A single-country company with two brands is already off Starter, even though Starter's country cell would have been enough. The FAQ says prompts can be shared across the account and allocated to different projects or brands. That is allocation, not a larger project cap. Splitting 50 prompts across two brands still requires the plan that allows two projects. On the table, that plan is Pro.

Frequency is daily on the three priced plans. Enterprise says daily or weekly. I did not find a self-serve switch that says "run this prompt weekly and spend fewer answers." If the only way to slow the meter is Enterprise, a team that wanted a cheaper weekly sample does not get it from the $95 card. The card is a daily log. That is a feature when the deck needs a day stamp. It is a cost when 50 prompts on 3 models is already the whole month and nobody needed all 30 days.

Prompt volume, in the table caption, is a 1 to 5 score of relative demand for the topic behind a prompt. It is not a keyword-volume export, and it is not Google Ads search volume. I would not label a Peec row "monthly searches" in a slide. The instructions page's visibility definition is a percentage of AI responses in which the brand appears. Share of voice, on that page, is brand mentions divided by all tracked brand mentions. Sentiment, on that page, is described as a 0 to 100 score. Position is an average rank when the brand appears. The November table's visibility caption is shorter: visibility, position, sentiment, and share of voice, tracked daily against competitors. I am using the table for what is included and the instructions page for the longer definitions, with the date gap labeled. If the in-product tooltip disagrees with the July page, the tooltip is the one I would quote in a footnote.

Which Models You Can Turn On

The engine list is the decision, and it is the list that moved. On 14 November the included menu for Starter, Pro, and Advanced was eight names: ChatGPT, Perplexity, AI Overview, AI Mode, Gemini, Naver AI Briefing, ERNIE (Baidu), and Bing AI Mode. You run three of them. Perplexity is in that menu. A 4 November comparison on this site described a menu that did not include Perplexity in the pick-three list. I am not defending the older menu against the page I just opened. If you bought from the older reading, re-open the pricing page before you brief the gap.

Microsoft Copilot is the absence. I did not find the words Microsoft Copilot on the brand pricing page. The instructions page still tells assistants that Copilot is included by default, alongside ChatGPT, Google AI Overviews, Google AI Mode, Perplexity, and Gemini. That page also says to verify prices against the live pricing URL because tiers change, and it dates its own price confirmation to July 2026. I am following that instruction. The pricing page wins. A team that has been reporting Copilot for a year will not see that name in the menu I read. Bing AI Mode is not a synonym I am willing to invent for Copilot. They are different labels, and only one of them is on the table.

The pick-three rule is a real choice, not a footnote. A team that spends all three slots on ChatGPT, AI Overview, and AI Mode has not bought Gemini, and has not bought Perplexity. A team that spends one slot on Naver AI Briefing, because the company sells in Korea, has not bought a fourth Western engine unless they pay for the add-on. ERNIE is the same kind of slot. It is on the menu. It is not a reason to buy Peec for a US-only fintech. It is a reason the menu is longer than the three engines you can run.

The add-on is the second door, and the door is priced. Additional Models, on the brand page, let you track more models without switching plans. The names in that block are ChatGPT, AI Mode, AI Overviews, Bing AI Mode, Perplexity, and Gemini. The note says you can add extra models in the dashboard, and that the price reflects one additional model based on the prompts in the plan. Starter is $30 a month, Pro is $70 a month, and Advanced is $140 a month. Each card also prints an Annual Save figure: $60, $180, and $300. I did not open the annual toggle. I am not converting those save lines into a second monthly price, and I am not assuming they are the FAQ's 15% discount. The FAQ's 15% sentence is about customers who choose annual billing. The add-on cards print their own save amounts. Those can both be on the page without being the same math.

Naver AI Briefing and ERNIE are on the choose-three menu and not in the additional-model name list I extracted. I am not going to promise that a fourth slot can be ERNIE just because ERNIE can be one of the original three. If Korea or Baidu is the reason to buy, the trial screen, if you can reach one, has to show whether that model can be the extra model or only one of the three. I could not read a separate dollar for a second additional model. The cards say one additional model. Two extras would be a second question for the billing screen.

