Best Black Friday AI Deals 2026: A Verified Tracker

What is confirmed today, and what is not

Black Friday 2026 falls on Friday, November 27. That date is known; most individual AI software offers are not. We are publishing this tracker early because the useful work begins before a discount appears: deciding which jobs need a tool, recording the normal price, testing the product, and setting a maximum budget. A countdown with a long list of predicted percentages would look busy, but it would not help you buy well.

At this review, this tracker does not label any offer as verified. That wording is deliberate. It does not mean no vendor anywhere has started a promotion. It means we have not entered an offer in our verified table until the evidence fields below are complete. When a vendor announces a promotion, we will link to the vendor page, record the exact plan and region, show the calculation, and date the check.

The distinction matters because an AI tool can advertise a large percentage while changing the comparison period, limiting the discounted credits, restricting the offer to new customers, or renewing at a much higher rate. A buyer who sees only the percentage cannot compare it with an ordinary annual plan. We treat the effective first-year cost and the expected renewal cost as separate numbers.

This is also an editorial page, not a merchant feed. Some links may become commissionable if a relevant partner has a real offer. A commission does not convert an unverified promotion into a recommendation. The selection test remains whether the tool solves a defined job at a defensible total cost.

The six facts every listed deal must show

A useful deal listing should let another person reproduce the conclusion. That is why every entry needs more than a tool name and coupon. We start with the exact product and plan. A discount on a limited creator plan is not evidence that a business plan or API package is discounted. We also record whether the offer applies to new customers, returning customers, upgrades, additional seats, or everyone.

Next comes the price baseline. The normal monthly price multiplied by twelve is not always the correct comparison if the vendor already sells an annual plan at a lower everyday rate. We compare the promotion with the normal way a sensible customer would buy the same entitlement. If the deal requires two years upfront, we do not describe the entire payment as a one-year saving.

Usage is the third fact. AI products often sell credits, minutes, generations, exports, seats, storage, or model access. A cheaper plan with half the usable allowance may be worse value even when the headline discount is larger. We record the unit that constrains normal work and note whether unused capacity expires.

Finally, we capture the end date, renewal behavior, refund rule, and source. Screenshots are useful evidence but can become stale, so the live vendor page or official announcement remains the primary reference. If checkout shows different terms from the landing page, checkout controls the buying decision and the discrepancy should be documented.

• Exact product, plan, billing period, currency, and customer eligibility.
• Everyday comparison price for the same entitlement, not an inflated anchor.
• Discounted checkout price and any tax, seat minimum, or required add-on.
• Included credits, limits, model access, export rights, and commercial-use conditions.
• Promotion start and end, cancellation path, refund policy, and renewal price.
• Official source URL, verification date, and a note when regional pricing differs.

A simple value score beats the biggest percentage

The biggest discount is rarely the best default. I score a candidate on job fit, expected usage, output quality, switching cost, and total cost. Price matters after the first four. A tool that saves ten hours in a real monthly workflow can be valuable at a modest discount. A tool bought for an imagined future workflow is expensive even at seventy percent off.

Start with one sentence: I need this product to perform a specific job for a named person or team at a stated frequency. Then estimate the number of outputs required during the first paid year. Divide the final checkout cost by that number. This produces a rough cost per useful output that can be compared across products with very different credit systems.

Add migration and review time. A new writing assistant may need brand examples, templates, approvals, and training. An AI video product may require repeated generations before one clip is usable. An agent platform may need engineering and monitoring. These costs are not reasons to reject the products; they are reasons to compare the real workflow instead of the sales page.

The final question is what happens after the sale year. If the renewal price would force an immediate migration, include the exit cost now. Export formats, ownership of generated assets, access to project history, and API portability can be more important than an extra ten percent reduction.

Build shortlists by job, not by AI category

Shopping by a broad label such as AI tools creates an endless list. A better shortlist begins with the work. A small business may need to publish product videos, answer support questions, schedule social posts, build a landing page, or research search demand. Those are five different buying decisions. A general assistant might cover three well enough, making three specialist subscriptions unnecessary.

For website building, judge the path from idea to a maintained site: responsive output, custom domains, source access, integrations, analytics, SEO controls, collaboration, and what happens when the generated application becomes more complex. A low introductory price is less valuable if the team cannot fix or move the product later.

For creative generation, test the unit economics. Count accepted videos, images, voice minutes, or exports rather than raw generations. Include reruns caused by model errors. Commercial rights, watermark rules, resolution, queue priority, and project storage can change the effective value dramatically.

For research, marketing, and SEO products, separate data access from generated commentary. Ask what proprietary or licensed data the tool supplies, how fresh it is, whether citations can be inspected, and whether exports fit the workflow. A polished summary of public information may overlap heavily with tools you already pay for.

• Assistant and research: compare source visibility, context limits, privacy controls, collaboration, and model access.
• Website and app building: compare ownership, hosting, export, integrations, responsive behavior, and maintenance.
• Video, image, and audio: compare accepted outputs per dollar, rights, watermarking, resolution, and iteration speed.
• Coding and agents: compare repository context, execution controls, review workflow, usage pricing, and observability.
• Marketing and SEO: compare underlying data, update frequency, export, team workflow, and measurable decisions supported.

