Most marketing teams now pay for AI twice. Once inside the platforms they already had, where AI features arrived as add-ons or tier upgrades, and again as standalone subscriptions for writing, images and video that nobody has reviewed since they were bought.

This is a guide to working out which is which, what a realistic AI budget looks like, and which parts are genuinely worth paying for separately.

Key takeaways

Start with an audit, not a purchase

Before adding anything, list what you already pay for and what AI it includes. Almost every major platform shipped AI features in the last two years, usually without a price change, and usually without anyone on the team noticing.

The specific question to ask of each tool: does this generate, or does it only assist? Rewriting a subject line is assistance. Producing a finished on-brand image is generation. Teams routinely buy a generation tool when their existing platform already covers the assistance they actually needed.

The four categories

The pattern: copy has commoditised, video and voice have not. If you are going to pay for one standalone AI tool, it is far more likely to be worth it in video than in writing.

What it actually costs

Standalone AI subscriptions are individually small and collectively significant. Realistic ranges for a small team, based on typical published pricing:

The three metering models

How an AI tool charges matters more than its headline price, because usage is far less predictable than seats.

Per seat

Predictable, and wasteful. You pay for the designer who generates forty images a week at the same rate as the account manager who generates two a month. Fine for small teams, expensive as headcount grows.

Credits or tokens

Usage tracks value, which is the fairest model, but it makes budgeting harder and creates a quiet incentive not to experiment. The thing to check is what happens when you run out: does work stop, or does it silently bill more?

Unlimited

Attractive and usually qualified. "Unlimited" typically means fair-use limits, slower queues at volume, or a lower-quality model. Read what unlimited actually covers before it becomes your planning assumption.

The cost nobody prices: getting assets back

Generation is the easy half. The asset then has to reach the campaign — the right size, on brand, in the email or ad or landing page it was made for.

When the AI tool is separate from the marketing platform, that is a human moving files between applications, checking brand consistency by eye, and re-exporting at different dimensions. It does not appear on any invoice, and for teams shipping campaigns weekly it is frequently the largest real cost in the whole stack.

It is also where brand consistency degrades. Four people generating images in a separate tool produce four interpretations of the brand, and nobody catches it until a client does.

How to decide what to keep

• List every AI subscription and its monthly cost. Use the card statement, not memory — there is almost always one nobody remembered.
• Check what your existing platforms already include. Cancel anything duplicated by a tool you already pay for.
• Identify the one category you genuinely need depth in. Usually video or images. Keep the specialist there.
• Count the file-shuffling. How many times a week does someone move an asset between tools? That is the consolidation argument.
• Check brand consistency. If generated assets do not look like each other, a stored brand kit is worth more than a better model.
• Model your volume at 3×. Credit-based tools that look cheap today are the ones that surprise you.

Questions to ask before you commit

Whichever way you are leaning, these are the questions that change the answer. Work through them against your own numbers rather than anyone's feature matrix — including this one.

• What does my current platform already generate? Check before buying anything.
• Does this generate or only assist? They are different purchases.
• How is it metered — seats, credits, or "unlimited" with conditions?
• What happens when I hit the limit? Work stops, or the bill grows?
• How do assets reach the campaign? Count the manual steps.
• Can it apply our brand automatically, or does consistency depend on whoever is prompting?

If you cannot answer the first three, you are not ready to choose between Scalry and an AI tool yet — you are still deciding what you need, which is a cheaper problem to solve first.

A worked example: what one campaign costs to produce

Take a single product launch: one landing page, four ad variants with images, one 30-second video, five emails and eight social posts. Modest by most standards.

On a separate AI stack

Copy comes from the writing tool. Images from the design tool, exported at three aspect ratios by hand. The video from a third tool, with the voiceover generated in a fourth and mixed manually. Each asset is then uploaded into the marketing platform, resized where it does not fit, and checked against the brand guidelines by whoever remembers them.

The subscriptions might total $120/month. The hidden cost is the afternoon spent moving files, and the fact that the four ad images do not quite look like each other.

On a bundled platform

The same assets are generated against a stored brand kit inside the workspace that hosts the campaign, so they arrive already on brand and already in the right place. The cost is metered from one balance, which means you can see what the launch cost to produce — a number almost nobody can currently answer.

Which is right

If you ship one campaign a quarter, the separate stack is fine and the file-shuffling is trivial. If you ship weekly, or you are producing for several brands, the integration cost dominates everything else in this article.

Frequently asked questions

How much should a small team spend on AI marketing tools?

A typical standalone stack — writing, design, video and voice — runs roughly $60 to $240 a month for a small team. Before committing to that, audit what your existing platforms already include, because much of it is now bundled at no extra cost.

Is bundled AI good enough, or do I need specialist tools?

For copy, bundled AI is close to parity with standalone writing tools and rarely worth a separate subscription. For video and voice, specialists still lead, and few marketing platforms generate either natively. Images sit in between, depending on how much control you need.

What is the real cost of running AI tools separately?

The subscriptions are the visible part. The larger cost is usually the manual work of moving assets into campaigns at the right size and checking they are on brand, plus the brand drift that happens when several people generate independently.

Which AI metering model is best?

Credit-based metering is fairest, because cost tracks usage rather than headcount, but it needs modelling at about three times your current volume before you can trust the budget. Per-seat is predictable but wasteful when only some people generate. Treat "unlimited" as fair-use until you read the conditions.

Should I consolidate AI into my marketing platform?

If your bottleneck is producing campaign assets and getting them live, yes — consolidation removes both the duplicate subscriptions and the file-shuffling. If one AI category is strategically critical and you need a specialist's depth, keep that one and consolidate the rest around it.