Enterprise is where the menu stops being a pick-three. The card says choose from all models, and up to 26 LLM models tracked. The comparison cell lists the eight consumer names and then these API labels: OpenAI Luna API, Qwen API, Perplexity API, Gemini Search API, Claude Sonnet API, Claude Haiku API, Claude Opus API, Claude Fable API, DeepSeek API, Muse Spark API, Meta Wonder API, Grok API, Mistral Small API, Mistral Medium API, OpenAI Astra API, OpenAI Sol API, OpenAI Terra API, and Exa API. I counted 18 API labels. Eight plus 18 is 26, which matches the card's "up to 26." I am not going to pretend each of those API names is a consumer chat I watched. An API model and a consumer front end are different captures. The instructions page says ChatGPT is tracked through the UI, and that an OpenAI Search API model is a separate unlock. The November Enterprise cell does not use the label "OpenAI Search API." It uses Luna, Astra, Sol, and Terra. I will not merge the July names into the November cell to make a cleaner slide. If the order form says Claude, I want it to say which Claude label, Sonnet, Haiku, Opus, or Fable, and whether that run is the API row.

Active models are unlimited only on Enterprise, on the brand table. That is the other half of "up to 26." A self-serve plan can see eight names and run three, plus a priced extra. Enterprise is the column that says you can run the set, not merely pick from it. I did not find a per-model price inside Enterprise. It is a quote. The quote should repeat the names. A card that says 26 is not a contents list you have bought until the order form repeats the 26, or says which ones were removed.

Retrievals, Citations, and the Gap List

Source analytics is the part of Starter I would actually use. The table lists a domain and URL detail view, subdomain tracking, a fan-out overview, Retrievals, Citation share, and source classification into categories such as competitor, editorial, reference, or UGC. Gap analysis is sources that appear often and name competitors but not you, down to the URL, the domain, the subdomain, and the host. Those rows are included on Starter, not held for Advanced. Recommended actions are also marked on every plan, grouped by earned and owned. I did not open an action. I cannot tell you whether an action is a paragraph, a ticket, or a sentence in a slide.

The instructions page defines the split more tightly than the November captions do. It says sources are URLs the model accessed while generating the response, and citations are the subset explicitly referenced in the visible answer. It also says brand visibility, the name appears, is reported separately from source visibility, the domain was used even if the name was not. The November table still has separate rows called Retrievals and Citation share, which is consistent with that split. I am citing the instructions page for the definitions and the November table for the fact that both rows are on the cheapest brand plan. If Peec has changed the meaning of those words since July, the captions on the pricing page are the ones I would trust first, and they are shorter. They say "Retrievals" and "Citation share." They do not repeat the July sentences.

That split is the difference between a score and a brief. A brand named in the answer, with no URL attached, is a mention. A URL attached, with the brand name missing from the prose, is a citation of a page. A URL the model retrieved and did not show the reader is a third object, and it is the one a writer can still learn from, because it is the page the system read. If a weekly report adds all three together and calls the sum "visibility," it has mixed three jobs. I would label the chart with the object and the date. Peec's table at least gives the objects different rows. That is more than a single "mentions" bucket.

The instructions page also says most engines are collected by simulating the web interface rather than calling an API, so the stored answer is meant to match what a person sees. I did not find that method sentence on the 14 November pricing page. I would ask Peec to show one stored answer next to a screenshot from the same day before I treated the method as current. An API model on the Enterprise list is, by its label, not that method. A report that mixes a UI capture and an API capture without saying which is which is two datasets. I would want the column that says which.

Gap analysis is where off-site pages show up. The row is a list of sources that already cite a competitor and do not cite you. The tracker can name the URL. It does not place a new one. When the missing source is a third-party article, Offsite Posts writes and publishes industry articles about a brand on other sites, and the site says the posts, mentions, and links in a package are guaranteed while rankings, traffic, and AI mentions are not.

