⌖Opportunity search

It automatically fetches current jobs and projects from the web and, with a click, creates them as opportunities — with a contact person and a verified email address where possible.

The results are then ready for sending. So the route from research to campaign is one step, not three.

Up to ten search terms, each confirmed with Enter, plus the region.

The region feeds straight into the search — no filtering afterwards. That is an important difference: results are not all fetched first and then thrown away; the search is targeted from the start. So a region is not a cost-saving measure after the event but a genuine restriction.

You can be generous with the terms — more terms means more coverage.

This search costs real money — Google queries, web search and AI evaluation. That is why it is one of the few places with a visible price tag.

Three safeguards apply:

• The cost preview before the click states the upper limit. It is an upper limit, not an exact figure; it is calculated with stored prices, and billing takes place at the respective provider.
• The monthly budgets are shown as they are used. If a budget is used up, the function pauses until the end of the month — everything else keeps working.
• The licence always remains the upper limit, even if no budget of your own has been set.

Enlarge

The most effective lever is contact enrichment: it only runs for the best-rated results — not for everything a search happens to throw up. Keeping this number low saves the most.

Other adjustments:

• The free sources are often enough. Three of the sources cost nothing and deliver most of the results; the most productive one brings around 25 results per term.
• In the free quota, it stays at one page of results per search term. More pages cost one request each.
• The cache prevents the same search from being paid for again shortly afterwards. Set to 0, every search goes out fresh and costs again.
• Web search is the most expensive single item — one call costs many times an ordinary AI request.

From several sources side by side, three of which are free and deliver most of the results. An official interface for public tenders is deliberately set more sparingly than the others: it responds slowly, and every term there costs a good one and a half seconds of waiting.

How far back tenders are considered can be configured; tenders with deadlines still running stay in the list either way.

You take results over one by one or with “Save all” — they land as opportunities in your opportunity list, and the contact person becomes a contact with the tag “Opportunity search”.

Two notes on the list:

• It is sorted by fit: to your search terms, to the location, to how recent it is and to how reachable the contact is.
• The badge “last session” means: these are results of the last run, not a fresh state. For current results, you have to search again, and that costs again.

Individual results can be ignored; they then no longer appear in future searches.

Yes, via the exclusion terms — comma-separated. They are checked against title and company, not against the whole text.

This is the usual way to keep staffing agencies or competitors out of the results, which would otherwise flood every search.

Because they only feed the preview. These figures do not bill anything — billing takes place at the respective provider at its rates.

They can be set because rates change and a wrongly guessed figure would make a cost preview worthless. If your preview is systematically off, these are the values to correct — not the invoice.

Those that actually appear in a tender — not those you use to describe yourself.

What works: the service in the client’s language (“Interim CFO”, “site management building construction”), ideally in several variants. What does not work: generic terms such as “consulting” and coinages from your own brochure.

You can be generous with the terms — more terms means more coverage, and the free sources deliver most of the results.

It applies as soon as a location is entered in the search — specified in kilometres.

Bear in mind that it applies to the location in the tender, not to your own location. For projects advertised supra-regionally, a narrow radius can therefore discard results you would take on.

It prevents the same search from being paid for again shortly afterwards. It is specified in minutes.

0 switches it off — every search then goes out fresh and costs again. That only makes sense if you really need brand-new results and accept the costs.

For the same reason, a results list carries the badge “last session”: you are then seeing results of the last run, not a fresh state.

Usually the terms, more rarely the technology. Go through it in order:

1. Are the terms too general? “Project” matches everything and nothing.
2. Are the exclusion terms maintained? Staffing agencies and intermediaries otherwise flood every search — they are checked against title and company.
3. Does the region fit? It feeds straight into the search.
4. Is the period too wide? Old tenders are not wrong, but they are useless.

The list is sorted by fit — if even the top results are off, the terms are the cause.

They are verified where possible — nothing is guessed. That is precisely why enrichment costs money: a guessed address following the pattern firstname.lastname@company.com would be free and produces bounces that cost sender reputation.

Nevertheless: spot-check the first results before you build a big campaign on them. And keep the first round small.

The same rules apply as for all other addresses: business outreach needs a recognisable factual connection and a case-by-case assessment.

With opportunities from a tender, this connection is usually obvious — someone is looking for exactly what you offer. That is the most favourable starting position there is, and the reason why this kind of research is less problematic legally than a purchased list.

As everywhere, the unsubscribe link is added automatically.

Because Google’s job search does not deliver its results immediately. The search still starts straight after the click: results from the other sources appear in the list at once, and the job adverts follow usually after about a minute.

Until then, above the list it says “Search running — job adverts follow in about a minute”. You do not need to reload; the list fills itself. You may also leave the page: the search keeps running, and when you come back, the result is there.

If the job search does not respond within ten minutes, the search is completed with the results available, and a note says so. Then it helps to search again later. If the search finishes normally and there are no job adverts, there simply were none for your terms in the chosen time window.

The same applies to “Find & prepare applications”: there too, job adverts from Google’s job search arrive a little later than the rest.

The time window in the opportunity search settings (last 3 days, last week, last month) is based on the publication date shown on the advert. Older adverts are sorted out.

How many adverts were hidden as a result is shown as a note below the search.

Not every advert states a date that can be read reliably. Such results are not dropped but stay in the list marked “date unknown”. The quickest way to see whether they are still current is in the advert itself.

Google sorts job adverts by relevance, not by date. A short time window therefore keeps the list fresh but often leaves only a few results. For a niche, “last month” is usually the better choice; the “Premium” setting brings more results per search term, but uses more research queries.