def format_metrics_efficient(extracted_metrics):
    """
    Converts extracted metrics into a clean, graphable DataFrame.
    This version filters for annual data points ('fp' == 'FY'), processes
    them based on 'end' dates, handles duplicates by prioritizing the most
    recent filing, removes rows with too many N/A values, and formats
    the 'end' date column to show only the date.
    """
    all_entries = []
    # Iterate through each metric and its corresponding entries.
    for metric_name, entries in extracted_metrics.items():
        for entry in entries:
            # Filter for entries that represent a full fiscal year ('FY').
            if entry.get('fp') == 'FY':
                record = {
                    'metric': metric_name,
                    'end': entry['end'],
                    'val': entry['val'],
                    'filed': entry['filed'],
                    'fy': entry['fy']
                }
                if 'start' in entry:
                    record['start'] = entry['start']
                all_entries.append(record)

    if not all_entries:
        return pd.DataFrame()

    # Create a DataFrame from the filtered list of annual entries.
    df = pd.DataFrame(all_entries)

    # Convert date strings to datetime objects for correct sorting.
    df['filed'] = pd.to_datetime(df['filed'])
    df['end'] = pd.to_datetime(df['end'])
    if 'start' in df.columns:
        df['start'] = pd.to_datetime(df['start'])

    # Sort by filing date to ensure the latest filing is first.
    df.sort_values(by='filed', ascending=False, inplace=True)

    # Drop duplicates for the same metric and end date, keeping the most recent filing.
    df.drop_duplicates(subset=['metric', 'end'], keep='first', inplace=True)

    # Pivot the DataFrame to get end dates as the index and metrics as columns.
    df_pivot = df.pivot_table(index='end', columns='metric', values='val')

    # Remove rows where more than 70% of the values are N/A.
    df_pivot.dropna(thresh=df_pivot.shape[1] * 0.3, inplace=True)

    # Standardize common metric names.
    metric_mapping = {
        'SellingAndMarketingExpense': 'SG&A',
        'SellingGeneralAndAdministrativeExpense': 'SG&A',
        'ResearchAndDevelopmentExpense': 'R&D',
        'ResearchAndDevelopment': 'R&D'
    }
    df_pivot.rename(columns=metric_mapping, inplace=True)

    # Consolidate columns if renaming created duplicates.
    if df_pivot.columns.has_duplicates:
        df_pivot = df_pivot.groupby(level=0, axis=1).first()

    # Reset index to make 'end' a regular column and clean up.
    df_pivot.reset_index(inplace=True)
    df_pivot.rename_axis(None, axis=1, inplace=True)

    # Format the 'end' column to remove the time component, showing only the date.
    df_pivot['end'] = pd.to_datetime(df_pivot['end']).dt.date

    # Sort the final DataFrame by the end date.
    df_pivot.sort_values(by='end', ascending=True, inplace=True)

    return df_pivot