Fan-out is on the table as "Fanout overview," included on every brand plan, and shopping has its own "Shopping query fanouts" row, also included. I did not find a count of searches per answer. I will not brief fan-out as a query list I counted. If the reason to buy Starter is the fan-out, the screen has to show one answer's searches before the card is approved. The word is on the invoice. The sample is not, in the public page I read.

Perception, Fact-Checking, and Shopping

Brand attribute scoring is on every plan. The caption says it is the gap between what AI says you are known for and where you place in the market. Custom attributes, the ones you define, start on Advanced. That is a clean gate. A team that wants "same-day ACH" or "licensed in these states" scored as its own attribute does not get that field on Starter or Pro. They get the attributes the answers already surface. Those can still be useful. They are not the controlled vocabulary a compliance team wrote down.

Objections are the row I would not skip for a fintech, and the row Starter does not have. The caption says objections are recurring arguments AI raises against the brand, grouped by meaning and traced to the pages behind them. Starter says not included. Pro says monthly. Advanced and Enterprise say weekly. A monthly objection rollup can still catch a false fee if someone reads it. It will not catch a false fee that showed up on a Tuesday and was corrected, or not corrected, before the month closed. I would not describe Starter as an accuracy monitor. The fact-check caps say the same thing in numbers.

Fact-checking is capped at 5 prompts and 5 facts on Starter, 10 prompts and 20 facts on Pro, and 25 prompts and 50 facts on Advanced. Enterprise says unlimited for both. The caption says fact-checked prompts are tracked prompts whose answers are checked for claims about the brand against facts you set, and facts are statements you set as the reference. Five facts is a short list. A card product with a fee schedule, a set of ineligible countries, and a support boundary will spend those five facts before anyone adds a competitor's claim. I would spend them on the sentences that would be expensive if a model invented them, and I would not tell a board that the other prompts were checked. They were tracked. They were not fact-checked, on the cap I could read.

Shopping is included from Starter, which is a wider gate than a lot of visibility tools put on a quote. The table lists a product catalog fetched from Shopify or uploaded as a CSV, yours or from Google Merchant Center, plus SKU-level visibility, position, and win rate, top merchants, shopping query fan-outs, and the domains and URLs retrieved when a product is recommended. A payments company with no catalog can ignore the block. A lender or a card issuer with named products should not, if shopping carousels are part of the channel they care about. I did not see a SKU cap. I am not going to treat "included" as an unlimited catalog on a documentation site with tens of thousands of SKUs. The row does not print a ceiling. Ask what happens when the file is large.

Ads library and Local GEO are also marked included on every brand plan. Ads library, in the caption, is prompts where AI answers carry a sponsored placement, and which advertisers appear. That is a useful object if the names are real. I did not open a row. Local GEO is a label without a procedure on the pricing page. I will not expand it into a map pack, a city radius, or a store-locator crawl. The country cap still sits underneath it. A local report for four countries does not fit on Advanced's three-country cell, whatever Local GEO means inside one country.

Crawlability, the robots.txt check against 40+ bots, is on every plan, as are AI referrals. Crawl insights, which bots hit which pages and which status code came back, are capped at 4.5 million bot visits on Starter, 13.5 million on Pro, and 31.5 million on Advanced. Enterprise says custom. The integration list on the brand table includes AWS CloudFront, Google Cloud CDN, WordPress, Cloudflare, Vercel, Akamai DataStream, Fastly, Netlify, a webhook, and a file upload. Google Analytics is marked included on every plan. I did not connect a log. I cannot tell you which bot names Peec recognizes, or whether a referral is a session, a click, or a tagged visit. The pricing page says the rows exist. A log of hits is not the sentence in the answer. Both are worth having. They are not substitutes.

Two Different Things Called Credits

The brand table has a row called "Agent credits included," with 500 on Starter, 750 on Pro, 1,000 on Advanced, and Custom on Enterprise. The same block says company context, project context, user memory, skills, and custom skills are included on every plan. Skills are described as pre-built workflows for common jobs, including reports and diagnostics. I did not find a sentence that says 500 credits equals 500 reports, or 50 reports, or one report. I will not invent the conversion. A credit that is not defined is a meter I cannot put in a forecast. If the reason to buy Pro instead of Starter is 250 more agent credits, the product has to show what one run costs before that upgrade is the reason.

