# Separating bot traffic, visitors and enquiries Requests, sessions and enquiries describe different events. Automated crawlers can generate many requests without any purchasing interest. A crawler fetching fifty articles creates many server requests but does not represent fifty prospective customers. Avoid presenting raw request totals as customer demand or using intrusive tracking without a clear need. ## Name the unit before discussing growth A request, a page view, a visitor and an application are not interchangeable. One visitor can trigger many requests for documents, images and scripts. An automated crawler can request hundreds of pages without any person considering the service. A genuine person can read useful information without completing the form. Begin every report by naming what the number counts. For a small business, separate three questions: is the site being requested, are people engaging with relevant content and are valid enquiries being recorded? Infrastructure logs help with the first; appropriately configured audience measurement may help with the second; the application database answers the third. No single total automatically explains all three. Imagine a sudden rise in requests after publishing an article library. That might reflect crawlers discovering the new pages. It could also include previews, monitoring or repeated asset requests. Calling the increase “new customers” would be unsupported. Investigate the source and behaviour before choosing a commercial label for the chart. ## Keep technical monitoring distinct from customer records A form error rate can reveal a broken service without requiring full form contents in logs. Record the type of failure and enough non-sensitive operational context to diagnose it. Do not turn routine monitoring into a second uncontrolled copy of names, email addresses and project descriptions. The private application store should have its own access and retention rules. Similarly, a notification is not the authoritative count of saved applications. An email service may fail after a database record has been created, or a retry may deliver the same notification twice. Use stable application references and a repeat-safe submission process to avoid interpreting duplicates as separate prospects. Document which system the operator should consult when numbers disagree. When exporting data, identify test submissions and remove them from business reporting according to the agreed process. An authorised launch test proves a workflow under its test conditions; it is not a customer enquiry. Keep the distinction visible to whoever reviews the list later, especially if different people handle development and sales. ## Turn anomalies into questions you can verify If requests rise but applications do not, first ask whether the traffic is relevant and human before redesigning the form. If relevant page views are stable but form failures rise, inspect the submission path. If saved applications exist but nobody has replied, the issue may be an operational handoff rather than a website problem. A clear measurement model points to the correct investigation. Use comparisons with a known period and note changes such as new pages, campaigns or monitoring tools. Avoid small-sample percentages that appear dramatic without their underlying counts. When the available evidence cannot identify a visitor's full journey, report that limitation rather than filling the gap with an assumption. The final business report can remain simple: published content, relevant observed interest, valid saved enquiries and outstanding follow-up. Each item should have a defined source. This gives the owner a useful operating picture while preventing crawler activity or technical checks from being presented as commercial success that has not occurred. **A simple monthly summary.** Use separate rows for observed public interest, successfully saved enquiries and enquiries awaiting a response. Note the source of each row and any known exclusions, such as authorised tests. If a data source is unavailable, mark it unavailable rather than replacing it with a different unit that happens to be easy to count. When a spike appears, annotate what changed around that date. New articles, a crawler visit, a campaign or a repeated monitoring request can create different patterns. Investigate before relabelling the spike as growth in customers. The business may still learn something useful from infrastructure activity without pretending it is human demand. Check the operational handoff too. A rising count of saved enquiries with a growing unreviewed queue is a different problem from a failing form. Assign someone to act on valid records and define how status is maintained. Measurement becomes valuable when it clarifies the next responsible action; a large total without ownership can hide work that customers are still waiting for. When presenting a count, include its definition beside the chart rather than only in a separate document. A future colleague should immediately recognise whether the figure describes technical requests, saved applications or work still awaiting a response. ## Practical checklist - Define the metric first - Separate automated traffic where possible - Count stored enquiries distinctly **A concrete example:** A crawler fetching fifty articles creates many server requests but does not represent fifty prospective customers. **A boundary to keep clear:** Avoid presenting raw request totals as customer demand or using intrusive tracking without a clear need. ## Your next step [Tell Orvunweb about your project](/en/start-a-project/) ## Related reading - [Designing a contact form people can complete](/en/blog/designing-a-contact-form-people-can-complete/) - [Protecting contact forms against spam](/en/blog/protecting-contact-forms-against-spam/) - [Understanding Search Console impressions and clicks](/en/blog/understanding-search-console-impressions-and-clicks/) - [Business website service](/en/services/business-website/) - [Compare packages](/en/pricing/) ## Related services - [Business website](/en/services/business-website/) --- Locale: en Canonical: https://orvunweb.com/en/blog/separating-bot-traffic-visitors-and-enquiries/ Updated: 2026-09-16