Three budgets that prevent subscription sprawl

A budget is most useful when it limits the number of overlapping products. At fifty dollars a month, choose one general assistant and at most one specialist that directly produces revenue or saves recurring labor. Free plans can cover experimentation. The purpose of the paid plan is reliable capacity for an existing workflow, not access to every new model.

At one hundred dollars a month, a small team can usually support a general assistant, one production tool, and a small automation or research layer. The exact combination depends on the business. A consultant may value research and presentations; a retailer may value product imagery and customer support; a developer may value coding assistance and monitoring.

At two hundred and fifty dollars a month, governance becomes part of the purchase. Seat ownership, shared workspaces, permission levels, audit logs, client separation, approved data, and renewal responsibility matter. Buying several individual accounts because they are discounted can create a fragmented system that costs more to administer.

Reserve part of every budget for the full-price month after a trial or promotion. That test exposes whether the workflow survives without sale psychology. If a product is valuable only while the discount is visible, it probably has not earned a permanent place in the stack.

A calmer calendar from September through Cyber Monday

September is for inventory. Export the last three months of software charges, list who uses each product, and mark the workflow it supports. Record monthly and annual renewal dates. This prevents a Black Friday purchase from duplicating a subscription that quietly renewed in October.

October is for trials and baselines. Test candidate tools with the same representative task and keep the outputs. Record the normal public price and annual-plan price with a date. Decide the acceptable price before promotional email begins. A precommitted ceiling makes it easier to ignore bundles that expand the scope of the purchase.

Early November is for terms. Watch official vendor channels, not coupon aggregators alone. Read plan limits, eligibility, refund policy, tax treatment, and renewal terms. If a promotion requires an immediate annual payment, confirm that the trial covered the features and capacity you intend to use.

During Black Friday week, recheck checkout and save the evidence. After purchase, record the owner, cancellation date, renewal price, and success measure. Cyber Monday is not a reason to reverse a sound decision; it is simply another deadline vendors may use. The correct deadline is the one that leaves enough time to verify the purchase.

1. Inventory existing subscriptions, owners, renewal dates, and overlapping jobs.
2. Test candidates on representative work and retain the results for comparison.
3. Set a maximum first-year cost and an acceptable renewal price before sale week.
4. Verify the live offer against the same plan, limits, region, and billing period.
5. Record checkout evidence, cancellation instructions, and the person responsible for renewal.
6. Review actual use after 30 and 90 days, then cancel purchases that did not earn their place.

Red flags that remove an offer from consideration

A promotion should become easier to understand as you approach checkout. If the plan name, allowance, currency, or renewal price keeps changing, pause. Screenshots shared on social media can omit region, customer eligibility, expiration, or required add-ons. Start again from the vendor domain rather than following an unfamiliar checkout link.

Treat a lifetime label as a business-model question, not a guarantee. Ask whose lifetime, what service level is promised, whether future models are included, and how an expensive inference product can support indefinite use from a single payment. A lifetime offer can still be worthwhile, but only if the current utility justifies the price without relying on years of future service.

Bundles require similar discipline. Access to fifty tools is not equivalent to fifty maintained direct subscriptions. Check account sharing, authorization, data handling, update speed, commercial rights, support, and whether access violates a vendor's terms. Do not upload sensitive client or company material to an uncertain intermediary.

Finally, urgency is not verification. A timer that resets, a coupon that always works, or a price visible only after supplying payment details should lower confidence. Useful vendors can state what is sold, what it costs, and when the terms end.

• The discount percentage is visible but the normal comparable price is not.
• The landing page and checkout disagree about the plan or included capacity.
• A third-party domain requests credentials, card information, or account-sharing access.
• The renewal price, cancellation process, data export, or refund rule cannot be found.
• The offer claims unlimited AI use while the fair-use definition is absent or discretionary.
• Affiliate or sponsored relationships are hidden far from the recommendation.

How this tracker will change during the season

The page will keep one stable URL throughout the 2026 season. Confirmed offers will be added to the category table with a dated source. Material changes such as a price increase, extended deadline, exhausted allocation, or changed renewal term will be recorded rather than silently overwritten. Expired offers will be labeled expired and retained long enough to explain what happened.

We will prioritize products that fit this site's audience: general AI assistants, prompt and research tools, website and app builders, creative generation, marketing and SEO software, productivity, coding, and small-business workflows. A vendor can be included without an affiliate relationship, and an affiliate relationship does not guarantee inclusion.

The tracker will also link to deeper category and verification guides when a table cell cannot carry the full decision. A video tool may require a test of accepted output per credit. An annual assistant plan may require a renewal analysis. A public prompt library may require a security and quality review rather than a purchase decision.

Readers should still verify the final checkout. Prices and availability can change after our last review, and regional tax or currency can alter the amount paid. Our role is to make the comparison legible and show our evidence, not to replace the vendor's contract or your own budget approval.

Questions readers ask