Skip to content

Estimate the workload before choosing a chatbot plan.

A useful budget includes the website’s content, visitor conversations, and the people maintaining the assistant. Use the planner below to explore your own assumptions.

First-party product guidance. How we maintain these guides · Report a correction

Interactive worksheet

Plan your monthly usage.

Change the example assumptions to estimate page refreshes and conversations. This is a capacity worksheet, not a price quote or a prediction of visitor behaviour.

Across the websites in this estimate.
Assumes every included page is refreshed each time.
Use your own analytics or a planning estimate.
Your assumption, from 0 to 100. This is not a benchmark.
Planned page refreshes / month
200
Estimated conversations / month
200

Page refreshes = pages × refreshes. Conversations = visits × your percentage, rounded up. Initial indexing, partial or failed refreshes, repeat conversations, and provider costs can change actual usage. Compare website and per-site page limits separately.

Compare the current plan allowances

Start with the costs you can actually identify.

Website assistant costs can come from the service subscription, AI processing, content indexing, optional integrations, and team review. Which charges apply depends on the product and the agreement; a model’s token price alone does not describe the cost of operating a service.

Sairo’s pricing page shows the Free allowance and active paid-plan allowances. Free requires no payment card and keeps a small Powered by Sairo mark on the widget. When no paid amount is published, the page asks you to contact the team. This guide does not invent a subscription quote or a promised saving.

Count useful pages and decide how often they change.

Count the pages the assistant needs, not every URL your CMS can generate. Archives, filters, duplicates, and obsolete offers can increase a crawl without adding useful answers. Start with the sources that support real visitor questions.

For a simple planning scenario, 100 included pages refreshed twice in a month represents 200 planned page refreshes. Initial indexing, retries, extra refreshes, and browser-rendered work can add activity. Compare the estimate with actual metering after setup.

Estimate conversation demand with explicit assumptions.

Use a website traffic figure you trust and choose a scenario for the share of visits that start a conversation. A site with 10,000 monthly visits and an assumed 2% start rate produces an estimate of 200 conversations.

That 2% is an example input, not a benchmark or a forecast for your website. Try lower and higher scenarios, then replace assumptions with observed usage after launch. Campaigns and changes in visitor intent can alter the result.

Understand why model spend can vary.

Generated replies and embeddings are different operations. Input context, conversation length, output length, the selected model, and provider pricing can all affect processing cost. An exact saved-answer match can avoid generation for that response, but it does not make the entire service free.

Sairo staff can configure a model and maximum reply-token allowance at supported scopes. A conversation allowance is not a promise that every conversation consumes an identical number of tokens. Ask which provider costs are included in your agreement.

Questions to settle before activation
AreaAsk about
SubscriptionPublished or quoted amount, currency, billing interval, and activation terms.
AllowancesSites, pages per site, page refreshes, conversations, and any custom limits.
AI processingSelected model, output allowance, and responsibility for provider charges.
OperationsWho reviews answers, maintains content, and completes follow-up.
ChangesHow a higher limit or a different plan is approved.

Include the work around the widget.

Someone needs to confirm the business facts, test the installation, review knowledge gaps, and handle enquiries. An agency should agree whether that work belongs to its team or the client before quoting a rollout.

For your own planning, you can multiply expected review hours by an internal hourly cost. Keep that assumption separate from the Sairo price and avoid presenting unmeasured time savings as money already earned.

Use a pilot to replace assumptions with evidence.

Choose a manageable page scope, record the initial content and conversation estimates, and review actual usage after a representative period. Check whether the assistant answered the intended questions and whether the follow-up process was sustainable.

Bring that evidence to a plan discussion. A useful quote starts with the number of sites, the included content, expected traffic, required controls, and the responsibilities around ongoing operation.

Common questions

Does the planner calculate my Sairo bill?

No. It estimates conversation and page-refresh workload from inputs you choose. Prices, metering rules, initial indexing, extra work, and custom allowances must be confirmed separately.

Is the example chat-start percentage an industry benchmark?

No. It is an editable scenario assumption. Replace it with your own observed data when available.

See what your website can answer.

Try a sample, review the result, and decide what comes next.