The operating problem
PromoStars began with a simple proposition: make promotional talent easier to discover directly through useful public profiles rather than forcing every search through an agency directory. The software was only one part of the challenge. The platform needed enough credible profiles, geographic coverage, useful search pages, safe contact flows and search-engine visibility to create value before a mature demand loop existed.
Engineering judgement tied to operating evidence.
Founder and software/electronics engineer at INESSOFT. Reviewed against current INESSOFT delivery practice and operating evidence.
Published 06 Jul 2026 · Last reviewed 06 Jul 2026The cold-start problem has at least four sides
A marketplace is often described as a two-sided problem: supply and demand. In practice, discovery and trust are separate systems. Supply can exist but remain invisible; demand can arrive but distrust the records; search engines can crawl pages that are too thin; and users can sign up without completing profiles that create value.
- Measure published, searchable supply rather than registrations alone.
- Treat discovery pages as product surfaces with real inventory.
- Design trust and moderation before aggressive growth creates cleanup debt.
Curation improved the product before it improved the brand
Automatically publishing every signup would have increased the visible count quickly, but it would also have produced incomplete profiles, weak images and inconsistent records. Manual enablement created operational effort, yet it protected the search experience and made public supply more useful.
- Separate account creation from public publication.
- Give administrators visibility into incomplete and low-quality records.
- Use reminders and assistance to improve profiles rather than lowering the publication standard.
Structured SEO worked because it matched real discovery intent
People search for talent by city, service and use case. Dynamic city/service routes created pages that could answer those searches with actual profiles and related discovery paths. The value did not come from generating thousands of keyword combinations blindly; it came from matching content structure to how buyers search.
- Only publish combinations supported by credible inventory or useful editorial context.
- Use internal linking to connect national, city, service and profile pages.
- Keep metadata, schema and sitemap behaviour tied to the same canonical route model.
Supply growth and demand growth are different programmes
Profile acquisition can create visible momentum while still failing to produce enough enquiries. Demand requires search rankings, brand recognition, buyer tools, confidence and a clear contact journey. The platform therefore needed separate measures for profile growth, profile quality, discovery traffic, shortlists and messages.
- Do not use signup count as a proxy for marketplace success.
- Track the buyer journey from landing page to profile, shortlist and enquiry.
- Invest in demand surfaces once supply is credible enough to satisfy the search.
Trust features are part of conversion
Removing public email addresses, providing controlled contact routes, publishing safety guidance and clarifying that the platform is not an agency all reduced ambiguity. Trust work can appear secondary to feature development, but it determines whether users are willing to expose information or contact someone through the system.
- Make the platform's role and limits explicit.
- Protect personal contact data from unnecessary scraping.
- Design reporting, moderation and privacy controls as first-class workflows.
Marketplace operations need internal tools early
Moderation, invitations, incomplete-profile reminders, publication controls, partner outreach and activity tracking quickly become operational workloads. Without internal tools, the team spends growth time manipulating the database manually. Admin capability is therefore part of the marketplace product.
- Build queues around the highest-volume operational decisions.
- Record why profiles were enabled, disabled or contacted where useful.
- Automate repetitive outreach carefully without losing consent and rate control.
The most important lesson was that distribution is engineered
Search visibility, internal linking, profile quality, content structure and enquiry flow were not separate marketing tasks added after development. They were design constraints that changed the database, routes, admin tools and measurement model. A marketplace has to engineer its route to users as seriously as its transaction workflow.
- Model acquisition and discovery events in the product.
- Preserve successful URLs while the platform evolves.
- Use operating data to decide which categories and locations deserve deeper investment.
Use the operating signal to choose the next action.
The same symptom can justify a custom system, a smaller integration, a stabilisation phase or no build at all. The decision should follow evidence rather than enthusiasm for a particular technology.
Many users register but few become publishable.
The bottleneck is onboarding quality, not acquisition volume.
Improve completion guidance, reminders and admin assistance.
Public supply is strong but enquiries remain weak.
Demand discovery and buyer confidence need investment.
Strengthen search routes, proof, shortlisting and contact flow.
Thousands of location/service combinations are technically possible.
Programmatic pages could create thin content.
Publish only combinations with inventory or substantial useful guidance.
Manual moderation is consuming founder time.
The internal operating system is underdeveloped.
Build admin queues, status reasons, bulk actions and measured automation.
Evidence that the current approach is becoming risky.
- Registration count is presented as active marketplace supply.
- Profiles publish automatically without minimum quality controls.
- SEO pages exist without real inventory or distinctive content.
- Buyer actions are not measured beyond page views.
- Personal contact data is exposed simply because direct contact is a product goal.
What to clarify before commissioning work.
- Define the minimum publishable supply record.
- Separate signup, completion, approval and public visibility states.
- Map national, location, service and detail routes around real search intent.
- Track profile views, shortlists, enquiries and completion quality separately.
- Build moderation, reminders and publication tools into administration.
- Treat trust, privacy and platform-role clarity as conversion features.
Case studies behind the lesson.
These links provide system context, architecture, workflows and engineering decisions connected to the article.
Questions that usually appear during scoping.
Why not publish every profile automatically?
Automatic publication increases visible quantity but can reduce buyer trust and search usefulness when records are incomplete. The correct threshold depends on the marketplace and its risk.
Is programmatic SEO inherently spammy?
No. It becomes weak when pages are generated without real inventory, unique utility or a coherent discovery architecture. Structured pages can be valuable when they answer genuine location and service intent.
Which marketplace metric matters most early?
No single metric is sufficient. Publishable supply, discovery traffic, buyer engagement and completed contact actions must be considered together.
Can marketplace trust be added after growth?
It is far cheaper to design publication, privacy, moderation and reporting controls before the platform accumulates low-quality or risky public records.