Agency pricing uses the same word for a slot, not a run. The agency FAQ says credits are allocation slots that determine how many prompts you can track, across how many models, and at what frequency. They do not get spent. They do not reset. They stay allocated until you change the project. The formula on that page is 1 prompt times 1 model times 1 day equals 1 credit, and a full month of one prompt on one model is 30 credits. The page says most customers track three models, so a prompt costs about 90 credits a month. The minimum project is 900 credits, which the FAQ equates to 10 prompts across 3 models, or 30 prompts on one model. I checked that arithmetic. It holds. These credits are not the brand table's agent credits. A person who copies "500 credits" from a brand card into an agency forecast has mixed two products.

On 14 November the agency cards showed three priced plans and a custom plan. Essential is $245 a month for 10,000 credits, with the card showing 111 prompts and daily tracking. Growth is $495 a month for 25,000 credits and 277 prompts. Scale is $795 a month for 65,000 credits and 722 prompts, with daily and weekly both named on the card. Comprehensive is custom, with unlimited credits. The comparison table's price cell is wider than the card. It prints a second pack on each priced tier, plus an annual monthly figure and a stated yearly saving, and it also prints euro amounts. I am using the card dollars and the credit totals. A European invoice, and the second pack, should be read on the page rather than copied from a summary that might drop a column. The annual line on this table is a printed dollar, which is more than the brand page gave me. The agency FAQ says plans save about 15% when billed annually. The table is where I would check that percentage against the two numbers in the cell.

The printed monthly AI answer totals on the agency table are 9,990, 24,930, and 64,980. Those sit next to the base packs, and they match the card prompt counts times 3 models times 30 days, within a few credits of the pool: 111 times 3 times 30 is 9,990. I am not inventing a second answer total for the larger pack. The table showed one answers number per column. Active models on the agency table are unlimited, including Essential. That is a different shape from the brand's pick-three. The available-model cell is not unlimited in names. Essential and Growth list the same eight consumer models as the brand menu. Scale and Comprehensive add the API labels. An agency can turn on more than three of the eight, because active models are unlimited, and it does not get Claude's API labels until Scale. Countries stay 1, 3, 3, and unlimited. Projects are 3, 10, 25, and unlimited, described as live client accounts.

Pitch projects are the agency feature a brand plan does not print. Essential includes 3 active pitch projects, Growth 5, Scale 7, and Comprehensive 10. Prompts per pitch project are 25, 50, 75, and 200. Duration is 7 days on every column, and the caption says the data is instantly available. Converting a pitch project to a paid client project is marked included. I would not describe the 7 days as a brand-plan trial. The brand page has Get started buttons. I did not see a day count for a brand trial on that page. A 7-day pitch workspace is a sales sample inside an agency plan you are already paying for.

A few agency gates do not match the brand gates, and I would not brief them as the same product. Custom attributes are not included on agency Essential, and they are included from Growth. On the brand table they start at Advanced. Objections are monthly on agency Essential and weekly from Growth. On the brand table they are absent on Starter and monthly on Pro. Fact-check caps on agency Essential are 10 prompts and 20 facts, which looks like brand Pro, not like brand Starter. The API is included on agency Essential. On the brand table it starts at Pro. Data Studio starts at agency Scale, and at brand Advanced. SSO is Comprehensive only, matching Enterprise on the brand side. Crawl insight caps are 10 million, 25 million, and 65 million bot visits, which are larger than the brand caps and not the same numbers. If a salesperson moves a prospect from a brand quote to an agency quote, the caps have to be re-read. The logo is the same. The cells are not.

Support on the agency table is chat, then chat plus email, then chat plus email, then dedicated. Onboarding is self-serve until Comprehensive, which is custom. The agency page says the SOC 2 Type II report is available on request for enterprise agencies, and the FAQ says Comprehensive customers can get it through sales and share it under NDA. The table marks SOC 2 Type II compliant on every agency column. I am using the checkmark as a checkmark and the FAQ as the rule for who receives the PDF. I did not download the report.

