Note
How to read an HER search before a desk-based assessment

The HER search comes back as a spreadsheet or a GIS export, a buffer around your red line, and a list of monument numbers that someone has to turn into a coherent picture before the brief gets answered. Most of the work in reading it well is knowing what each record type is actually telling you, and what it isn't.
What comes back in an HER search
A standard search returns three kinds of record, and they don't carry the same weight. Monument records (PRNs, HER numbers, whatever the local naming convention is) describe something somebody has identified and logged: a scheduled barrow, a findspot, a cropmark complex from a 1970s flight. Event records sit alongside them and tell you what work has happened on the ground: a watching brief, a geophysical survey, a trial trench evaluation. Then there's the grey literature list: titles and a reference number, pointing at reports that sit in a county archive, an OASIS record, or nowhere you can get at without emailing the HER officer.
The trap is treating the monument count as a density map. A search area with forty records and one with four isn't necessarily telling you the first has ten times the archaeology. It might just mean the first has had ten times the fieldwork, or sits under a parish with an active local society that's been fieldwalking since the 1980s. Read the event records before you read the monument records. They tell you how much looking has happened, which is the number that lets you interpret everything else.
Pay attention to date ranges on the monument records too. A "Roman settlement (probable)" logged from a single field survey in 1962 means something different from the same description attached to a 2019 geophysics report with a plot attached. HER staff are generally good about flagging confidence, but the flagging convention varies between counties and sometimes between officers within the same county, so don't assume consistency across a search area that crosses an administrative boundary.
Where the HER search runs out of road
The HER only knows about what's been reported to it. That's the limit that matters most for a desk-based assessment, and it's worth stating plainly in the report rather than leaving the curator to infer it. A field with no monument records might be archaeologically empty. It might just as easily be a field nobody has ever looked at. Arable land that's been in continuous cultivation, never geophysed, never cropmark-surveyed, never fieldwalked by anyone who wrote it up, will come back clean on an HER search regardless of what's sitting forty centimetres down.
This is why a competent desk-based assessment treats the HER search as one layer among several. Historic mapping, LiDAR where it's available, the aerial photo collections held by the NMR and local flying clubs, and any geophysics commissioned for adjacent sites sit alongside it, and each covers a different kind of gap. Most search areas have no cropmark photography at all, simply because nobody flew that parish in a dry summer at the right time.
That's the gap Buried Feature Detection is built to close without waiting on an archive flight that may never have happened: it reads NIR imagery over the search area during the current growing season and returns a georeferenced cropmark layer you can drop straight into the assessment or into QGIS alongside the HER export.
Writing the gap into the report
Once you've worked through the records, say what the search area has and hasn't had done to it, not just what's been found. A sentence like "no cropmark evidence is recorded for this search area; no aerial reconnaissance has been flown over it in the HER's holdings" is more useful to a curator than silence on the point, and it's the kind of line that heads off a follow-up request for geophysics you could have scoped for in the first draft.
If that gap is one you'd rather close before the brief forces your hand, it's worth seeing what a current-season cropmark pass over your search area looks like.