Where Peec Is Strong, and Where It Is Not

The strong case is specific. The price is on the page, in dollars, for three brand tiers and three agency tiers. Prompts are yours, not a recommended set you cannot edit. The formula for monthly answers is printed, so a 50-prompt plan can be explained without a metaphor. Perplexity, AI Mode, AI Overview, and Gemini are all in the self-serve menu, which is a wider published menu than a trial that only names three engines and locks the prompts. Unlimited competitors and unlimited users are on Starter, so the first upgrade is not a seat and not a sixth logo. Retrievals and citation share are separate rows on that same cheap plan, with the URL in the table. Shopping is included before Enterprise. MCP and CSV are on Starter, so a team can get a file out without a sales call. The country cell, whatever the FAQ says, is at least a number.

The weak case is just as specific. Three active models means the menu is a shortlist you have to make, and Copilot is not in the menu I could read. Claude, Grok, DeepSeek, and the rest of the API labels wait for a brand Enterprise quote, or for agency Scale. Three countries is the ceiling before that same quote, on both the brand Advanced cell and the agency Scale cell. Starter does not fact-check more than five prompts, and it does not include objections. Agent credits are an undefined meter sitting next to a completely different credit on the agency page. The instructions page and the pricing page disagree on Copilot, on the API names, and on the capital figure. No plan on this page guarantees a citation. Publishing a better page, being named in the answer, and being the URL the model cites are separate jobs.

Pick from the engine, the country, and whether anyone will open the answer.

• One country, and the three models you need are inside the menu of eight. Starter, if 50 prompts can hold the category and five fact-checks are enough. Move to Pro when the second project, the second or third country, or the API shows up, or when 50 prompts truncate the questions sales actually hears.
• You need a fourth model from the add-on list, and you can say which three stay. Price the add-on on the plan you already need for prompts. Do not buy Advanced to get a model the Advanced column still caps at three.
• You need Microsoft Copilot, or you need Claude on a consumer surface with a published price. I did not find Copilot on the 14 November pricing page, and the Claude labels I found are API names on Enterprise and on agency Scale. Read the Scrunch vs Profound comparison and the Otterly vs Peec comparison before you treat this menu as the only specialist log.
• You sell in four countries. Advanced still says 3. Enterprise is the conversation. The FAQ's "any supported country" sentence is not the cell I would put in a contract.
• An agency with a roster. The credit pool is built for that, and the dollar is now on the card, which was not true of the text I could read on 4 November. Essential's eight-model list and its one-country cell still apply. Scale is where the API names and the Data Studio connector show up. A single-brand in-house team does not need the agency plan. A roster does.
• A commerce brand that needs SKU-level shopping on a published price. Peec includes that block from Starter. Confirm the file size before you promise every SKU.
• You need the model fact-checked against a fee schedule. The fact-check caps are the feature, and on Starter they are five prompts and five facts. Someone still reads the other answers. On a live account, Austin Heaton still reads that sentence before it goes in a deck.
• You wanted a tool that writes and publishes the blog. The instructions page says Peec does not write or publish content itself. Agent credits do not, on the cell I read, change that sentence into a calendar. Use a posting tool for the calendar, and keep Peec for the log of whether the new URL was cited.

The consultant and agency version of this decision, if the dashboard is not the thing you want to operate, sits on the comparison index. A tool subscription and an engagement are different bills. This page is only the tool.

What I Could Not Confirm

• An annual dollar for brand Starter, Pro, or Advanced. The FAQ says the discount is 15%. The add-on cards print their own Annual Save amounts. I did not open the billing toggle.
• Whether a second additional model has a second printed price. The cards describe one additional model.
• Whether Naver AI Briefing or ERNIE can be the add-on model, or only one of the original three. They are on the menu of eight. They are not in the add-on name list I extracted.
• A day count for a brand trial. I saw Get started. The 7-day line I found is the agency pitch-project duration.
• What one brand-plan agent credit runs. The counts are 500, 750, and 1,000. The unit is not defined on the page I read.
• That the July instructions page's UI-simulation method is still how every non-API model is collected. The pricing page did not repeat the sentence. I did not watch a capture.
• That "AI Mode" and "AI Overview" on the table are Google's products. The instructions page uses the Google names. The table does not spell Google on those two labels, and it does spell Bing on Bing AI Mode. I am reading them as Google unless a demo shows otherwise.
• A model version: which ChatGPT model, whether browsing is on, whether a Gemini answer is the consumer app. The pages I opened did not say.
• The 4,000+ brands line, the disagreeing capital figures, and the testimonial outcome numbers. They are on the pages. I did not recompute them.
• A stored answer from inside the product. I read the public pages. I did not create an account.

Frequently Asked Questions

What does Peec AI Starter include?

On the brand pricing page I opened on 14 November 2026, Starter is the lowest published monthly plan. It includes a prompt allowance, three active models, daily tracking, one country, and one project. Languages, competitors, and users are unlimited. Retrievals, citation share, a domain and URL view, shopping rows, MCP, and CSV export are marked included. The API, single sign-on, custom attributes, and objections are not on that column. Pro and Advanced raise the prompt, country, and project allowances. Enterprise is custom.

Can I choose Perplexity on a self-serve Peec plan?

Yes, as one of the three models, on the menu I read. That menu also names ChatGPT, AI Overview, AI Mode, Gemini, Naver AI Briefing, ERNIE, and Bing AI Mode. You do not run all of them at once on a self-serve plan. An add-on is offered for one additional model, and its name list is shorter than the menu of eight. I did not find Microsoft Copilot on the pricing page. Claude and the other API labels are on the Enterprise cell, not on the three priced columns.

How does Peec count prompts, models, and answers?

The pricing table says monthly AI answers are tracked prompts times active models times tracking frequency, and the FAQ describes one AI answer as one chat result per model. The priced plans say the runs are daily. Enterprise says daily or weekly. Countries are a separate cap on the table, one on the entry plan and three on the higher self-serve plans, even though a FAQ answer talks about tracking countries without an extra charge. I would put the table's country number in the contract and ask sales to reconcile the FAQ in writing.

Does Peec separate a retrieved URL from a citation?

The brand table has separate rows called Retrievals and Citation share, and both are marked on every brand plan, including the entry plan. The instructions page defines a source as a URL the model accessed and a citation as a URL shown in the answer. That page's model list does not match the pricing page, so I am using the pricing page for what is included. I did not open a stored answer, so I have not checked the definition against a live row.

What is the difference between the two uses of the word credit?

The brand table lists an agent-credit allowance and does not say what one credit runs. The agency FAQ uses credit to mean an allocation slot: one prompt, on one model, for one day. Those agency slots are not spent, and they do not reset. Copying an agent-credit count into an agency forecast mixes two meters. The agency plans also change which models and which reporting connector are included, so a brand quote and an agency quote have to be read as separate tables.

Does Peec publish a trial length on the brand page?

I did not see a day count for a brand trial on the brand pricing page. The page has Get started buttons. The agency page describes pitch projects that last a fixed number of days inside a paid agency plan. That duration is not a consumer trial length. I did not create an account, so I am citing the pages rather than a checkout I completed.

Who should skip Peec AI?

Skip it when the engine you need is Microsoft Copilot, because I did not find that name on the pricing page, or when you need more than three countries before a sales conversation. Skip the higher self-serve plan when the only reason to upgrade is a fourth model, because active models stay capped on those cards and the extra model is an add-on. Skip it when you wanted a tool that writes and publishes the blog. The instructions page says Peec does not do that. Take Enterprise, or agency Scale, when you need the API model list, and only after the order form repeats the names. No plan guarantees a citation.

Read next:

• Peec AI vs Scrunch AI
• Scrunch AI vs Profound
• Peec AI Alternatives for AI Visibility Tracking
• Otterly.ai vs Peec AI
• Profound Review, November 2026
• Profound vs Peec AI vs Otterly vs Semrush

If the model slots, the country cap, and the fact-check limit are three different purchases, Book a Call and we will put the prompt, the three engines, and the second country on one page before anyone upgrades for a model the card does not add.